2017-09-01 00:37:06 skarnet: --branch mater is mostly useless, --depth 10 puts a lot of load on the server 2017-09-01 00:37:10 err 2017-09-01 00:37:13 skazz, but they're gone. 2017-09-01 00:37:23 fwiw, --depth isn't necessary as faster as people think it it to be 2017-09-01 02:44:20 clandmeter: we're slowly getting more mirrors 2017-09-01 02:44:34 http://speglar.siminn.is/alpine/ this one through rsync, http and ftp 2017-09-01 02:45:38 hosted by the Icelandic ISP Síminn hf 2017-09-01 04:45:02 so there are cases where `which` is not available on the system, and it is making docbook-x{m,s}l triggers fail. should I convert these to POSIX portable `command` calls, or should `which` be added as a dep? 2017-09-01 05:07:14 foxkit: which is a handy tool, but to keep it light weight, i vote for converting to command calls 2017-09-01 05:09:35 btw, we now have an icelandic mirror 2017-09-01 05:09:38 the entire repo 2017-09-01 05:12:05 morning fabled 2017-09-01 05:12:31 morning danieli 2017-09-01 05:12:42 as i said, alpine now has an icelandic mirror 2017-09-01 05:12:48 probably gonna be more soon enough 2017-09-01 05:13:08 has alpine considered setting up something like mirrorbits? 2017-09-01 05:14:19 <_ikke_> https://github.com/etix/mirrorbits ? 2017-09-01 05:14:27 _ikke_: yep! 2017-09-01 05:14:46 we have dl-cdn as cdn; but we sometime ago spoke about geodns 2017-09-01 05:14:53 yeah, that's doable 2017-09-01 05:15:05 I could write something using powerdns, or you could use something like mirrorbits 2017-09-01 05:15:32 geodns (+ round robin?) could be very nice, and exploits the usefulness of worldwide mirrors 2017-09-01 05:16:00 is mirrorbits another geodns, or it's somesort of http level redirector? 2017-09-01 05:17:25 ah, json api and/or http redirect 2017-09-01 05:17:31 oh sorry, yes 2017-09-01 05:17:38 geodns would be better imo 2017-09-01 05:18:22 geodns is more transparent 2017-09-01 05:18:32 exactly 2017-09-01 05:18:44 and easier to manage due to no API and no HTTP level redirects 2017-09-01 05:18:47 mirrorbits probably wants to control all mirrors to keep them up to date 2017-09-01 05:18:47 also, quote, "I’ve added Alpine Linux to the mirror and it’s currently syncing, since we have the resources then I’m syncing the entire repo, we will sync at a 1 hour interval when the initial sync is done." 2017-09-01 05:19:01 yeah, I would assume so 2017-09-01 05:19:15 need to ping ncopa on the geodns thing 2017-09-01 05:19:21 yeah, indeed 2017-09-01 05:19:27 there are some things I want to discuss with him 2017-09-01 05:19:38 is there something like an #alpine-meetings or is it all behind closed doors? 2017-09-01 05:21:08 <_ikke_> danieli: most disussion is happening in these channels and on the mailing list iirc 2017-09-01 05:21:15 I see 2017-09-01 05:22:01 there isn't really an IRC chan for administrative stuff, there's offtopic, dev and community support / general channel 2017-09-01 05:22:28 <_ikke_> danieli: there is also -infra 2017-09-01 05:22:33 cool, I didn't know 2017-09-01 05:22:43 what is -infra for? 2017-09-01 05:28:47 it's been pretty quiet, it's been mostly for committers and those running builders 2017-09-01 05:28:59 discussions or notifications from bots? 2017-09-01 05:29:02 assuming discussions 2017-09-01 05:29:19 discussion, usually when action is needed on some technical level. to fix builders, etc 2017-09-01 05:29:24 I see 2017-09-01 05:29:36 or the hosting infra 2017-09-01 05:29:42 like the web site 2017-09-01 05:29:46 or bug tracker 2017-09-01 05:29:47 etc 2017-09-01 05:30:01 point taken 2017-09-01 05:38:36 fabled: https://p.duniel.no/bVNq 2017-09-01 05:38:38 i translated it 2017-09-01 05:40:22 yes, we spoken about having an org... so far it's been sort of mix of 'do-ocracy' and 'meritocracy' 2017-09-01 05:40:34 I can help fixing a legal org 2017-09-01 05:41:15 what worries me more is the lack of transparency 2017-09-01 05:41:18 we've been reluctant for legal org originally, but after recent growth it's probably mandatory to have one 2017-09-01 05:41:27 my thoughts exactly 2017-09-01 05:41:52 originally this was ncopa-dictator ship 2017-09-01 05:42:03 but it has a been good one :) 2017-09-01 05:42:04 ncopa has a lot to do, I can imagine 2017-09-01 05:42:08 yeah, without doubt 2017-09-01 05:42:36 i think we try to let people work, and stay out of the way as much as possible. if there's problems we see what can be done. 2017-09-01 05:42:59 the legal org stuff isn't too much of a worry to me, more transparency and the leadership structure 2017-09-01 05:43:03 it has worked pretty well, since everyone has mutual understanding and respect... so far we've known when to ask for other's opinion 2017-09-01 05:43:06 it's just very vague and diffuse 2017-09-01 05:43:25 but also the increased amount of contributors might warrant something a bit more formal 2017-09-01 05:43:50 generally on big decisions we just poll the people with git push access 2017-09-01 05:44:11 maybe we should publish that list 2017-09-01 05:45:08 the lack of formal steering committee has avoided us from much lobbying :) 2017-09-01 05:45:20 but yes, i also understand the concerns it creates 2017-09-01 05:45:33 the problem is that the people working on alpine are technical ones who love to hack 2017-09-01 05:45:44 and would prefer not to do any of the administrative paper work 2017-09-01 05:45:48 it's been done 2017-09-01 05:48:37 yeah, and that's a good thing - but there's things like licenses, donations, where said donations go, leadership 2017-09-01 05:49:25 i guess we should just document the current flow 2017-09-01 05:49:29 agreed 2017-09-01 05:51:17 that's more or less my point, I guess 2017-09-01 06:16:49 morning 2017-09-01 06:17:22 documentation has always been a struggle imho 2017-09-01 06:17:31 im not good at writing good documentation 2017-09-01 06:18:19 I can help 2017-09-01 06:18:22 We need to start using docs.a.o 2017-09-01 06:18:40 docs.a.o? 2017-09-01 06:18:42 danieli: that would be great! 2017-09-01 06:18:52 ncopa: did you see the email from the icelandic dude? :) 2017-09-01 06:19:25 i think i saw it 2017-09-01 06:19:28 mm 2017-09-01 06:19:30 Git based wiki 2017-09-01 06:19:31 we got an icelandic mirror 2017-09-01 06:19:39 markdown is great 2017-09-01 06:19:57 markdown is simple 2017-09-01 06:20:03 which is great :) 2017-09-01 06:20:05 it's simple indeed, and functional as hell 2017-09-01 06:20:08 ncopa: i'm about to fire off a couple more emails, do you want to be a cc or bcc on them? 2017-09-01 06:20:13 It supports more than just markdown 2017-09-01 06:20:24 well yeah, but I was just saying, markdown is noice 2017-09-01 06:20:42 i like markdown too, im just afraid it is too limited 2017-09-01 06:20:46 agreed 2017-09-01 06:20:55 why not put documentation on the wiki? 2017-09-01 06:20:55 We need to create a proper structure 2017-09-01 06:21:05 depends if the markdown parser lets you also use arbitrary html 2017-09-01 06:21:05 make a category and section for it 2017-09-01 06:21:15 Xe: that's implementation specific :/ 2017-09-01 06:21:18 if so then 👌👍 2017-09-01 06:21:29 Because git has advantages 2017-09-01 06:21:37 Xe: that is my point, md itself is limited so you have to resort to html 2017-09-01 06:21:54 could probably edit a markdown parser to support arbitrary html 2017-09-01 06:21:57 the problem with wiki is that it grows organically 2017-09-01 06:22:01 yeah true 2017-09-01 06:22:08 we need someone who can architect the docs 2017-09-01 06:22:23 someone that has the overview 2017-09-01 06:22:26 Structure is the key 2017-09-01 06:22:30 exactly 2017-09-01 06:22:36 someone that can build the structure 2017-09-01 06:22:38 sceleton 2017-09-01 06:22:45 And time ;-) 2017-09-01 06:22:56 and qualitycheck the contributions to docs 2017-09-01 06:23:02 so git makes alot sense 2017-09-01 06:23:03 i can write some structural templates 2017-09-01 06:23:24 then we need to migrate much of the wiki contents to the git 2017-09-01 06:23:27 and remove it from wiki 2017-09-01 06:23:33 we can still have a community wiki 2017-09-01 06:23:44 but we need something for official documentation 2017-09-01 06:23:57 Yes and we should push devs to write manuals 2017-09-01 06:24:08 documentation is key 2017-09-01 06:24:09 Or ask them nicely 2017-09-01 06:24:16 yeah, asking nicely is probably a better solution hahaha 2017-09-01 06:24:22 kick them in the rear parts 2017-09-01 06:24:24 :) 2017-09-01 06:24:27 i like docs.a.o better than the current mediawiki :p 2017-09-01 06:24:39 OH THAT docs.a.o 2017-09-01 06:24:40 i'm slow 2017-09-01 06:24:46 re:above -- I see the advantages of a legal org, but also the distinct disadvantages it can give 2017-09-01 06:24:57 i wonder if we could model it after how something like openbsd does it 2017-09-01 06:25:10 without doubt, but I'm more worried about transparency and the organizational (leadership) *structure* 2017-09-01 06:25:20 The repo is on GitHub 2017-09-01 06:25:42 norway is funky 2017-09-01 06:26:00 danieli: would you like to work on the docs? 2017-09-01 06:26:06 ncopa: hell, why not 2017-09-01 06:26:11 that would be wonderful 2017-09-01 06:26:25 https://github.com/alpinelinux/alpine-wiki 2017-09-01 06:26:40 come with a big picutre or overview suggestion on how you want it generally organized 2017-09-01 06:26:59 eg what goes into official docs 2017-09-01 06:27:04 Probably best to share it on the ml 2017-09-01 06:27:04 and what goes to community wiki 2017-09-01 06:27:16 then a plan how to get there 2017-09-01 06:27:27 yeah, I'll write some suggestions on the structure 2017-09-01 06:27:36 i have some opinions too 2017-09-01 06:27:44 Ok I'm going back to holiday mode. Ttyl 2017-09-01 06:27:52 clandmeter: enjoy! 2017-09-01 06:27:57 have fun, laters 2017-09-01 06:28:58 danieli: once we have the overall structure we want, you can ask help filling in the empty parts 2017-09-01 06:29:09 for example how the team is organized etc 2017-09-01 06:29:36 where we write the policies, packaging, infra etc 2017-09-01 06:29:41 :) 2017-09-01 06:29:41 yeah, i just need to map the existing services and such 2017-09-01 06:29:53 danieli : without doubt, but I'm more worried about transparency and the organizational (leadership) *structure* 2017-09-01 06:29:57 I agree 2017-09-01 06:30:12 transparency is the big thing for me, and probably others too 2017-09-01 06:30:14 if it helps alleviate concerns. I'm not sure it's transparent to even core devs either 2017-09-01 06:30:16 :p 2017-09-01 06:30:27 doesn't really help LOL 2017-09-01 06:30:40 also 2017-09-01 06:30:52 at least it means we're not a shadowy dev cabal 2017-09-01 06:30:57 once we have the overall documentaion plan worked out, we need prioritize the order 2017-09-01 06:31:17 Shiz: yeah true 2017-09-01 06:32:21 the documentation project is a huge one 2017-09-01 06:32:26 without doubt 2017-09-01 06:32:43 ncopa: that reminds me, what about geodns for dl.a.o? 2017-09-01 06:32:45 so im very happy someone wants take care of it 2017-09-01 06:32:47 dl* 2017-09-01 06:32:54 we have talked about it 2017-09-01 06:33:03 isn't dl-cdn already a.... 2017-09-01 06:33:05 well CDN 2017-09-01 06:33:12 what *is* dl-cdn? 2017-09-01 06:33:32 I have newapkbuild.1 and parts of APKBUILD.5 written, and plan to write apk.8 at some point soon, if that helps the documentation 2017-09-01 06:33:35 dl-cdn is a CDN mirror sponsored by faaasttttlyyyy 2017-09-01 06:33:43 define CDN mirror 2017-09-01 06:33:44 up real late.. keyboard has had a fault.. arrow keys stopped working 2017-09-01 06:33:52 whoops, mmy bluetooth keyboard is nnnnnnoottt happy :) 2017-09-01 06:33:55 it's just a mirror where they handle the redirection and stuff? 2017-09-01 06:34:13 ncopa: iiiiit''''sss okaaayyyyyy 2017-09-01 06:34:19 lol 2017-09-01 06:34:51 foxkit: nice :) 2017-09-01 06:35:04 seeeqaklø 2017-09-01 06:35:13 foxkit: this keyboard occasionally flashes like a christmas tree and dies 2017-09-01 06:35:14 bad cable 2017-09-01 06:36:04 danieli: this isn't a bad cable.. I just received it with a pallet (almost literally) of other Sun accessories, and I think this keyboard came with said fault. very old. but worth it.. 2017-09-01 06:36:12 sun..? damn 2017-09-01 06:36:13 i'm jealous 2017-09-01 06:36:17 to some, the Model M is perfection... for old unix nerd like me... sun type 6 2017-09-01 06:36:47 i dream of one day acquiring PC-UX/V 2017-09-01 06:36:51 ACTION dozes off 2017-09-01 06:36:56 http://i.imgur.com/arhl9du.jpg 2017-09-01 06:37:24 dl-cdn is a service sponsored by fastly 2017-09-01 06:37:37 bascially, it is varnish caches 2017-09-01 06:37:52 iirc they use dl-4.alpinelinux.org as backend 2017-09-01 06:38:00 im not sure how they do the geodns thing 2017-09-01 06:38:07 but i think they do geodns 2017-09-01 06:38:49 they don't do geodns 2017-09-01 06:38:55 they do ip multicast 2017-09-01 06:38:59 ok 2017-09-01 06:39:21 <_ikke_> ip anycast? 2017-09-01 06:39:56 I can write something geoip for assorted mirrors using powerdns 2017-09-01 06:40:18 sorry yeah anycast 2017-09-01 06:40:20 it's early here 2017-09-01 06:41:30 Shiz: where are you? 2017-09-01 06:41:35 .nl 2017-09-01 06:41:42 me, you and ncopa are in the same time zone 2017-09-01 06:41:45 foxkit: thanks for working on manpages. I appreciate 2017-09-01 06:42:00 and clandmeter too 2017-09-01 06:42:02 foxkit: btw, i love your "projects" idea 2017-09-01 06:42:21 "projects" idea? 2017-09-01 06:42:24 fill me in 2017-09-01 06:42:40 <_ikke_> and mee 2017-09-01 06:42:46 http://lists.alpinelinux.org/alpine-devel/5811.html 2017-09-01 06:42:54 read the mailing list more often ya bums 2017-09-01 06:43:09 I haven't even subscribed to it to be honest 2017-09-01 06:43:11 i should do that 2017-09-01 06:43:48 we have an #alpine-infra channel that might be of interest too 2017-09-01 06:43:56 I was told so, and joined it 2017-09-01 06:44:14 ncopa: it would also be nice if we could step up our securiyt game 2017-09-01 06:44:22 security.. I'm in 2017-09-01 06:44:33 not to poke, but i handed off that gpg remailer thing months ago 2017-09-01 06:44:34 security on what level? 2017-09-01 06:44:35 :p 2017-09-01 06:44:42 infrastructural mostly 2017-09-01 06:44:48 having an email to report vulns to 2017-09-01 06:44:52 membership of the distros lists 2017-09-01 06:44:57 I can fix that! 2017-09-01 06:44:57 maybe tracking cves 2017-09-01 06:45:04 i fixed it months ago 2017-09-01 06:45:06 :P 2017-09-01 06:45:08 I'm into security a lot, especially lately 2017-09-01 06:51:04 so - where do we want what documentation? 2017-09-01 06:51:10 what do we put on the wiki, what do we put on the docs? 2017-09-01 06:54:25 i think we want all the official documentation in git 2017-09-01 06:54:42 it's currently fractured 2017-09-01 06:55:06 what is the purpose for wiki. and docs. respectively? 2017-09-01 06:55:19 wiki: community docs 2017-09-01 06:55:29 docs.a.o: anything to be considered as official documentation 2017-09-01 06:55:41 so we want to move docs.a.o to git? or how does it work? 2017-09-01 06:55:55 docs.a.o = git 2017-09-01 06:56:00 aha 2017-09-01 06:56:34 the idea is to have better structure and quality check of official docs 2017-09-01 06:56:44 what i think we want for official docs: 2017-09-01 06:56:58 install docs 2017-09-01 06:57:01 user guide 2017-09-01 06:57:12 developer docs 2017-09-01 06:57:18 eg hot to build packages 2017-09-01 06:57:21 how* 2017-09-01 06:57:25 agreed 2017-09-01 06:57:28 packaging policies 2017-09-01 06:57:36 docs directly about alpine 2017-09-01 06:57:37 but also on the organizational docs 2017-09-01 06:57:42 ah I see 2017-09-01 06:57:45 like release yccles 2017-09-01 06:57:45 how the team is organized 2017-09-01 06:57:50 yes 2017-09-01 06:57:54 release policies 2017-09-01 06:57:58 packaging policies 2017-09-01 06:58:06 developer policies 2017-09-01 06:58:12 eg how to become a developer 2017-09-01 06:58:13 *cough*coc 2017-09-01 06:58:17 coc 2017-09-01 06:58:23 thats an important bit 2017-09-01 06:58:32 how to become a mirror 2017-09-01 06:58:33 what /is/ the coc? 2017-09-01 06:58:37 decision making process too :) 2017-09-01 06:58:39 missing so far 2017-09-01 06:58:43 right 2017-09-01 06:58:46 danieli: currently none, and wanting to work on it 2017-09-01 06:58:56 there was a proposal on the ML a few months back but it lead to nothing people were happy with 2017-09-01 06:59:04 yeah, i heard 2017-09-01 06:59:09 but it's important imo 2017-09-01 06:59:12 agree 2017-09-01 06:59:18 agreed 2017-09-01 06:59:51 so basically, anyting that can be considered as official docs, should be tracked in git 2017-09-01 06:59:57 yeah 2017-09-01 07:00:00 and published on docs.a.o 2017-09-01 07:00:06 anything that's considered canonical :) 2017-09-01 07:00:31 the wiki can be more for how to fix hardware problems 2017-09-01 07:01:01 or how to do this or that 2017-09-01 07:01:16 how to install gnome, desktop wwhatever 2017-09-01 07:02:27 so much can probably be moved from wiki to docs.a.o 2017-09-01 07:02:48 i think the current docs.a.o alos support mediawiki format so we could in theory just copy the pages one by one 2017-09-01 07:03:13 Correct 2017-09-01 07:03:47 i think we should have an overall plan, what we want to be in the docs.a.o 2017-09-01 07:03:49 why shouldn't everything just be on docs.a.o in git? 2017-09-01 07:03:54 Converting between formats is a bit more complicated 2017-09-01 07:03:58 seems kind of pointless to fracture stuff between the wiki and docs.a.o 2017-09-01 07:04:25 danieli: because i think not all belongs there 2017-09-01 07:04:29 Because not everyone is familiar with git 2017-09-01 07:04:39 im afraid the documentation will be cluttered 2017-09-01 07:04:43 point taken 2017-09-01 07:04:48 we just need a clear boundary on what to put where 2017-09-01 07:04:56 too much of non-essential stuff will make the important stuff drown 2017-09-01 07:04:56 and move whatever shouldn't be on the wiki to docs 2017-09-01 07:05:11 good point 2017-09-01 07:05:46 in fact, i think we want try to keep it short 2017-09-01 07:05:55 so we just extract the pages from wiki and push them to git? 2017-09-01 07:06:01 sounds fairly easy 2017-09-01 07:06:05 that is away to do it 2017-09-01 07:06:13 we just need the overall plan 2017-09-01 07:06:15 or do we want to convert them to another format? 2017-09-01 07:06:16 how to organize it 2017-09-01 07:06:18 yeah indeed 2017-09-01 07:06:35 so this is one of the important things: what goes where 2017-09-01 07:06:59 "how to fix hardware", troubleshooting, and "non important stuff" goes to the wiki 2017-09-01 07:07:03 the rest goes to docs? 2017-09-01 07:07:11 i think, anything that is canonical, or official docs, should go to docs.a.o 2017-09-01 07:07:14 yes 2017-09-01 07:07:28 neat tricks etc should be documented in wiki 2017-09-01 07:07:40 agreed 2017-09-01 07:07:56 hardware compatibility lists etc is probably better in wiki 2017-09-01 07:08:00 agreed 2017-09-01 07:08:03 so anything that is a policy should go to docs 2017-09-01 07:08:06 do we have a HCL? 2017-09-01 07:08:18 i don think we do 2017-09-01 07:08:27 but thats a candidate for wiki 2017-09-01 07:08:41 those accumulate slowly, should probably make one and suggest people add to it 2017-09-01 07:08:51 yeah 2017-09-01 07:08:54 but that less important 2017-09-01 07:08:58 indeed 2017-09-01 07:09:13 ok 2017-09-01 07:09:14 so 2017-09-01 07:09:47 does it makes sense? 2017-09-01 07:09:54 do you have enough to start with? 2017-09-01 07:10:07 maybe we could go over the wiki pages and tag the stuff that are policies 2017-09-01 07:10:07 yeah i've made some notes 2017-09-01 07:10:11 agreed 2017-09-01 07:10:18 and move those over 2017-09-01 07:10:19 and tag out everything that gets migrated to docs.a.o 2017-09-01 07:10:32 make a redirction 2017-09-01 07:10:40 yeah i suppose 2017-09-01 07:11:02 we should probably fairly soon also have a notice on docs.a.o what it is 2017-09-01 07:11:08 what we want on the official docs: https://p.duniel.no/47PY 2017-09-01 07:11:09 yeah 2017-09-01 07:11:22 an index page stating its purpose too 2017-09-01 07:11:33 hcl can go to wiki 2017-09-01 07:11:50 oh yeah, my bad 2017-09-01 07:12:14 other wise i think its good 2017-09-01 07:13:17 write a plan how you want organize it and email to alpine-devel 2017-09-01 07:13:22 will do 2017-09-01 07:16:36 is there a place to look to see how to setup package build service? 2017-09-01 07:18:16 ncopa: there's things on docs.a.o that should go to wiki.a.o too, lots of it 2017-09-01 07:18:28 some of the tutorials and howtos i can imagine 2017-09-01 07:49:32 i think the current docs.a.o was mostly an example 2017-09-01 07:49:45 a demo that it is possible to move mediawiki to git 2017-09-01 07:50:09 I see 2017-09-01 07:50:20 dopry: i don tthink we have such doc 2017-09-01 07:50:31 might be we have some outdated 2017-09-01 07:51:15 ncopa, thanks. I'm looking at some containers people have put together. Mostly trying to get a local aarch64 CI setup running. 2017-09-01 07:52:53 Looking to make sure we can package some tools we'll be using on projects and we'll need to run on several 32/64 arm and x86 architectures. 2017-09-01 07:54:17 If anyone has a reference it would be greatly appreciated, otherwise I'm pretty sure I can figure something out with drone and the docs for package maintainers. 2017-09-01 07:56:50 what we have done so far, use a lxc for each builder 2017-09-01 07:56:59 there is a script: aports-build 2017-09-01 07:57:03 which has most of it 2017-09-01 07:57:19 note that it will build the entire repo 2017-09-01 07:57:24 and not individual packages 2017-09-01 07:57:53 there are some work going on in abuild to build a package in a namespace using bwrap 2017-09-01 07:58:03 abuild rootpkg 2017-09-01 07:58:27 i think it mostly works, but have some unresolved issues still 2017-09-01 07:58:29 cool.. I'll keep an eye out.. probably going to base something on https://hub.docker.com/r/andyshinn/alpine-abuild/~/dockerfile/ 2017-09-01 07:59:14 we use mqtt for notifying the builders on git push 2017-09-01 07:59:35 but we dont provide a stable api for mqtt messages 2017-09-01 07:59:37 I noticed the msg.alpinelinux.org mentioned in a few places. 2017-09-01 07:59:44 thats mqtt 2017-09-01 07:59:57 we also use it to report the status of build servers 2017-09-01 08:00:20 we may close it for public in future 2017-09-01 08:00:29 hm 2017-09-01 08:00:54 i suppose it woudl be nice to have read-only access to it though 2017-09-01 08:01:05 I'll probably end up using drone and it's Github integration to drive my own process. that's how I'm used to running CI anyway. 2017-09-01 08:01:44 hmm 2017-09-01 08:01:47 there's mirrors that aren't listed 2017-09-01 08:02:00 https://nanomirrors.vip.s.tuna.tsinghua.edu.cn/alpine/ for instance 2017-09-01 08:03:08 when I worked at an ISP we mirrored every thing for local. most colleges will as well. 2017-09-01 11:58:33 who has dns access? 2017-09-01 12:06:44 clandmeter: ncopa ^ 2017-09-01 12:12:19 danieli: me, nangel, rnald and fabled 2017-09-01 12:14:04 ncopa: all right 2017-09-01 12:14:21 ncopa: does your current DNS infra support wildcards? 2017-09-01 12:25:30 not sure, but i would believe it does 2017-09-01 12:25:50 i suppose i can add a * to test? 2017-09-01 12:33:52 ncopa: try *.test.a.o -> 8.8.8.8 2017-09-01 12:33:56 an A record 2017-09-01 12:34:04 nsd, you said? 2017-09-01 12:34:15 yes 2017-09-01 12:34:41 I think nsd supports wildcard 2017-09-01 12:35:06 i would be surprised if it didnt 2017-09-01 12:35:45 yes, it supports * 2017-09-01 12:35:46 i checked 2017-09-01 12:36:22 I don't think it supports partial wildcard though 2017-09-01 12:36:48 I was thinking dl*.a.o -> change NS to the PowerDNS server 2017-09-01 12:37:17 nsd is pretty bae 2017-09-01 12:37:32 unbound is nice too, for small to mid size deployments 2017-09-01 12:37:38 lets move this discussion to #alpine-infra 2017-09-01 12:37:40 powerdns is nice for programmatic DNS servers 2017-09-01 12:37:41 agreed 2017-09-01 16:46:29 ncopa: https://p.duniel.no/u8Jr rough structure, as far as I can imagine 2017-09-01 16:46:42 anything specific I should add? 2017-09-01 18:32:51 danieli, you might consider adding some meta information about the docs. How they're organized, who they're written for, etc. that will help others maintain it over time. 2017-09-01 18:34:41 dopry: just wrote a template for the categories and overall structure this far, will add it to my notes 2017-09-02 07:00:30 good morning 2017-09-02 07:19:31 hi 2017-09-02 07:19:36 hihi 2017-09-02 08:27:13 Hello, I created the standard iso from edge, the iso output all makes sense compared to live, but apk wont run from the initfs, apk complains about missing libssl.so.43, libcrypto.so.41 and libz.so.1 ... 2017-09-02 08:27:44 Is the compiled apk binary against older versions of libssl so edge cannot be built? 2017-09-02 08:36:52 Sorry, is this the wrong channel by the way? I thought it was to do with making a distribution so I figured here was the right place to ask... 2017-09-02 09:15:44 this is fine 2017-09-02 09:16:00 do all packages come from the same release? (3.5/3.6/edge?) 2017-09-02 12:54:59 Shiz: I used the following command: ./mkimage.sh --tag edge --outdir /var/mkiso/iso --arch x86_64 --repository http://liskamm.alpinelinux.uk/edge/main --profile virt --workdir /var/mkiso/work 2017-09-02 12:55:14 Also did the profile as standard, same result both images 2017-09-02 15:10:31 skazz: is that mirror up-to-date? try with cz. 2017-09-02 15:10:39 if it still appears, then yes it seems to be an error on our side i think 2017-09-02 15:53:33 hey all 2017-09-02 15:53:37 I notice that cfssl does not have a maintainer: https://git.alpinelinux.org/cgit/aports/tree/testing/cfssl/APKBUILD 2017-09-02 15:53:52 is a maintainer needed? 2017-09-02 15:58:18 if you volunteer, you are probably more than free to maintain it :) 2017-09-02 16:01:35 > is a maintainer needed? 2017-09-02 16:01:36 festivemongoose you can create a pull request claiming maintainership 2017-09-02 16:03:39 reminds me, we should doc what it means to be a maintainer 2017-09-02 16:04:05 i would also like to have a way to flag PRs that are maintainer updates 2017-09-02 16:04:27 or allow maintainers to ack a PR or patch to mailing list 2017-09-02 16:04:45 so those can be handled fast 2017-09-02 16:08:25 Yeah 2017-09-02 18:30:28 ncopa, I would like to possibly become a maintainer, however, I would like to know "what it means to be a maintainer" 2017-09-02 20:07:10 festivemongoose, it mostly means you send updates to APKBUILD when needed (usually when there's new upstream release), and take a look at related bugs 2017-09-02 20:21:50 so sometimes I see 2017-09-02 20:21:57 make -C "$builddir" 2017-09-02 20:22:01 instead of 2017-09-02 20:22:06 cd "$builddir"; make 2017-09-02 20:22:27 which is recommended for new APKBUILD? the latter seems more popular but I am looking at for instance libpng APKBUILD and it does first one 2017-09-02 20:25:36 i prefer the latter personally 2017-09-02 20:25:40 not sure if there's any strict policy on it 2017-09-02 20:33:18 explicit cd seems better, because there may be other commands working in "$builddir", and if so, then for consistency sake I would stick to it. 2017-09-02 20:35:46 -C is also not posix but it's not like anything uses posix make anyway 2017-09-02 20:40:19 shimmy doeso/ 2017-09-02 20:40:22 * o/ 2017-09-02 20:51:12 *ahem* 2017-09-02 20:51:18 why does libsm depend on e2fsprogs...? 2017-09-02 20:51:45 an X11 library really should not need an ext2 fs right? and I don't see any use of it or linkage to it.. 2017-09-02 20:54:06 one could say ext2 is a form of S&M... 2017-09-02 20:54:14 bad jokes aside, does it build without? 2017-09-02 20:54:59 yes it does 2017-09-02 20:55:18 well it doesn't, because I didn't define 'check' phase yet, and I modified abuild to die if check isn't run (instead of just warning)... but it does build 2017-09-02 20:58:48 foxkit: likely historical reasons 2017-09-02 20:59:04 foxkit: e2fsprogs used to contain libuuid, which has been split out 2017-09-02 20:59:33 kaniini: the configure phase doesn't do --with-libuuid so even with e2fsprogs-dev or libuuid-dev installed, it won't use it. 2017-09-02 20:59:52 but if that is the intention, I can fix that 2017-09-02 20:59:55 sure 2017-09-02 20:59:59 i am just saying :p 2017-09-02 21:07:33 clandmeter: any reason why freetype is !check? 2017-09-02 21:11:20 i think i was the one who did that 2017-09-02 21:11:27 and i turned it off because it was blocking builders 2017-09-02 21:15:53 is there a way to specify arch-specific dependencies without being ugly? libjpeg-turbo is pulling in nasm everywhere. it definitely cannot use x86 asm on arm/ppc/etc, haha 2017-09-02 21:16:22 it's nbd, it still works and compiles fine (and nasm supports cross, so it will even build fine on other arches). it's just waste of time. 2017-09-02 21:40:04 i think abuild should be extended for that 2017-09-02 21:52:49 ugh, half-wrote an email and then realised emails take too long; will just patch abuild later 2017-09-02 21:52:59 trying to finish this "modernise ALL the X" sprint first 2017-09-02 21:53:05 by X I mean X11 2017-09-03 11:50:26 i think community/llvm3.9 needs to be rebuilt against the latest libc++, getting relocation errors that look to be due to abi changes 2017-09-03 11:50:55 example https://gist.github.com/mitchty/76b6a94dacfde6dd46ed71aeface9527 2017-09-03 11:51:09 i'm rebuilding it locally quick to be sure 2017-09-03 20:54:14 is there any ARM hardware available for testing? 2017-09-03 20:54:20 would love to crack the icu bus error test thing 2017-09-03 20:54:33 (time for weekly git merge) 2017-09-03 20:58:23 I got some raspberry pi's :') 2017-09-03 20:58:25 pis* 2017-09-03 20:59:14 I have an rpi 1 but it does not run alpine 2017-09-03 20:59:17 foxkit, what do you need? 2017-09-03 20:59:19 maybe I can put on sdcard 2017-09-03 20:59:29 clandmeter: just something I can ssh in to and try to inspect icu test failure. 2017-09-03 20:59:47 only ARM hardware I have is rpi 1b.. not sure if that can run alpine.. if it can, I can probably set that up 2017-09-03 21:00:02 it should work 2017-09-03 21:00:16 i can get you a container to test, but im still on holidays 2017-09-03 21:00:23 considered qemu? 2017-09-03 21:00:31 either way, yeah, clandmeter is your build master 2017-09-03 21:01:32 so i have limited internet access. 2017-09-03 21:08:49 clandmeter: let me try my rpi first, enjoy your holidays :) 2017-09-03 21:09:48 foxkit: you can borrow an rpi2 of mine 2017-09-03 21:09:52 or i can spin up a qemu vm 2017-09-03 21:10:41 danieli: normally I would use qemu or similar, but my main computer is not yet on apk-based system (not sure if alpine or adelie), it is still on gentoo, and it is /broken/ badly. it needs a complete purge of /usr honestly.. 2017-09-03 21:11:47 things randomly crash and break due to bad merge of mesa libs, `iconv` has stopped working, pidgin can't send messages that are utf-8 (only ascii), firefox can't connect to ipv6 addresses.. 2017-09-03 21:11:54 it's real bad, I can't wait to be able to wipe it :) 2017-09-03 21:12:05 foxkit: I'm on debian :P 2017-09-03 21:12:05 hence why I'm trying very hard to fix up X11 packages :P 2017-09-03 21:12:12 I see.. hmm. 2017-09-03 21:12:35 right now I'm trying to edit glibc source to use some compat stuff 2017-09-03 21:13:07 has to do with msgpack and backtraces (execinfo) 2017-09-03 21:16:34 we've got libexecinfo 2017-09-03 21:16:52 we do?! 2017-09-03 21:17:10 I needed that some time ago.. tried to rip it out of gcc source but I couldn't ever make it work.. 2017-09-03 21:17:14 Shiz: i'll try it and rebuild, thanks for telling me 2017-09-03 21:17:26 :) 2017-09-03 21:17:41 we ported the BSD version 2017-09-03 21:17:50 https://git.alpinelinux.org/cgit/aports/tree/main/libexecinfo?h=3.6-stable 2017-09-03 21:18:03 kick-ass! thank you 2017-09-03 21:18:31 alright building llvm is not quick, its 7%% done, then another 6 hours to find out if i was right, arm boxes remind me of the late 90's compile time wise 2017-09-03 21:18:42 err 75% rather 2017-09-03 21:18:52 llvm is very slow yes 2017-09-03 21:18:54 clang too 2017-09-03 21:19:07 i'm just glad i don't need clang in this case :) 2017-09-03 21:19:08 main/heimdal does not seem to build here due to secfixes comment error? can anyone confirm this? 2017-09-03 21:19:22 I don't know what sexfices comment is supposed to look like so I can't confirm visually 2017-09-03 21:19:27 er secfixes 2017-09-03 21:19:37 ah yes 2017-09-03 21:19:43 foxkit: add a colon after r0 2017-09-03 21:19:46 that should fix it 2017-09-03 21:20:07 it does indeed, thank you! 2017-09-03 21:20:23 i can't push to main/ so i'll poke kaniini to do so 2017-09-03 21:20:25 ;p 2017-09-03 21:23:47 god damn it, compat/b64_ntop.c isn't building 2017-09-03 21:24:00 watcha working on? 2017-09-03 21:24:30 making a package for tmate 2017-09-03 21:24:50 basically a fork of tmux with functionality for session sharing via ssh 2017-09-03 21:25:50 also mirror geodns stuff, tinkering with tiering, status page, docs 2017-09-03 21:26:15 and a huge state-of-the-art py3.6 chatbot, but that's a lower priority right now 2017-09-03 21:28:55 https://code.foxkit.us/IRC/PyIRC contributions welcome :) 2017-09-03 21:29:29 >this library aims to follow [...] RFC2812 [..] 2017-09-03 21:29:31 slight doubt :P 2017-09-03 21:30:05 i'm working on another transport to begin with, might contribute to that once i've gotten the core and API completeness for that transport done 2017-09-03 21:30:30 i'm using "transport"s and "protocol"s as abstractions 2017-09-03 21:31:12 [my bot uses 'transport' to differentiate between IRC, Discord, etc.] 2017-09-03 21:32:34 i'm using transports to differentiate between sockets and websockets, and protocol for the protocol/platform 2017-09-03 21:34:02 i'm doing it event based, with a core event loop and pubsub (with patterns) kinda stuff 2017-09-03 21:34:34 ill do it in a bit. assembling some outdoor furniture right now 2017-09-03 21:38:05 foxkit: i'm generating the iso to build the VM using archivemount and genisoimage 2017-09-03 21:38:08 it'll be up shortly 2017-09-03 21:38:45 i'm using the alpine-uboot-3.6.2-armhf.tar.gz image 2017-09-03 21:39:48 danieli: I really appreciate this 2017-09-03 21:41:38 foxkit: which arm cpu? any preference? 2017-09-03 21:42:28 danieli: whatever 'armhf' is. I guess ARMv6? 2017-09-03 21:43:06 foxkit: https://git.alpinelinux.org/cgit/abuild/tree/functions.sh.in 2017-09-03 21:43:10 armv6 indeed 2017-09-03 21:43:36 iirc debian calls armv6 armhf 2017-09-03 21:43:48 "hard float" 2017-09-03 21:45:09 armv7* my bd 2017-09-03 21:45:10 bad* 2017-09-03 21:45:58 mv "$pkgdir"/usr/lib/slapd "$pkgdir"/usr/sbin/ || reteurn 1 2017-09-03 21:46:17 I guess it hasn't ever failed, so the 'command not found: reteurn' message has never been seen.. 2017-09-03 21:53:32 Shiz: it supports RPL_BOUNCE and their stupid WHO syntax, so, yes, 2812 is supported. just very begrudgingly and with a frowny face 2017-09-03 21:53:47 doesn't rfc2812 have a whole lot more silly stuff 2017-09-03 21:54:48 most of it deals with casemapping and funny numerics, which are handled 2017-09-03 22:11:32 bah, i cba messing around with qemu, haven't used it in years 2017-09-03 22:11:46 Shiz: any idea how i easily can run a qemu arm vm with alpine? 2017-09-03 22:12:15 just boot it with the iso mounted, i guess? 2017-09-03 22:12:20 which iso? 2017-09-03 22:12:30 there's just .tar.gz files on the downloads page 2017-09-03 22:14:10 oh right 2017-09-03 22:16:55 yep 2017-09-03 22:57:21 so OpenLDAP's test suite is still running, over one hour later on my 8-core xeon 2017-09-03 22:57:31 does that mean I should mark it !check due to time? 2017-09-03 23:53:06 so it appears postgres has discontinued their FTP service 2017-09-04 03:32:33 >>> libxdamage: Updating ./config.sub 2017-09-04 03:32:35 cp: cannot create regular file './config.sub': Permission denied 2017-09-04 03:32:46 -r-xr-xr-x 1 awilcox awilcox 35442 Jan 2 2013 config.sub 2017-09-04 03:32:50 I can confirm it is not writeable 2017-09-04 03:32:53 should it be rm'd first? 2017-09-04 06:25:57 i suppose so 2017-09-04 06:26:03 or perhaps just fix the permissions 2017-09-04 09:02:25 i dell asleep last night 2017-09-04 09:02:26 fell* 2017-09-04 09:48:02 as we all should :) 2017-09-04 09:48:38 Shiz: good day! how are you, and what's up? 2017-09-04 09:52:04 good morning :) 2017-09-04 09:52:06 just working 2017-09-04 09:52:52 alpine or work-work? 2017-09-04 10:05:43 work 2017-09-04 10:05:50 ah all right 2017-09-04 10:50:46 danieli: http://immunity.shiz.me/apk-licenses.txt 2017-09-04 10:51:20 Shiz: frequency? 2017-09-04 10:51:26 yes 2017-09-04 10:51:54 nice, i extracted all the package: license pairs 2017-09-04 10:52:22 that's a lot of "GPL" 2017-09-04 10:52:47 # apk search | rev | cut -d- -f3- | rev | xargs apk info --license | grep -vF license | grep -v '^\s*$' | sort | uniq -c | sort -nr 2017-09-04 10:53:36 Shiz: think i can assume "public domain" = CC0? 2017-09-04 10:53:44 CC0 is a public domain dedication 2017-09-04 10:53:48 i'll just check them i guess 2017-09-04 10:53:50 there are multiple public domain dedications 2017-09-04 10:53:55 indeed 2017-09-04 10:53:58 not an assumption you can make 2017-09-04 10:54:03 i'll check the packages 2017-09-04 10:54:37 the 1490 packages just labeled GPL irk me 2017-09-04 10:54:51 me too 2017-09-04 10:54:54 that's why i'm doign this 2017-09-04 10:54:55 doing* 2017-09-04 10:55:03 my fingers are cold, i need to turn up the heater :< 2017-09-04 10:56:17 maybe we can xref packages with other distro databases 2017-09-04 10:57:08 that's what i'm about to do :) 2017-09-04 10:57:16 the ones that definitely match anyway 2017-09-04 11:34:31 i bought a new keyboard 2017-09-04 11:34:40 apple magic keyboard 2017-09-04 11:34:56 this thing is fantastic 2017-09-04 11:35:27 favorite parts: when plug in the usb cable to charge it, it will appear on the USB bus as an USB keyboard 2017-09-04 11:35:38 i was able to pair it with bluetooth 2017-09-04 11:35:51 so i can use it as bluetooth keyboard too 2017-09-04 11:36:26 (my previous was other apple keyboard ove bt, but with 2 AA batteries) 2017-09-04 11:36:36 however... 2017-09-04 11:36:41 the Fn key does not work 2017-09-04 11:36:43 i was never too much a fan of the key feel on apple keyboards :( 2017-09-04 11:36:49 i love it 2017-09-04 11:36:50 Shiz: same, i'm more of a mechanical geek 2017-09-04 11:37:02 apple style chicklet can be really nice, but i prefer mechanical ones 2017-09-04 11:37:03 the macbook keyboards are fine for laptop keyboard standards, though 2017-09-04 11:37:16 and i prefer to have a numpad, or i just slice my hand through the air once in a while 2017-09-04 11:37:19 i found out why the Fn key does not work 2017-09-04 11:37:25 yeah, i like apple keyboards, except layouts 2017-09-04 11:37:31 ncopa: i'm guessing it sent the wrong keycode? 2017-09-04 11:37:37 nope 2017-09-04 11:38:00 what happens is that with bt it will use a generic keyboard driver 2017-09-04 11:38:08 but with usb it will use hid-apple 2017-09-04 11:38:15 so with ub it works as expected 2017-09-04 11:38:19 usb* 2017-09-04 11:38:22 aah okay 2017-09-04 11:38:29 but with bluetooth the Fn does not work at all 2017-09-04 11:38:44 yeah, there's some custom apple juice in the hid-apple driver 2017-09-04 11:39:42 and what happens is that with USB it will report vendor id as 0x05ac, which is USB_VENDOR_ID_APPLE 2017-09-04 11:39:53 can yo uforce it to hid-apple over bt? 2017-09-04 11:40:12 im preparing a patch 2017-09-04 11:40:29 when connecting with bt, the vendor id is 0x004c 2017-09-04 11:40:37 strange 2017-09-04 11:40:52 i think its because there are USB is and BT id 2017-09-04 11:41:00 https://www.bluetooth.com/specifications/assigned-numbers/company-identifiers 2017-09-04 11:41:16 yeah, that sounds likely 2017-09-04 11:41:16 in there the 0x004c is defined as Apple 2017-09-04 11:41:38 drivers are found using vendor + device ID, is it fucking up? 2017-09-04 11:42:09 drivers are found using vendor + device id 2017-09-04 11:42:10 correct 2017-09-04 11:42:16 and it does not find the apple id 2017-09-04 11:42:31 apple vendor id 004c 2017-09-04 11:42:44 im gonna reboot and test my patch ;) 2017-09-04 11:42:46 brb 2017-09-04 11:43:20 tyt 2017-09-04 11:47:20 hum 2017-09-04 11:47:23 didnt work 2017-09-04 11:48:25 hmm 2017-09-04 12:16:41 it didnt work because my patch was wrong 2017-09-04 12:16:52 where was the mistake? 2017-09-04 12:20:03 https://bugzilla.kernel.org/show_bug.cgi?id=99881#c38 2017-09-04 12:20:29 I used HID_USB_DEVICE(0x004c, ... 2017-09-04 12:20:42 where it should been HID_BLUETOOTH_DEVICE(.... 2017-09-04 12:21:46 aah okay, gotcha 2017-09-04 12:21:51 but thats good news, then i may make it work :) 2017-09-04 12:21:57 i've made the same mistake before :P 2017-09-04 12:22:02 yeah true 2017-09-04 12:22:15 didnt know you hacked bt kernel driver :) 2017-09-04 12:23:18 i've done a lot of RF and kernel hacking, most in terms of network stacks and other drivers related to RF comms 2017-09-04 12:23:28 cool 2017-09-04 12:23:50 i have no experience with this at all 2017-09-04 12:24:10 and very little experience with bt 2017-09-04 12:24:51 most of what i did was related to subsea communication and sensor systems - acoustic/laser modems, custom RF drivers, hacking existing drivers, porting, ++ 2017-09-04 12:25:32 tl;dr ICS, subsea comms, and custom RF impls 2017-09-04 12:25:41 sounds cool :) 2017-09-04 12:26:56 a lot of it was a massive pain in the ass, but it was something else. 2017-09-04 12:30:56 hum 2017-09-04 12:31:02 still now workey 2017-09-04 12:31:25 it should be enought to just rmmor hid-apple, right? 2017-09-04 12:31:28 tried tshark for usb? 2017-09-04 12:31:40 and it should, in theory... 2017-09-04 12:31:57 my older keyboard connects with BT though 2017-09-04 12:36:23 unrelated tip: the `mdv` python package is pure magic, to render markdown files in terminals 2017-09-04 12:37:56 or the `msee` npm package 2017-09-04 12:49:40 ok some progress here with keyboard driver 2017-09-04 12:50:39 [ 61.271210] input: Magic Keyboard as /devices/pci0000:00/0000:00:14.0/usb1/1-5/1-5.4/1-5.4.3/1-5.4.3:1.0/bluetooth/hci1/hci1:11/0005:004C:0267.000D/input/input24 2017-09-04 12:50:40 [ 61.271971] apple 0005:004C:0267.000D: input,hidraw0: BLUETOOTH HID v0.67 Keyboard [Magic Keyboard] on 00:19:0e:0c:dc:21 2017-09-04 12:51:07 at least it loads the hid-apple driver instaed of hid-generic 2017-09-04 12:51:21 i'd say that's progress :) 2017-09-04 12:51:28 Fn key still dont work 2017-09-04 12:52:30 and none of the function keys (volume, brightness, media keys) does not work at all 2017-09-04 12:53:11 try tshark on the usb port 2017-09-04 12:54:28 im gonna test the module params first 2017-09-04 12:55:14 setting fnmode to 2 made the f1-f12 keys work 2017-09-04 12:55:32 so something is apparently working 2017-09-04 12:59:17 Shiz: think "GPL-2.0+ AND GPL-3.0+ AND (Artistic-1.0 OR GPL-1.0+)" would be correct for this license file? https://github.com/networkupstools/nut/blob/master/COPYING 2017-09-04 12:59:53 yes 2017-09-04 13:00:03 okay, thanks 2017-09-04 13:00:12 just had a little doubt, needed to be sure i wasn't insane 2017-09-04 15:01:29 man 2017-09-04 15:01:32 there's a lot of incorrect licenses here 2017-09-04 17:13:57 it honestly looks like people are putting random stuff into the license="" in APKBUILDs 2017-09-04 17:45:37 Hey. Im trying to package `mlite` (a qt5 library I think) for Alpine 2017-09-04 17:45:53 <_ikke_> hello 2017-09-04 17:46:38 I have qt5 installed, but the build fails because `qdbusxml2cpp` isnt found, but it does exist under the name of `qdbusxml2cpp-qt5`, 2017-09-04 17:46:44 hi _ikke_ 2017-09-04 17:47:45 so In the build function (although I guess this should be in prepare thikning on it) in my `APKBUILD`, I tried making symlink for it, but whenever I try I get `permission denied`... 2017-09-04 17:48:51 <_ikke_> You should not alter the system in an APKBUILD anyway 2017-09-04 17:49:00 <_ikke_> abuild itself is not run as root 2017-09-04 17:49:56 Ahh, right... So how best would I go about fixing this issue _ikke_ ? I've heard I should use $pkgdir for it, but that directory doesnt exist when I try making the symlink there 2017-09-04 17:50:11 <_ikke_> Bloo[m]: You should fix the issue indeed 2017-09-04 17:51:31 <_ikke_> $pkgdir is the directory where you place the built files, where they get packaged 2017-09-04 17:52:52 <_ikke_> making the symlink in $pkgdir does not make sense if you need it to build the package, it's only if you need it after the package is installed 2017-09-04 17:53:18 _ikke_: Yeah thats what I thought.. Would I be better off patching the source to use the right package name? 2017-09-04 17:55:24 <_ikke_> Yeah, something like that 2017-09-04 17:55:52 right, Il do that then. Cheers for the run down _ikke_ :) 2017-09-04 18:32:37 danieli: I had started making script for this, thanks for your help! we should coordinate and work on it together :) 2017-09-04 18:36:49 foxkit: script for what, exactly? for the licensing stuff? 2017-09-04 18:37:34 danieli, yes. I was going to compare license values with alpine, gentoo portage, and fedora rpmspec. 2017-09-04 18:38:14 dammit, net split 2017-09-04 18:38:14 foxkit: have you written up a script for it? 2017-09-04 18:38:14 i see, i can write something 2017-09-04 18:38:14 not yet 2017-09-04 18:38:17 I had gotten as far as Shiz did 2017-09-04 18:38:52 we agreed that SPDX compliance would be nice 2017-09-04 18:38:59 https://spdx.org/licenses/ 2017-09-04 18:41:06 also ncopa https://doc.powerdns.com/md/authoritative/backend-geoip/ :) 2017-09-04 18:41:36 powerdns, according to itself, hosts 30%+ of all EU domains, and around 90% of all EU domains using DNSSEC 2017-09-04 18:41:47 its geoip backend uses yaml 2017-09-04 18:43:00 oh right, -infra 2017-09-04 18:43:03 my bad 2017-09-04 18:47:11 omg.... 2017-09-04 18:47:25 i think i got the Fn key work on the apple magic keyboard 2017-09-04 18:47:32 ncopa: congrats! :') 2017-09-04 18:47:36 <_ikke_> lol 2017-09-04 18:47:48 i'm getting the mirror dns ideas written down 2017-09-04 18:48:44 i have no idea *why* it works 2017-09-04 18:49:08 *magic* keyboard, remember? 2017-09-04 18:49:16 lol 2017-09-04 18:49:18 indeed 2017-09-04 18:49:36 i wonder if they have a black magic keyboard 2017-09-04 18:49:44 reminds me of https://github.com/endrazine/wcc tbh 2017-09-04 19:07:48 _ikke_: not sure if you can help me with this, got the patch working and its compiling fine, but fails the tests at the end 2017-09-04 19:08:11 this is due to lrelease not being found, because its named lrelease-qt5 and not lrelease... But again, because the build isnt complete, $pkgdir isnt there yet 2017-09-04 19:08:51 <_ikke_> yeah, $pkgdir only makes sense for the package() phase 2017-09-04 19:09:42 <_ikke_> Bloo[m]: can you share the APKBUILD (and anyhting else necessary)? 2017-09-04 19:10:50 _ikke_: https://hastebin.com/usupanofip.bash cheers 2017-09-04 19:11:54 searched for anything to do with `lrelease` in the project and cant find anything 2017-09-04 19:15:28 <_ikke_> I also need the patch 2017-09-04 19:15:38 oh yeah ofc 2017-09-04 19:16:03 _ikke_: https://hastebin.com/johalaziju.hs 2017-09-04 19:16:10 thats the patch 2017-09-04 19:16:11 worked a treat 2017-09-04 19:16:19 but sadly nothing to do with `lrelease` in that file either 2017-09-04 19:17:39 ncopa: over bluetooth? nice! 2017-09-04 19:18:05 <_ikke_> Bloo[m]: qt5-qtwayland cannot be found.. 2017-09-04 19:18:45 _ikke_: oh yeah sorry, this is postmarketOS 2017-09-04 19:19:03 wel, its not actually a different OS even, its Alpine OS and postmarket bootstrapper is what handles it 2017-09-04 19:19:10 so we have our own repo of packages 2017-09-04 19:21:21 <_ikke_> right 2017-09-04 19:21:30 _ikke_: https://github.com/postmarketOS/pmbootstrap/blob/master/aports/kde/qt5-qtwayland/APKBUILD 2017-09-04 19:21:35 will be upstreamed to Alpine repos eventually 2017-09-04 19:22:22 thanks PureTryOut 2017-09-04 19:23:52 <_ikke_> conflicting packages apparently 2017-09-04 19:24:05 <_ikke_> qt5-qtbase-dev and qt-dev 2017-09-04 19:25:00 Ohhh 2017-09-04 19:25:10 <_ikke_> I removed qt-dev for now 2017-09-04 19:25:25 <_ikke_> (filesystem conflicts) 2017-09-04 19:25:31 Right, so I should just uninstall qt-dev then 2017-09-04 19:26:11 but I didnt even have it installed in the first place though 2017-09-04 19:26:14 well, at least I dont think I did 2017-09-04 19:26:45 <_ikke_> I'm not saying that's your issue 2017-09-04 19:26:58 <_ikke_> that's what I ran into when trying to build qt5-qtwayland 2017-09-04 19:27:39 ohh 2017-09-04 19:28:01 I get confused easily 2017-09-04 19:28:03 lol 2017-09-04 19:28:06 qt-dev is a qt4 package right? 2017-09-04 19:28:12 yes 2017-09-04 19:28:24 all the Alpine qt5 packages start with qt5-* 2017-09-04 19:28:53 yeah thought so. I don't think qt4 is much used anymore anyway 2017-09-04 19:31:37 PureTryOut[m]: I am trying to deprecate it, bumping last packages that use it 2017-09-04 19:32:00 it has been EOL for two years 2017-09-04 19:32:07 it has no business being in main 2017-09-04 19:32:13 (imo) 2017-09-04 19:32:37 foxkit: yeah makes sense 2017-09-04 19:32:49 but first I have to build qt5... 2017-09-04 19:32:53 which requires mesa... 2017-09-04 19:36:59 I must say I find the `qt5-qt` namingscheme a bit weird though. why not just `qt5-`? 2017-09-04 19:37:29 <_ikke_> ACTION agrees 2017-09-04 19:37:35 I wondered the same 2017-09-04 19:38:36 me aswell gracias 2017-09-04 19:39:30 <_ikke_> Bloo[m]: built wayland, now trying mlite agian 2017-09-04 19:39:40 Thanks! 2017-09-04 19:39:52 Trying to figure it out, but cant seem to find much on it 2017-09-04 19:40:16 unless there might be some osrt of argument that can be supplied to `make` to change lrelease 2017-09-04 19:42:26 interesting: https://groups.google.com/d/msg/golden-cheetah-users/H8r2M4hu9L4/mVy7Qfv0UcUJ 2017-09-04 19:45:59 <_ikke_> Bloo[m]: tests/ut_mdesktopentry.pro contains lrelease 2017-09-04 19:46:44 wtf, grep didnt find that at all 2017-09-04 19:47:20 _ikke_: well thanks a lot! Sorry about that, I did try doing some searches but none of them worked for some reason 2017-09-04 19:47:29 opefully thatl get things moving again! 2017-09-04 19:48:08 <_ikke_> Bloo[m]: aren't you just missing the qt5-qttools-dev dependency? 2017-09-04 19:48:15 <_ikke_> it contains plain lrelease 2017-09-04 19:48:27 <_ikke_> or has that been renamed under PostmarketOS 2017-09-04 19:48:51 _ikke_: think it might be renamed or something as I have that already 2017-09-04 19:49:12 <_ikke_> might be that they did that to be able to have both qt4 and qt5 2017-09-04 19:49:16 *postmarketOS, small p, big OS 😉 2017-09-04 19:49:20 <_ikke_> ah ok 2017-09-04 19:55:25 foxkit: iirc you suggested getting rid of qt4 2017-09-04 19:55:29 and had a plan for it 2017-09-04 19:55:39 we need to do that 2017-09-04 19:56:01 i think wpa_supplicant was the reason qt stayed in main 2017-09-04 19:56:56 ncopa: if wpa_gui is really needed, then we can use qt5 2017-09-04 19:57:18 yeah 2017-09-04 19:57:19 how about making wpa_gui a separate package from wpa_supplicant... 2017-09-04 19:57:30 it is a separate package 2017-09-04 19:57:31 this 2017-09-04 19:57:34 just built at same time 2017-09-04 20:30:15 ncopa: yeah. I'm trying to build qt5 locally so I can port everything over. 2017-09-04 20:30:37 requires some patience... 2017-09-04 20:30:55 ncopa: I have modified abuild locally to die on no-tests instead of warn. so.. more than some.. 2017-09-04 20:31:05 I've done about 80 X11-related packages now 2017-09-04 20:31:11 I'd say 30 of them are !check, and the rest have test suites 2017-09-04 20:31:25 but that's all approx. I haven't done any formal counting 2017-09-04 20:32:16 i think we owe you a beer for each :) 2017-09-04 20:32:17 testing is a big part of the Adélie project because that is how we found many bugs that we otherwise would not have known. and especially on arches nobody tests on, like ppc. 2017-09-04 20:32:39 but I'm doing it the "right way" so that it can be provided to Alpine too 2017-09-04 20:32:42 yes, we enabled make check right before 3.6 2017-09-04 20:32:47 I hope the work is useful 2017-09-04 20:32:50 and it turned out to be very good 2017-09-04 20:33:03 i believe it is 2017-09-04 20:33:11 would probably mass-sed everything otherwise XD 2017-09-04 20:40:33 how do i build wpa_supplicgant with qt5? 2017-09-04 20:40:51 there is a dir wpa_gui-qt4 2017-09-04 20:44:36 why would you? :P 2017-09-04 20:45:30 ncopa: it 'magically' detects which version of qt you have installed 2017-09-04 20:45:37 ncopa: the 'qt4' dir is really 'qt' dir 2017-09-04 20:47:52 https://gitweb.gentoo.org/repo/gentoo.git/tree/net-wireless/wpa_supplicant/wpa_supplicant-2.6-r2.ebuild#n284 for instance 2017-09-04 20:48:03 the fedora rpmspec still uses qt4, and I can't find the debian one 2017-09-04 20:48:23 but yeah you just call qmake 5 instead of qmake 4 on it. the source is compatible with either. 2017-09-04 20:52:54 https://anonscm.debian.org/cgit/collab-maint/wpa.git/tree/debian/rules debian also uses qmake_qt4 it seems 2017-09-04 20:53:14 i suppose the -qt4 suffix made me think it was only qt4 2017-09-04 20:53:24 it does work with qmake-qt5 2017-09-04 20:54:56 pushed 2017-09-04 20:55:05 those are left in main: http://tpaste.us/8QZ6 2017-09-04 20:57:30 Woop! 2017-09-04 20:57:36 Got it compiling 2017-09-04 20:57:51 ACTION sent a long message: Bloo[m]_2017-09-04_20:57:50.txt  2017-09-04 21:00:46 Bloo[m]: it just installs it in wrong location 2017-09-04 21:01:29 ncopa: shouldnt this suffice? `make INSTALL_ROOT="$pkgdir" install`? 2017-09-04 21:01:47 might need set PREFIX=/usr or similar 2017-09-04 21:02:07 my guess is that prefx=/usr/local 2017-09-04 21:02:22 grep /usr/local Makefile 2017-09-04 21:06:36 ncopa: https://hastebin.com/amatihegas.bash heres my APKBUILD btw. Did try setting PREFIX=/usr/local and /usr but had no baring on the results sadly 2017-09-04 21:09:27 Bloo[m]: looks like prefix should be correct 2017-09-04 21:09:32 did you build qt5 yourself? 2017-09-04 21:09:51 ncopa: using ones from my repo 2017-09-04 21:09:56 well the repo* 2017-09-04 21:10:10 not sure if its from ours or not (postmarketOS's repos) 2017-09-04 21:10:33 we don't have qt5 packaged, except for qt5-qtwayland 2017-09-04 21:10:38 ohhh 2017-09-04 21:10:44 so no, you did not build it yourself, you got it from the Alpine repos 2017-09-04 21:10:54 Right 2017-09-04 21:10:57 -prefix /usr \ 2017-09-04 21:11:09 i dunno 2017-09-04 21:11:14 hum 2017-09-04 21:11:20 i think i messed up wpa_supplicant 2017-09-04 21:11:35 which depends on qt5 2017-09-04 21:11:45 which is in community 2017-09-04 21:12:18 looks like we need have a separate aport for wpa gui 2017-09-04 21:12:25 or drop it 2017-09-04 21:13:44 thanks ncopa il test that now 2017-09-04 21:17:21 well, using `make INSTALL_ROOT="$pkgdir" -prefix /usr \ install ` 2017-09-04 21:17:33 I get this: http://paste.debian.net/984448/ at least its a different output 2017-09-04 21:23:54 tried `DESTDIR` instead of `INSTALL_ROOT` and got this instead: http://paste.debian.net/984450/ 2017-09-04 21:23:59 slightly more hopeful? I guess... 2017-09-04 21:27:43 ACtually, whats the difference between `DESTDIR` and `INSTALL_ROOT`? 2017-09-04 21:35:18 I just end up getting `mkdir: can't create directory '/usr/include/mlite5': Permission denied` sadly 2017-09-04 21:38:25 ncopa: any ideas? Sorry to ping you! 2017-09-04 22:38:56 kaniini: ncopa: do you think adding a check for *.service files in abuild would be a good idea? 2017-09-04 22:39:03 we've got far too much stray systemd service files in package atm 2017-09-05 07:39:27 god morgen ncopa :) 2017-09-05 07:39:39 jeg holder på å sette opp den nye maskinen. hva skal hostname på den være? 2017-09-05 07:44:19 #alpine-eekhoorntjesbrood 2017-09-05 07:44:55 scadu: why squirrel bread 2017-09-05 07:45:13 je bent zo raar 2017-09-05 07:45:51 svennebanan.se? 2017-09-05 07:46:13 danieli: why not, actually. anyway → -offtopic :P 2017-09-05 07:47:21 it turns out I missed some core packages during bootstrap ;.; 2017-09-05 07:47:51 perl, texinfo, and gcc don't have test suites or !check. I am disappointed in myself 2017-09-05 08:33:19 i think we need a package for pandoc. anything i have to take care of for haskell stuff? 2017-09-05 08:33:57 im sure there is a reason why there is no pandoc package ;-) 2017-09-05 08:46:38 I wouldn't mind Haskell packages! 2017-09-05 10:28:17 [edge] to be able to start chromium, which already pulled 64 packages beside chromium, in headless mode I still needed to install eudev-libs (FATAL:udev_loader.cc(38)] Check failed: false.), ttf-opensans (FATAL:platform_font_linux.cc(63)] Check failed: typeface. Could not find any font: sans, sans). 2017-09-05 10:30:42 was going to ping clandmeter, as he is the official maintainer, according to APKBUILD, but reality seems to differ 2017-09-05 10:32:31 so I'll only ping the last one touching it: ncopa ;) 2017-09-05 10:33:56 not sure whether we want to add even more deps by default or simply issue somewhere helpful warning regarding additional packages needed for headless mode. well, maybe it's not only headless mode problem, not sure, because I don't use alpine w/ X. 2017-09-05 10:40:51 przemoc: interesting 2017-09-05 10:41:02 i wonder why eudev-libs was not pulled in automatic 2017-09-05 10:41:06 maybe they dlopen it? 2017-09-05 10:42:16 the ttf-opensans... maybe it works with other font? 2017-09-05 10:42:24 im ok to add them as explicit depends 2017-09-05 10:43:22 lib: quite likely (doubt there is any other explanation), ttf: quite likely (just installed first sensibly looking result from apk search sans ;>) 2017-09-05 10:44:05 I'm complete noob when it comes to X-related stuff on AL 2017-09-05 10:54:51 jirutka: regarding "main/postgresql: build with -O2 instead of -Os" and "It should increase performance a bit". well, without any tests it's often hard to tell. and can vary greatly between archs and programs. from my past experience -Os often gave faster execution for arms or ppcs. I never tried building postgres with different opt flags, though. 2017-09-05 10:58:01 (but I weren't using gcc 6 then, so maybe there were improvements in that regard and -O2 is always faster than -Os on any platform, but somehow I doubt it) 2017-09-05 11:16:03 having an issue where building `pdns` isn't finding `yaml-cpp` libs, they're in /usr/lib/ 2017-09-05 11:16:20 libyaml-cpp.so.0.5 and libyaml-cpp.so.0.5.3 2017-09-05 11:17:53 it possibly wants libyaml-cpp.so 2017-09-05 11:18:09 and you maybe installed only -libs that doesn't provide it (just guessing) 2017-09-05 11:18:25 it explicitly asks for libyaml-cpp 2017-09-05 11:18:28 yaml-cpp* 2017-09-05 11:18:42 yaml-dev perhaps 2017-09-05 11:18:54 should add the .so files 2017-09-05 11:19:56 nope, nada 2017-09-05 11:20:36 installing yaml-cpp-dev didn't help? 2017-09-05 11:20:54 let's see 2017-09-05 11:21:01 got it, thank you przemoc 2017-09-05 11:21:22 thought i'd ask instead of pound my head into a wall, i appreciate it 2017-09-05 11:22:15 how do i specify dependencies for a sub-package in an APKBUILD file? 2017-09-05 11:24:13 in their packaging functions 2017-09-05 11:24:38 feel free to override just about any parameters inside a package function 2017-09-05 11:24:51 ah, gotcha - thanks 2017-09-05 11:27:07 i figured it out, this packages uses a backend_$something() { _mv_backend $subpackage $depends; } structure 2017-09-05 11:28:45 ncopa: just out of curiosity, what was wrong with gdb 8? it simply didn't build or it wasn't working correctly? 2017-09-05 11:31:09 I also think that before AL 3.7 gdb should start using python3, and maybe, just maybe, provide gdb-python2 (like debian does) if it may seem crucial to anybody 2017-09-05 11:38:54 przemoc: i dont know 2017-09-05 11:39:10 yeah, we should use python3 2017-09-05 11:39:41 agreed 2017-09-05 11:40:56 ncopa: sorry, I mislabelled the commit reverting it to you, but it kaniini 2017-09-05 11:41:09 kaniini: what was wrong with gdb 8? 2017-09-05 11:41:11 kaniini: why was gdb 8 reverted? 53d4afb9e Revert "main/gdb: upgrade to 8.0" 2017-09-05 11:41:58 (for future it may be good to add some, even explanations, to reverts) 2017-09-05 11:42:22 s/even explanations/even short, explanations/ 2017-09-05 11:42:32 <_ikke_> I agree 2017-09-05 11:43:14 i'm gonna go through a ton of forensics packages and try getting them into alpine 2017-09-05 11:45:11 fine. remember to diligently push needed fixes (if there are any) upstream. 2017-09-05 11:46:12 of course :) i've done some package maintaining on other distros 2017-09-05 11:47:33 ok, seeing your question regarding lacking .so just wasn't sure how well-informed you are 2017-09-05 11:47:50 not too familiar with the package structure on alpine, still getting familiarized with alpine in general 2017-09-05 11:50:10 sure, that's fine. e.g. this -dev -libs is actually very similar to other distros, others name it slightly differently only, debian's -libs is lib*, but it also doesn't install .so, so you need -dev, but they name it again lib*-dev 2017-09-05 11:50:34 i noticed 2017-09-05 11:52:03 both naming conventions have advantages and downsides, but keeping original project name as base name seems nicer in the end for maintainers, but for clueless user lib* seems definitely better 2017-09-05 11:53:02 the "lib" prefix doesn't matter when there's a -lib suffix 2017-09-05 13:11:29 it would be nice if someone could test & bump chromium to 60.*. its chromedriver (2.29) has some bugs that are supposedly fixed in newer versions. chromedriver binaries from google (http://chromedriver.storage.googleapis.com/index.html) aren't meant for musl users. 2017-09-05 13:11:43 s/its/current/ 2017-09-05 13:12:03 lots of new stuff in 60 2017-09-05 13:12:11 iirc they changed up the dev tools too 2017-09-05 13:13:20 chromium upgrade is practically never painless, maybe only if x.y.patch part of the version is changed ;) 2017-09-05 13:26:01 maybe it would be better to provide chromedriver separately in AL, but apparently they don't tag it in any way within chromium repo. 2017-09-05 13:26:37 there is someone who provides his own repo with extracted chromedriver: https://github.com/bayandin/chromedriver 2017-09-05 13:27:02 and tags releases according to commits mentioning release 2017-09-05 13:28:31 latest one (2.32) still supports v59 2017-09-05 13:38:14 ok, it seems to be unbuildable when strimmed like that, so you need whole chromium 2017-09-05 13:39:00 s/str/tr/ 2017-09-05 13:39:13 ACTION 's mistake ratio is definitely too high today 2017-09-05 13:55:11 i forget. it was probably blocking builders 2017-09-05 14:20:25 przemoc: i built chromium 60 some time ago but the tabs crashed 2017-09-05 14:20:38 i havent had time to investigate it 2017-09-05 14:22:40 http://tpaste.us/mMXE 2017-09-05 14:30:02 ncopa: is it possible to build the same package, stable in v3.6 and a release candidate in edge? how do I do that? 2017-09-05 14:30:10 just add the new package to edge? 2017-09-05 14:32:15 or.. how is it done? not familiar with the ecosystem on this subject 2017-09-05 14:33:00 how do you mean? 2017-09-05 14:34:09 update the one in edge, basically - how is versioning done? 2017-09-05 14:34:19 how does it know which one to retain in v3.6? is it done by git branches? 2017-09-05 14:40:01 yes 2017-09-05 14:40:15 every 6 months we do stable x.y.0 release 2017-09-05 14:40:25 and at that time we create x.y-stable git branch 2017-09-05 14:40:38 the stable builders are configured to track that git branch 2017-09-05 14:41:46 aah, all right 2017-09-05 14:42:16 ncopa: thanks, but I'm X-less w/ AL, so cannot help either. 2017-09-05 14:43:01 also, shouldn't 3.6.3 be coming soon? 2017-09-05 14:43:47 or there won't be 3.6.3 before 3.7? 2017-09-05 15:18:10 ok so my test was more successful after rebuilding llvm3.9, I'm guessing community/llvm37 and community/llvm39 should be rebuilt for all arches due to libc++ changing for 4.0, as it stands there are a ton of linker errors 2017-09-05 15:27:06 https://gist.github.com/mitchty/76b6a94dacfde6dd46ed71aeface9527 as an example of what you see when you try linking to libLLVM-3.9.so, the libc++ changes cause some "fun" symbol not found errors 2017-09-05 15:33:18 ncopa: sent a patch for alpine-mksite 2017-09-05 17:03:01 ola guys, need support:) iam using this dockerfile https://github.com/xsteadfastx/dockerfiles/blob/master/mopidy/mopidy/Dockerfile, just added python-dev, but when i run mopidy, warrning shows up (gst-plugin-scanner:10): GStreamer-WARNING **: Failed to load plugin '/usr/lib/gstreamer-1.0/libgstdtls.so': Error relocating /usr/lib/gstreamer-1.0/libgstdtls.so: BIO_set_data: symbol not found. Its because of missing glibc ? gst-pl 2017-09-05 17:03:13 ldd output: https://pastebin.com/K3kKYc5s 2017-09-05 17:04:04 for support, i suggest #alpine-linux 2017-09-05 17:04:06 let me see, one sec 2017-09-05 17:05:02 asdf_: do you have libressl-dev? 2017-09-05 17:09:50 still same error :( 2017-09-05 17:11:42 i'm testing 2017-09-05 17:37:52 danieli: any progress? 2017-09-05 17:41:37 had to finish something, dorry 2017-09-05 17:43:38 itsok:) 2017-09-05 17:44:16 asdf_: i did `apk add python2-dev libressl-dev` before running `mopidy` 2017-09-05 17:44:17 https://p.duniel.no/4Nar 2017-09-05 17:45:14 oh... 2017-09-05 17:45:20 i have python-dev and libressl-dev 2017-09-05 17:46:00 asdf_: there's no "python-dev" in edge 2017-09-05 17:46:03 https://pkgs.alpinelinux.org/packages?name=python-dev&branch=edge&repo=&arch=&maintainer= 2017-09-05 22:31:06 Anyone here know what the difference between `INSTALLDIR` and `INSTALL_ROOT` is? 2017-09-05 22:31:53 Using `INSTALLDIR` on my APKBUILD gives me permission denied errors, `INSTALL_ROOT` gives me the error that you can use the /usr or /sys direcotyr for packages or something 2017-09-05 22:46:11 `DESTDIR`* 2017-09-05 23:04:16 yeh, the latter is an enforcement of the filesystem hierarchy standard; you're not supposed to install stuff from packages into /sys and a couple other places 2017-09-05 23:04:49 but if you absolutely have to do that, you can use, if my memory serves me correctly, options="!fhs" to override that check 2017-09-06 04:01:11 I am thinking of squashing all my *proto commits for X11 packages since they all do the same thing (modernise, add util-macros to makedepends, add -doc where missing, change all arch to noarch where necessary, and mark as not having test suite) 2017-09-06 04:01:20 and if I don't, the X11 revamp is approaching 100 commits... 2017-09-06 04:01:26 about 40 of them are *proto packages 2017-09-06 04:02:24 I also have version bump for gobject-introspection and glib; they just released yesterday, and pass all tests 2017-09-06 12:03:02 RE: http://bugs.alpinelinux.org/issues/7614 2017-09-06 12:03:19 i think this should be fixed upstream 2017-09-06 12:04:11 fcolista: agreed 2017-09-06 12:15:43 done 2017-09-06 12:15:44 https://github.com/Atoptool/atop/pull/7 2017-09-06 12:17:48 nice! 2017-09-06 12:18:15 fcolista: any idea what .~tmp~ in the repos are? 2017-09-06 12:18:23 i've seen it somewhere in code but can't remember 2017-09-06 12:18:42 where have you seen this? 2017-09-06 12:18:52 syncing a mirror and indexing the files in it 2017-09-06 12:19:00 2017/09/06 12:17:40.277 UTC /edge/main/x86_64/.~tmp~/xtables-addons-hardened-4.9.44-r0.apk: SHA1 d3eb75b0dbb35e700b346f305d53b414b5408072 2017-09-06 12:19:18 something rsync related perhaps 2017-09-06 12:19:19 maybe rsync ? 2017-09-06 12:19:25 yes TBB 2017-09-06 12:19:28 i think so 2017-09-06 12:19:28 hmm, that might actually be 2017-09-06 12:19:51 i'll reindex once rsync is done 2017-09-06 12:19:58 setting up mirrorbits on uk.a.o 2017-09-06 12:20:10 you can avoid this by using --inplace 2017-09-06 12:20:17 https://stackoverflow.com/questions/22626214/rsync-without-creating-hidden-file-in-destination 2017-09-06 12:20:29 it was indexing while rsync was running the initial sync 2017-09-06 12:20:33 it seems the default behaviour of rsync 2017-09-06 12:20:34 i'll add --inplace to the rsync args 2017-09-06 12:20:50 I just had a quick read 2017-09-06 12:20:58 so, check better first 2017-09-06 12:21:12 yeah, i'll read up on it 2017-09-06 12:21:19 google-fu is a wonderful thing 2017-09-06 12:21:21 since you can also specify the tempdir to use 2017-09-06 12:21:21 --temp-dir=../tmp 2017-09-06 12:42:35 re: http://build.alpinelinux.org/buildlogs/build-edge-aarch64/testing/filebeat/filebeat-5.5.2-r0.log 2017-09-06 12:42:42 i wonder if go should depend on binutils-gold 2017-09-06 12:42:55 seems to be: https://github.com/nodejs/node/issues/4212#issuecomment-217904671 2017-09-06 16:50:40 fcolista, in https://github.com/Atoptool/atop/pull/7 ^^ maybe use pidof if possible 2017-09-06 21:05:25 TBB: did you manage to avoid the hot icelandic law ladies last night? ;) 2017-09-06 21:10:59 icelandic law ladies? 2017-09-06 21:11:04 wut 2017-09-06 21:12:03 ACTION is icelandic and norwegian 2017-09-06 21:32:27 Ok so next challange in relation to mlite 2017-09-06 21:33:18 I have it compiled, works great... Next issue is that my next package (`lipstick`) requires mlite5 (which is what I compiled, but did set the pkgname to mlite5 because its called mlite in the sources for some reason), 2017-09-06 21:33:37 now even though mlite5 is installed, I get an error tha tit cant find the files, even though they are there 2017-09-06 21:34:44 `mlite5 development package not found` is what I get 2017-09-06 21:34:48 despite the fact that mlite5 is in the directories it should be iun 2017-09-06 21:37:51 for example, `/usr/include/mlite5/` exists 2017-09-06 22:20:34 Bloo[m]: unfortunately, yes, there were a couple who were eyeing me in a manner that suggested to go talk to them, but the whole night was so busy I just couldn't. and none of them returned tonight :( 2017-09-06 22:22:08 ;c bummer! 2017-09-06 22:22:44 TBB: hopefully some better luck in future! lol 2017-09-06 22:23:11 my regular bar/poolhall is unfortunately of the kind where everything works against me in that regard :D 2017-09-06 22:23:43 I've got just about a hundred stories of how I've met girls there and how the environment has been like "oh that can't be allowed to happen" 2017-09-06 22:56:49 TBB: similar experience here... Its a pain in the ass! 2017-09-07 06:18:38 vkrishn, pidof is still part of procps, as well as "ps -p". While with ps|grep you should not rely on procps 2017-09-07 07:16:23 o/ 2017-09-07 07:22:11 fcolista, isn't it part of BusyBox too? 2017-09-07 08:22:04 clandmeter oh 2017-09-07 08:22:08 you are right :) 2017-09-07 08:22:15 I didn't know that :) 2017-09-07 08:36:51 fcolista, I also pm'd some txt, hope your irc client supports it 2017-09-07 08:37:40 vkrishn, yeah, got it, but you were not online 2017-09-07 08:37:47 "fcolista> vkrishn, pidof is still part of procps, as well as "ps -p". While with ps|grep you should not rely on procps" 2017-09-07 08:38:01 but clandmeter pointed out that pidof is also part of bb 2017-09-07 08:38:06 I didn't know that 2017-09-07 08:38:09 so, you're right 2017-09-07 08:38:20 i'll send an update 2017-09-07 08:38:21 thx 2017-09-07 08:39:00 nope, this... 2017-09-07 08:39:02 what if atop's pid is 123 and 2017-09-07 08:39:02 ps | grep `cat $PIDFILE` would list pids for 123, 1234... etc 2017-09-07 08:39:10 ;) 2017-09-07 08:39:30 pidof is just alt suggestion 2017-09-07 08:39:34 yes, that's a bug with ps|grep 2017-09-07 08:39:43 yeah i noticed that 2017-09-07 08:39:47 ok 2017-09-07 08:39:49 thx 2017-09-07 09:56:46 clandmeter: tpaste.us is broken btw 2017-09-07 09:57:08 the data: url doesn't work directly in anchor tags in newer chrom(e|ium) releases 2017-09-07 11:15:52 I know 2017-09-07 11:15:59 I will fix it soon 2017-09-07 11:16:21 But not now 😎 2017-09-07 11:20:41 Or send a PR 2017-09-07 11:20:47 clandmeter: will do 2017-09-07 11:21:16 I can merge it from webif 2017-09-07 11:21:52 i forked resprunge too, added a /paste endpoint that serves the html 2017-09-07 11:22:33 I saw it 2017-09-07 11:24:25 Should work just fine 2017-09-07 13:11:02 how do i have an incorrectly flagged package unflagged? 2017-09-07 13:11:20 clandmeter: ^^^ 2017-09-07 13:11:27 firefox-esr in this case 2017-09-07 13:11:39 i got an email about it 2017-09-07 13:11:46 the person said sorry 2017-09-07 13:11:48 whoever flagged it confused -esr with normal 2017-09-07 13:11:49 ah okay 2017-09-07 13:12:03 yeah, latest esr is 52.3.0, not 55.x 2017-09-07 13:12:53 i suppose we can wait til ff 52.4.0 is out 2017-09-07 13:13:02 yeah, that works 2017-09-07 14:14:47 what deems a package unmaintained? 2017-09-07 14:14:56 like, what is the process? 2017-09-07 14:15:46 from what I've seen on the Github PR's, just remove the commented line on top of the APKBUILD 2017-09-07 14:17:36 PureTryOut[m]: I mean on alpine 2017-09-07 14:17:41 oh that 2017-09-07 14:17:52 uhm, there's lots of packages in aports/unmaintained that has a maintainer field 2017-09-07 14:18:10 ooh sorry I missed that repo 2017-09-07 14:18:23 aports is one repo with subdirs 2017-09-07 14:18:43 anyway, 739 out of 743 packages in unmaintained has a Maintainer: name comment at the top 2017-09-07 14:23:45 danieli: unmaintained packages don't have a maintainer, the maintainer commet usually indicates the previous maintainer of that package. In most cases the comment is removed when moving the package to unmaintained/ 2017-09-07 14:23:49 the difference between unmaintained/, testing/, main/ and community/ is that packages in unmaintained/ are not built on the builders 2017-09-07 14:23:58 *isn't 2017-09-07 14:24:45 nmeum: I see, so it's manually done 2017-09-07 14:24:48 and yeah, gathered as much 2017-09-07 14:46:38 _ikke_: i fixed zabbix 2017-09-07 14:46:56 3.4.1 :) 2017-09-07 14:47:12 no sqlite though, it's not supported as a main backend, only for proxies 2017-09-07 14:47:40 configure: error: SQLite is not supported as a main Zabbix database backend. 2017-09-07 14:53:09 <_ikke_> danieli: nie 2017-09-07 14:53:12 <_ikke_> nice 2017-09-07 14:53:17 <_ikke_> I would have looking into it otherwise 2017-09-07 14:53:24 <_ikke_> and yes, sqlite for backend is known 2017-09-07 14:53:29 <_ikke_> not a good idea anyway imo 2017-09-07 14:53:31 just gotta fix sqlite3 for agent atm 2017-09-07 14:53:35 yeah true 2017-09-07 14:53:36 <_ikke_> agent? 2017-09-07 14:53:47 <_ikke_> iirc, agent does not do anything with sqlite3 (only proxy) 2017-09-07 14:54:09 I meant proxy 2017-09-07 14:54:11 my bad 2017-09-07 14:58:48 it's ready for patching and building 2017-09-07 14:58:50 @ _ikke_ 2017-09-07 15:03:13 <_ikke_> danieli: nice, you have a link to the APKBUILD? 2017-09-07 15:03:25 uhm, i sent the patch to aports 2017-09-07 15:03:29 i'll paste it 2017-09-07 15:03:44 APKBUILD: https://p.duniel.no/AZbZ 2017-09-07 15:04:10 i build tested it a couple times to figure out missing deps and issues 2017-09-07 15:11:32 <_ikke_> danieli: thansk 2017-09-07 15:11:46 <_ikke_> Then I can build it myself 2017-09-07 15:11:57 :) 2017-09-07 15:18:03 ncopa: mind pushing the patch to alpine-mksite? 2017-09-07 15:18:12 it was sent from d@duniel.no to ncopa@a.o 2017-09-07 15:29:34 thanks! 2017-09-07 15:29:46 danieli: pushed 2017-09-07 15:30:11 I saw :) 2017-09-07 20:11:52 ncopa: is it possible to exclude infra/mirrors from #alpine-commits? it's not really relevant there 2017-09-07 20:36:03 . 2017-09-07 20:37:11 <_ikke_> .. 2017-09-08 07:55:24 Hi all, I'd like to help with wiki documentation/proofreads. Is there a guideline of some sort, or should I just dive in and edit? 2017-09-08 08:27:58 radsy: just dive in and edit 2017-09-08 09:00:15 ncopa: thanks 2017-09-08 12:19:51 yeah, it's gonna be a bit restructured soon 2017-09-08 12:19:53 but dive in 2017-09-08 12:46:16 can i trigger a rebuild of a package? 2017-09-08 12:46:57 why? 2017-09-08 12:46:59 there are some problems with the fish package and a fish dev and i tried to debug it. we cant reproduce the problem with a own built package... but with the one on the repos 2017-09-08 12:47:07 https://github.com/fish-shell/fish-shell/issues/4387 2017-09-08 12:47:11 I'll test 2017-09-08 12:48:46 If I could ask a maintainer to take a look at my PR (https://github.com/alpinelinux/aports/pull/2217), it would be much appreciated. 2017-09-08 12:48:59 thanx danieli 2017-09-08 12:49:24 chambbj: that's just a mirror, make a git patch and send it to alpine-aports@lists.alpinelinux.org 2017-09-08 12:49:43 clone git://git.alpinelinux.org/aports 2017-09-08 12:49:56 github PRs are good, we do accept those too 2017-09-08 12:50:35 I can go either way. The GH PRs had been working, so I stuck with that, but if one is preferred... 2017-09-08 12:50:52 works for me 2017-09-08 12:50:52 preference depends on developer, i try to look at both 2017-09-08 12:50:56 some others look at github only 2017-09-08 12:51:03 and others at mailinglist only :) 2017-09-08 12:56:02 Thanks as always @fabled. I'll address your comments shortly. 2017-09-08 12:57:57 xsteadfastx: https://i.imgur.com/EEXRrAV.png 2017-09-08 12:58:05 that's what I'm getting with the package from the alpine builders 2017-09-08 12:58:15 I built it myself, works fine - it does seem to need a rebuild 2017-09-08 12:59:24 the build logs aren't in http://build.alpinelinux.org/buildlogs/build-edge-x86_64/main/ either 2017-09-08 13:03:39 fabled: ncopa: either one of you mind triggering a rebuild of the 'fish' package when you have time? 2017-09-08 13:14:00 danieli: thats what i think so too 2017-09-08 14:45:55 is there a telnet client in alpine? a `telnet`, not netcat 2017-09-08 14:48:17 there's one in busybox 2017-09-08 14:49:18 thanks, was just wondering if there was one 2017-09-08 14:49:20 it's in busybox-extras 2017-09-08 17:30:27 hi it seems checkbashisms does not work with perl 5.26, somebody else that have noticed this? 2017-09-08 17:32:48 the problem with checkbashisms is the same as this https://lists.gnu.org/archive/html/bug-automake/2017-06/msg00002.html 2017-09-08 18:09:56 ncopa: attempted a fix for checkpatch in this PR https://github.com/alpinelinux/aports/pull/2284 2017-09-08 18:11:38 ncopa: sorry checkbashisms ... 2017-09-08 19:21:40 Compiling something new for another package and Im getting this error: 2017-09-08 19:21:41 protocol.c:289:10: error: implicit declaration of function 'writev' [-Werror=implicit-function-declaration] 2017-09-08 19:22:57 https://git.merproject.org/mer-core/libdsme/tree/master is what Im trying to port 2017-09-08 19:44:03 that error usually means its trying to call a library without importing it 2017-09-08 19:44:10 or including it 2017-09-08 19:45:55 Bloo[m]: try adding #include to the protocol.c under line 40 2017-09-08 19:46:25 arch3y: yeah thanks, managed to solve it 2017-09-08 19:46:35 Bloo[m]: no worries 2017-09-08 19:46:39 just thought Id help out 2017-09-08 19:47:02 thanks Making some good progress :D 2017-09-08 19:47:38 yep shame on the dev for not explicitly including it 2017-09-08 20:34:08 <_ikke_> danieli: looks like zabbix built on 3.6 :-) 2017-09-08 20:39:50 _ikke_: nice! 2017-09-08 20:39:55 i tested it on edge 2017-09-08 20:40:00 <_ikke_> Right 2017-09-08 20:41:59 the mirror http://dl.alpinelinux.org/ (can append /alpine instead) is the http redirector btw 2017-09-08 20:42:27 <_ikke_> nice 2017-09-08 20:48:36 brb setting up a bouncer 2017-09-08 20:50:53 <_ikke_> weird, I installed 3.4.1, but it still reports to be 3.2 2017-09-08 20:51:57 wat, strange 2017-09-08 20:52:09 it should pull the correct version, right? 2017-09-08 20:52:13 <_ikke_> ah, d'oh 2017-09-08 20:52:25 <_ikke_> It's only the common package I updated :D 2017-09-08 20:52:31 aah hahaha 2017-09-08 20:53:32 brb 2017-09-08 21:27:26 https://hastebin.com/rujujezomi.coffeescript 2017-09-08 21:27:44 that dreaded `Packages must not put anything under /srv, /usr/local or /opt` error :( 2017-09-08 21:28:40 ohh 2017-09-08 21:28:45 <_ikke_> Bloo[m]: Did you forget to set a prefix? 2017-09-08 21:28:54 _ikke_: exactly that :p 2017-09-08 21:29:07 thats $pkgdir right? 2017-09-08 21:30:48 <_ikke_> No 2017-09-08 21:30:53 <_ikke_> /usr 2017-09-08 21:31:04 <_ikke_> DESTDIR = $pkgdir 2017-09-08 21:33:27 <_ikke_> danieli: upgrade succeeded :-) 2017-09-08 21:34:53 _ikke_: nice! 2017-09-08 21:34:57 i set up znc, finally 2017-09-08 21:35:15 <_ikke_> ha 2017-09-08 21:35:27 <_ikke_> I have a makeshift bouncer :-) 2017-09-08 21:35:33 i would do that but lazy 2017-09-08 21:38:03 Hell. Yess! 2017-09-08 21:38:10 Thanks for the reminder, _ikke_ :) 2017-09-08 21:38:30 Im backing my way out of the infinite porting tunnel of hell 2017-09-08 21:38:31 bit by bit 2017-09-08 22:12:44 god damn it! Project ERROR: mlite5 development package not found 2017-09-08 22:12:57 I've built it and everything :c 2017-09-08 22:13:53 but did you install it? :') 2017-09-08 22:14:54 danieli: ah thought I did but turns out I didnt ;/ 2017-09-08 22:15:12 ohh got the name wrong in the makedepends 2017-09-08 23:01:48 Having trouble with thie error now... Dont understand it 2017-09-08 23:01:48 https://hastebin.com/owajeyokog.vbs 2017-09-08 23:02:12 `` ` CC dbus-proto.o 2017-09-08 23:02:13 make[2]: *** No rule to make target '../src/libresource.la', needed by 'libresource-glib.la'. Stop. 2017-09-08 23:02:14 make[2]: *** Waiting for unfinished jobs....``` 2017-09-08 23:03:40 oh dbus 2017-09-08 23:04:14 something to do with dbus? cant seem to find an issue in the code 2017-09-08 23:04:40 although https://git.merproject.org/mer-core/libresource/blob/master/src/resource-glib-glue.c#L27 looks particularly suspect 2017-09-08 23:04:50 what? 2017-09-08 23:04:52 theres no glib.h at all 2017-09-08 23:04:54 i just said "oh dbus" 2017-09-08 23:05:03 oh 2017-09-08 23:05:06 nvm then :p 2017-09-08 23:05:13 i need to look around a bit to say anything about it 2017-09-08 23:07:13 Bloo[m]: do you have glib-dev? 2017-09-08 23:07:40 dont think so lemme check 2017-09-08 23:07:47 apk add it 2017-09-08 23:08:02 just tried thta and got this response: `│OK: 549 MiB in 222 packages` so I gather its installed 2017-09-08 23:11:29 Bloo[m]: what's your apkbuild, so i can test? 2017-09-08 23:11:36 willing to share? 2017-09-08 23:11:56 danieli: absolutely! Sorry its a bit of a mess, Il send it over in a sec 2017-09-08 23:12:02 no worries 2017-09-08 23:14:33 danieli: https://hastebin.com/nobipabaye.bash here ya are! 2017-09-08 23:14:35 Thanks a lot 2017-09-08 23:15:24 Bloo[m]: do remember to add a maintainer and contributor field 2017-09-08 23:15:46 danieli: yeah definitely, Im jut slamming my APKBUILDs together as quickly as possible just to get my port working for now 2017-09-08 23:15:54 point taken 2017-09-08 23:15:58 but Il go over them all and credit the correct people when Im done :) 2017-09-08 23:16:04 you can fix them later on 2017-09-08 23:16:12 maintainer, as in, the maintainer of the alpine package 2017-09-08 23:27:51 ACTION sent a long message: pablog[m]_2017-09-08_23:27:51.txt  2017-09-08 23:30:18 Bloo[m]: for some reaso it seems to build fine when using only one thread 2017-09-08 23:30:27 (without -j) 2017-09-08 23:30:55 it also needed check-dev 2017-09-08 23:32:02 https://p.duniel.no/NBFg 2017-09-08 23:33:05 Woah, that's odd. Thanks a. Lot man ! 2017-09-08 23:33:19 Bloo[m]: does it work? 2017-09-08 23:33:48 danieli: going for a shower , but will test it when I get back and let you know 2017-09-08 23:33:57 all right, have a good one 2017-09-08 23:56:47 This branch is 351 commits ahead, 145 commits behind alpinelinux:master. 2017-09-08 23:57:22 some day I will stop fixing so many packages... lol 2017-09-08 23:57:41 i'm gonna fix licenses soon enough 2017-09-08 23:57:49 ALL the conflicts 2017-09-08 23:57:56 that might just be a bitch to merge 2017-09-08 23:58:01 hahaha 2017-09-08 23:58:04 what kinda stuff did you fix? 2017-09-08 23:58:30 danieli: modernising, fixing configure lines, and adding test suites to every X11 package 2017-09-08 23:58:40 danieli: it appears I have done about 200 2017-09-08 23:58:41 that's nice! 2017-09-08 23:58:56 last weekend I did 30 per day both days.. 2017-09-08 23:59:02 that's not bad at all 2017-09-08 23:59:15 also dragging all the packages kicking and screaming from py2 to py3 2017-09-08 23:59:17 where possible 2017-09-08 23:59:23 not always possible, but I've done quite a few 2017-09-08 23:59:30 yeah, modernizing is nice 2017-09-08 23:59:42 did you use `.patch`es, or? 2017-09-08 23:59:50 mostly it was just version bumps 2017-09-08 23:59:57 and then change the depends line from python2 to python3 2017-09-09 00:00:30 there were only two packages that needed .patch to work: itstool, which just merged it 20 days ago after working on it two years; and libxml2, which the main software supports py3 but the test suite did not. 2017-09-09 00:01:37 ncopa gonna kill me when I make PR and have 351 commits hahaha... need to probably squash all the *proto commits, as they are all the same thing (s/all/noarch/, +util-macros to makedepends, options='!check') 2017-09-09 00:01:42 that probably is 60 or 70 of them 2017-09-09 00:02:09 i bet there's gonna be some squashing, but you've been efficient 2017-09-09 00:02:33 it looks like xorg-server is pulling in a few more protos I missed, hmm 2017-09-09 00:02:44 better test build them all first 2017-09-09 00:02:48 i guess you did 2017-09-09 00:03:06 we almost have X in Adélie, which means alpine almost has all X packages with test suites and modern style 2017-09-09 00:03:15 I changed our local abuild to die on default_test instead of just warning 2017-09-09 00:03:33 pablog[m]: do you have virglrenderer-dev as a build dep? 2017-09-09 00:03:33 er default_check 2017-09-09 00:04:17 foxkit: ncie work :) 2017-09-09 00:04:41 ^ 2017-09-09 00:04:47 thanks ^^ 2017-09-09 00:05:06 i was angry at golang while implementing some algorithms for a mirror redirector 2017-09-09 00:05:14 turns out i had the arguments for math.Pow in the wrong order 2017-09-09 00:14:42 Shiz: Yes! I have virglrenderer-dev. I've uploaded my code to https://github.com/PabloCastellano/pmbootstrap/blob/feature/qemu-virtio-gpu-driver/aports/main/qemu/APKBUILD 2017-09-09 00:22:58 pablog[m]: hmm, it should find -lvirglrendere just fine, as virglrenderer-dev contains /usr/lib/libvirglrenderer.so 2017-09-09 00:31:19 Shiz: That is the weird thing ;( I will try wiping and compiling again everything from the beginning 2017-09-09 00:31:51 can you verify that if you install virglrenderer-dev yourself, /usr/lib/libvirglrenderer.so is a non-broken symlink? 2017-09-09 00:56:04 danieli: Am I missing something? Dont think im using -j in my APKBUILD :c 2017-09-09 00:56:40 Bloo[m]: try building it manually and see if it reproduces 2017-09-09 00:56:47 cd src/thepack[tab] 2017-09-09 00:56:58 there was an über-long dir name 2017-09-09 00:57:08 might be something in the flags handed to make or configure by abuild doing it 2017-09-09 01:03:57 Bloo[m]: check /etc/abuild.conf and $HOME/.abuild/abuild.conf for JOBS and MAKEFLAGS. if your build cannot handle multiple jobs, you need to explicitly put "make -j1" in your APKBUILD. some builders have insane number of job configured like ppc64el (has 160 cores, so..) 2017-09-09 01:10:37 s390x ayy 2017-09-09 01:17:32 CONFLICT (content): Merge conflict in main/wpa_supplicant/APKBUILD main/util-linux/APKBUILD main/strace/APKBUILD main/postgresql/APKBUILD main/pixman/APKBUILD main/libpng/APKBUILD main/hunspell/APKBUILD 2017-09-09 01:17:36 that's not so bad really 2017-09-09 01:17:42 considering there's basically 200 commits on each side 2017-09-09 01:18:09 that's not bad at all 2017-09-09 01:18:12 should be fairly easy to merge that 2017-09-09 01:52:42 Shiz: I have checked /usr/lib/libvirglrenderer.so and everything looks correct 2017-09-09 01:53:28 If you have any other idea, please tell me! :D 2017-09-09 02:12:23 is there a good way to defer running $pkgname.post-install? 2017-09-09 02:12:28 i.e. run it in the background 2017-09-09 02:12:38 nohup? 2017-09-09 02:14:16 errr 2017-09-09 02:14:27 that would be a massive policy violation :D 2017-09-09 02:14:30 agreed 2017-09-09 02:14:45 but downloading and gunzipping the files takes a little bit 2017-09-09 02:15:14 i made a geoip-extras, and i want it to actually make sure the database files are downloaded and installed after the install 2017-09-09 02:15:16 it's just a downloader 2017-09-09 02:15:34 danieli: wouldn't it be cleaner to package the database files instead? 2017-09-09 02:15:45 ollieparanoid[m]: i'd have to update the package every month 2017-09-09 02:15:51 and they're available from a static URL 2017-09-09 02:16:16 the `geoip` package is what debian calls "geoip-bin" + a cron script for downloading the country database 2017-09-09 02:16:28 it should be split up even further imo 2017-09-09 02:17:17 then I would personally just not download it automatically and let the user do it. but better ask for more opinions :) 2017-09-09 02:17:40 rnalrd maintains `geoip` 2017-09-09 02:17:58 i was merely extending on his idea 2017-09-09 02:18:37 but yeah.. i agree, i need opinions 2017-09-09 02:19:39 1. split `geoip` into `geoip-bin` and `geoip-database`? 2017-09-09 02:19:57 2. use a cron script for downloading geoip databases, or package the files themselves? it's updated once a month, and available from a static URL 2017-09-09 02:20:11 should honestly package the files themselves, not everyone who installs an APK is online 2017-09-09 02:35:10 it's too big to send as an email, the patch 2017-09-09 02:35:11 hmm 2017-09-09 02:53:55 there could be a new section configured 2017-09-09 02:54:02 volatile or such 2017-09-09 02:54:15 elaborate? 2017-09-09 02:55:05 debian has a section for packages which need to be updated frequently 2017-09-09 02:55:12 yeah, true 2017-09-09 02:55:27 volatile is a good idea 2017-09-09 02:55:41 especially in a rolling release distro there's gonna be a few of those 2017-09-09 02:55:59 alpine is not rolling release 2017-09-09 02:56:01 :p 2017-09-09 02:56:02 edge 2017-09-09 02:56:29 i was aiming at edge :< 2017-09-09 02:56:39 poor phrasing 2017-09-09 02:56:51 volatile would be for things like virus definition files and such 2017-09-09 02:56:55 should packages that gets updated on an interval go there? or just packages that overall gets updated frequently? 2017-09-09 02:56:57 i see, interval 2017-09-09 02:57:18 yeah, agreed. i'd assume accumulated data, mostly 2017-09-09 02:57:44 things which are low impact but not security/reliability updates 2017-09-09 02:58:11 gotcha 2017-09-09 02:58:19 would be a good idea 2017-09-09 02:58:41 debian has a separate security repo, but i think alpine is a little small for that yet? 2017-09-09 02:59:01 the security repo is a bug in debian 2017-09-09 02:59:32 alpine main repos receive security updates directly 2017-09-09 02:59:38 in a sense, yes 2017-09-09 02:59:44 it shouldn't need to be separate 2017-09-09 02:59:58 it's separate because of bureaucracy 2017-09-09 03:00:06 screw that shit 2017-09-09 03:00:12 right 2017-09-09 03:08:49 there is no reason to separate security repo, agreed 2017-09-09 03:23:58 it's been ages since i've done web design. this is nice. 2017-09-09 03:24:18 https://static01.nyt.com/images/2016/08/05/us/05onfire1_xp/05onfire1_xp-facebookJumbo.jpg 2017-09-09 04:48:20 ^ all web design tbh 2017-09-09 06:00:11 so.. how can i send a large aports commit with binary? 2017-09-09 06:00:15 email won't work 2017-09-09 06:01:57 errr 2017-09-09 06:01:58 what 2017-09-09 06:02:53 what? 2017-09-09 06:06:39 danieli: what binary are you putting in aports, and why? 2017-09-09 06:07:07 geoip 2017-09-09 06:07:20 instead of a cron script, i'm packaging the .dat files 2017-09-09 06:07:27 why not just download the .dat files 2017-09-09 06:07:28 danieli: that should be downloaded from the geoip server itself. 2017-09-09 06:07:31 as part of the APKBUILD :p 2017-09-09 06:07:46 the source line can be computed, if needed 2017-09-09 06:07:51 `geoip` uses a cron job 2017-09-09 06:08:16 i.e. you can define year=2017 month=september and do http://geoip/${year}${month}.dat or such for source= 2017-09-09 06:08:20 (I don't know the format) 2017-09-09 06:08:24 is it okay if it takes a while to perform the install? 2017-09-09 06:08:42 yes. the invisible island server, where ncurses lives, downloads at 20k/s :) 2017-09-09 06:08:48 if you install it together with other packages, the $pkgname.post-install is gonna take a little while to complete 2017-09-09 06:08:49 jesus christ 2017-09-09 06:08:54 that's bad 2017-09-09 06:09:06 what is it doing in post-install? 2017-09-09 06:10:04 triggering /etc/periodic/monthly/geoip-extra 2017-09-09 06:10:23 a couple `wget`s, `gunzip`s and `mv`s 2017-09-09 06:10:49 what we are saying 2017-09-09 06:10:51 this seems like a bad idea for a few reasons. what if the server is low on disk space? the transaction could abort in the middle due to enospc 2017-09-09 06:10:56 is that the APKBUILD should have it in $sources 2017-09-09 06:11:12 ah gotcha 2017-09-09 06:11:45 download URL never changes btw 2017-09-09 06:11:57 foxkit: it could indeed 2017-09-09 07:22:41 w 31 2017-09-09 07:30:45 kaniini: are you /sure/ the failures in gc are due to grsecurity? because they are failing here too, on vanilla. and on gentoo-based Adélie we shipped fix, because the segv was caused by improper stack size assumption: https://code.foxkit.us/adelie/patches/blob/master/dev-libs/boehm-gc/boehm-gc-7.4.2-testsuite.patch 2017-09-09 07:31:12 I won't fix it if it still doesn't work on grsecurity because obviously that would not be upstreamable 2017-09-09 07:31:47 it also required two /other/ patches: https://code.foxkit.us/adelie/patches/tree/master/dev-libs/boehm-gc 2017-09-09 07:32:02 otherwise, applications like w3m which use it crash bad on 32-bit arches like x86 and arm 2017-09-09 07:32:18 (doesn't seem to 'poison' 64-bit since 64-bit gc does not seem to tag memory the same) 2017-09-09 07:33:19 this package was actually the /reason/ Adélie started caring so much about test suites; if we had enabled it at the beginning, I would not have wasted two weeks pouring over w3m code in a red herring chase 2017-09-09 16:34:17 it's quiet today 2017-09-09 16:35:02 <_ikke_> yea 2017-09-09 16:35:07 <_ikke_> Mostly in the weekend though 2017-09-09 16:35:22 I've noticed 2017-09-09 17:32:51 Im trying to mnow compile qtsystems 2017-09-09 17:32:54 https://hastebin.com/xofedugaba.py 2017-09-09 17:33:27 ACTION sent a long message: Bloo[m]_2017-09-09_17:33:26.txt  2017-09-09 17:35:39 Take this for example: https://github.com/qt/qtsystems/blob/ffcf1ff783827171c3fb911cc90b2d0ecb2931af/src/publishsubscribe/qvaluespace.h#L37 its clearly listing the wrong directory 2017-09-09 17:35:54 my initial instinct is to patch it but I dont want to incase theres something I can change in the build settings to fix it 2017-09-09 17:37:42 ACTION sent a long message: Bloo[m]_2017-09-09_17:37:41.txt  2017-09-10 00:34:52 In the end, I only ended up needing systeminfo, so I compiled that on its own with a couple patches and it worked :D 2017-09-10 00:35:13 I think all `lipstick` needs now is systemd patching out and Im good 2017-09-10 00:35:16 (hopefully) 2017-09-10 21:32:37 https://github.com/alpinelinux/abuild/pull/26 :) 2017-09-10 22:32:26 foxkit: <3! 2017-09-10 23:31:07 the main/apr is so far out of date that the tarball is a 404 because it was moved to apache archive dir 2017-09-10 23:31:14 trying to bump, but need to find a way to disable test of dlclose first 2017-09-10 23:44:04 jesus, that's bad 2017-09-11 00:09:58 _ikke_ or kaniini: mind fixing the commit msg and applying the erlang patch? I forgot the "community/erlang: " part, it's labelled "bump to OTP 20.0" 2017-09-11 00:10:09 wait, there's even newer versions of it, dammit 2017-09-11 07:47:23 could someone force a rebuild on the fish package? 2017-09-11 08:09:51 could someone trigger a rebuild of the fish package? there is some kind of bug.... and compiling it by myself and others tested that too, worked fine 2017-09-11 08:10:01 https://github.com/fish-shell/fish-shell/issues/4387 2017-09-11 08:19:42 xsteadfastx: it smells like a real bug in fish 2017-09-11 08:20:16 the "illegal instruction" is probably fortify-headers 2017-09-11 08:29:37 there are three people just compiling the apk and the failure is gone 2017-09-11 08:29:56 i check the fortify-headers 2017-09-11 08:36:17 but i cant reproduce it with rebuilt of the same APKBUILD... but on all boxes i use the precompiled package it crashes 2017-09-11 08:36:58 xsteadfastx: is there a special make completion package somewhere? can you link it to me? 2017-09-11 08:39:42 i use the built in one 2017-09-11 08:39:54 the bugreport i use has all steps to reproduce it 2017-09-11 08:45:48 xsteadfastx: fixed(-ish), ncopa can we have a rebuild on fish? this is that fortify-headers bug again 2017-09-11 08:45:58 the one that was fixed in 0.9 2017-09-11 08:50:48 so what did you do? 2017-09-11 08:51:12 added fortify-headers as dependency? 2017-09-11 08:52:45 xsteadfastx: no. there was a bug in fortify headers. fish needs to be rebuilt against new version, then it is fixed (and that is why it works for you when rebuilt) 2017-09-11 08:52:56 xsteadfastx: so you were right first time, it needs a rebuild 2017-09-11 08:53:17 okidoki... yeah :) 2017-09-11 12:37:53 ncopa: CFP for FOSDEM 2018 is open 2017-09-11 12:38:20 how did you submit your talk last year? were you part of the devroom organizers or did you contact them later? 2017-09-11 12:39:33 because if you want to make a devroom for "simple systems" or something, I have about 4-5 hours of content :D 2017-09-11 13:38:41 i was not part of devroom organizers 2017-09-11 13:38:45 i think i just submitted a talk 2017-09-11 13:39:26 so the devroom was created, but not complete? 2017-09-11 13:39:44 you were lucky to find some room it seems :) 2017-09-11 17:43:19 fyi: i'll be going on holiday from wednesday on for a little bit over a week 2017-09-11 17:43:25 will be unavailable 2017-09-11 18:00:29 have a nice vacation :) 2017-09-11 18:05:28 thanks :) 2017-09-11 18:05:49 yeah, have a good one 2017-09-11 18:46:08 Shiz: anything new on system-llvm? 2017-09-11 21:31:07 ncopa: what was that we discussed with coreos/grub? 2017-09-11 21:31:13 i lost my notes on that 2017-09-11 21:44:12 I built it with some ever so slight modifications 2017-09-12 02:49:26 Hey Shiz :D 2017-09-12 02:50:10 I wanted to check back in with you now that llvm 5.0 has been released 2017-09-12 02:51:00 Has there been any work towards an llvm5 APKBUILD or should I start it? 2017-09-12 03:03:02 Going offline, will pop back tomorrow :) 2017-09-12 03:07:10 not that i know of 2017-09-12 03:07:15 xentec: little sadly 2017-09-12 03:13:56 i can have a crack at a llv5 apkbulid 2017-09-12 03:13:59 build* 2017-09-12 04:09:01 thought Shiz is on vacation :D 2017-09-12 04:43:39 tmh1999: not until Wednesday 2017-09-12 04:46:10 you're keeping track :) foxkit 2017-09-12 04:46:14 also, dare I say good morning? 2017-09-12 04:46:15 some years ago, I had requested afterstep package, and was told that if I wanted to package and maintain it that it could be added 2017-09-12 04:46:27 I said no because I was mainly using BSD at that point (2009) 2017-09-12 04:46:32 is that offer still valid? ^^ 2017-09-12 04:46:54 danieli: 23:46 here. been awake since around noon. hoping to get some hacking done before bed 2017-09-12 04:47:03 martell, Shiz: I am working on llvm5 right now. everything seems fine except one failing test [DynamicLibraryTest.cpp]: it expects destructors of static objects to be called after dlclose which musl denies 2017-09-12 04:47:03 foxkit: ah all right, I see 2017-09-12 04:47:08 I'm more productive at night 2017-09-12 04:47:18 it's 06.57 here 2017-09-12 04:47:20 47* 2017-09-12 04:47:21 yup, as am I 2017-09-12 04:47:31 oh early morning for you, well good morning to you! 2017-09-12 04:47:36 thank you <3 2017-09-12 04:47:46 xentec: hmm.. 2017-09-12 04:48:00 xentec: I see same test failure in apr-util. I just disabled that test. can LLVM be told to disable the test? 2017-09-12 04:49:03 note that as specified here: http://pubs.opengroup.org/onlinepubs/9699919799/functions/dlclose.html 2017-09-12 04:49:06 dlclose does not have to do anything 2017-09-12 04:49:16 so that test is technically non-posix-conformant 2017-09-12 04:49:27 yes 2017-09-12 05:00:54 is it all right if i go through flagged packages and update them, unless a major update is expected soon? 2017-09-12 05:00:56 or how is that done? 2017-09-12 05:18:29 hi, 2017-09-12 05:18:29 if anyone planning for new CI that fits alpinelinux philosophy, I could suggest 2017-09-12 05:18:29 https://github.com/insteps/mqtt-dirpub, this is modified version of clients. 2017-09-12 05:19:05 just curious, do you have ulterior motives of suggesting that? 2017-09-12 05:19:14 it can create /proc or /sys like hierarchy for mqtt msgs. 2017-09-12 05:19:15 eg. mosquitto_sub -h msg.alpinelinux.org -t '#' --fmask /tmp/msgs/@date/@topic --nodesuffix 'msg' 2017-09-12 05:19:15 the /tmp/msgs folder can be processed at anytime later 2017-09-12 05:19:39 danieli, no, but we already have mqtt service 2017-09-12 05:19:50 oh i see the profile 2017-09-12 05:19:56 next time, remember full disclosure - you wrote it 2017-09-12 05:20:03 though the msgs are not nomalized yet 2017-09-12 08:20:12 skarnet, is there a demo/poc iso that uses full s6 stack that I can try ? 2017-09-12 08:20:32 would be nice if based on alpinelinux 2017-09-12 08:21:35 not Alpine-based, but you can make your own system with https://github.com/skarnet/lh-bootstrap 2017-09-12 08:22:12 there was an online s6-based jor1k VM, but it got unplugged, unfortunately 2017-09-12 09:09:54 ncopa, you around? 2017-09-12 09:10:16 re: krb5 package. Are we usingthe "Currente release" of the "Maintenance Release"? 2017-09-12 10:50:06 xentec: the test is invalid and should be disabled, yes 2017-09-12 10:50:15 danieli: yes that is fine 2017-09-12 18:39:03 someone knows what is wrong with gdb 8.0 in Alpine? I saw that gdb was upgraded to 8.0 but the patch was reverted. I just built gdb 8.0.1 in alpine ppc64le and seems it is working (at least a simple debug is) 2017-09-13 01:21:46 what *is* the best way to merge in all our changes? we're past 400 packages. it's about time for me to do the monthly merge thing, but at this point, I think it would crash travis... and probably crash ncopa's brain too. perhaps batches of related commits, in smaller PRs? it should definitely slow down now, since we have built 1700 of our target 2400 packages for alpha3 and most of the rest are 2017-09-13 01:21:48 things we are packaging ourselves. 2017-09-13 01:23:13 so doing PRs in small increments should not cause a runaway effect.. 2017-09-13 01:31:23 smaller batches would be better, to be honest 2017-09-13 01:31:35 would lessen the load on travis 2017-09-13 01:32:26 it makes me sad that I cannot test on the arm platform, I feel it is causing undue burden on alpine. 2017-09-13 01:32:40 how come you can't test on ARM? 2017-09-13 01:32:42 but I can't manage to get any of these silly things to boot 2017-09-13 01:32:57 are you virtualizing or using physical hardware? 2017-09-13 01:33:42 I don't have any experience with ARM in the slightest. I have decades with x86, sparc, ppc, etc. so right now even the rudimentary task of "show me uboot" is shrouded in mystery and not happening... I feel like I did in 1998 with x86. 2017-09-13 01:33:58 danieli: I have a pine a64 and an rpi b+ 2017-09-13 01:34:21 ah.. yeah, point taken. I've mostly done work with SPARC, x86, x64, some IA-64, mips and arm 2017-09-13 01:34:40 where are you failing to boot alpine? 2017-09-13 01:34:56 have yet to touch an itanium machine. MIPS, I've only done DEC and SGI. 2017-09-13 01:35:19 I've only messed around a slight bit with itanium 2017-09-13 01:35:34 I'm not old enough for most of the arches and platforms you're mentioning :) 2017-09-13 01:35:37 danieli: the rpi just shows grey screen forever, it doesn't seem to like the SD card for some reason, though I can read it back fine here. as for the pine... it just has some LEDs flashing... trying to get my arduino to latch on to the serial port.. 2017-09-13 01:36:07 hmm.. 2017-09-13 01:36:52 while I have so much experience with firmware, I am not a 'hardware person'. the closest I've ever come to hardware-level stuff is soldering a new battery in to the 386 2017-09-13 01:37:12 well, and the typical assembly of computer systems from their parts. but that's not fixing, that's just clicking things together 2017-09-13 01:38:06 I'm better with hardware than firmware, but proficient with firmware 2017-09-13 01:38:16 I'll rig up another vlan and do some testing with a couple of my rpis 2017-09-13 01:38:27 foxkit: qemu-system-arm works rather nice in my experience 2017-09-13 01:38:29 so wiring up a custom harness for JTAG using an arduino and spare bits of copper... well it is a whole new world! 2017-09-13 01:38:37 with qemu-system-aarch64 you can boot tianokit efi directly 2017-09-13 01:38:44 and boot an aarch64 linux kernel :) 2017-09-13 01:38:46 yeah, qemu is neat 2017-09-13 01:38:54 tianocore* 2017-09-13 01:39:30 Shiz: qemu is useful for low systems development (and I've used it for kernel work before) but I would never use it for testing/QA, which is my exact need for ARM. I know it has many deficiencies in emulation that could cause tests to pass/fail in qemu and do the opposite irl. 2017-09-13 01:39:42 fair enough 2017-09-13 01:39:47 on that note, i'll be leaving you 2017-09-13 01:39:54 got a flight to catch tomorrow :) 2017-09-13 01:40:06 don't burn down the distro while i'm gone please 2017-09-13 01:40:15 for instance, the CVE found in linux kernel 4.3 where you could access debug registers of x86 from userland... that was caused by qemu not emulating the crX registers properly 2017-09-13 01:40:21 and I discovered it on my real quark :P 2017-09-13 01:40:33 Shiz: have a lot of fun! we'll try to be good 2017-09-13 01:40:48 danieli: you got the matches? 2017-09-13 01:40:50 XD 2017-09-13 01:41:47 I got the matches, explosives and alcohol 2017-09-13 01:42:21 great! 2017-09-13 01:44:21 hmm 2017-09-13 01:44:24 I cannot find some of my SD cards 2017-09-13 01:45:34 I found one, it'll do 2017-09-13 02:56:31 god damn it, vlc made my DE crash 2017-09-13 04:01:15 so why is llvm support in clamav disabled? 2017-09-13 04:18:31 ooo 2017-09-13 04:18:46 in clamav daemon subpkg, it is installing files to $pkgdir instead of $subpkgdir, that's a no-no 2017-09-13 04:18:54 ACTION adds another commit to long list :/ 2017-09-13 04:20:32 hm 2017-09-13 05:34:12 that was intense. my network failed to come back up after some changes 2017-09-13 05:47:07 danieli: it's more intense when it is on a remote server XD 2017-09-13 05:47:27 I had that happen once... changed the network settings, rebooted, waited 10 minutes... welp.. time to drive to thedc 2017-09-13 05:47:29 the dc* 2017-09-13 05:48:20 I have in the past made the same mistake 2017-09-13 05:48:25 mostly with networks, but servers too 2017-09-13 05:48:33 generally I don't make mistakes, especially expensive ones 2017-09-13 05:52:36 my PC was directly connected to an ISP, it gets unlimited(?) public addresses on that port 2017-09-13 05:52:48 I connected a mikrotik RB between, so I can hook up RPis and more 2017-09-13 06:21:10 danieli: there is a limit! the limit is your ISP's subnet ;) 2017-09-13 06:21:14 hah true 2017-09-13 06:21:20 should try dhcp starvation just to be an ass 2017-09-13 06:21:24 this isp's core network is crap 2017-09-13 06:21:29 that would be a pretty funny DoS... just connect a /24 or /23 worth of rpis to it 2017-09-13 06:21:31 lol 2017-09-13 06:21:36 meh 2017-09-13 06:21:46 or just run a program making dhcp requests continuously 2017-09-13 06:22:57 dhcpc test suite lol 2017-09-13 06:23:18 lol 2017-09-13 06:23:27 I booted alpine on an rpi2 fine by the way 2017-09-13 06:23:38 I can create you some qemu VMs if you want to build stuff 2017-09-13 06:29:09 @ foxkit 2017-09-13 06:31:39 danieli: thanks! I will probably like to do that this weekend, so no ush. 2017-09-13 06:31:41 rush* 2017-09-13 06:31:52 I'll have it up and going by then in that case 2017-09-13 06:32:05 I'll just set up a build env on edge 2017-09-13 06:35:34 @ncopa did you see what I wrote about checkbashisms being broken in master due to new perl? 2017-09-13 06:37:56 no "@" for tagging on IRC :) 2017-09-13 06:38:06 oh, nice, that is the regex brace thing 2017-09-13 06:38:16 I had to fix same syntax error in autoconf 2017-09-13 06:38:31 and I think f colista fixed it in automake, but I'm not sure 2017-09-13 06:38:31 danieli: thx ;-) 2017-09-13 06:39:50 foxkit: is my fix OK you think? 2017-09-13 06:40:03 did you say there's a PR? 2017-09-13 06:40:06 HRio: I don't see the fix. can you link me? is it on the ML? 2017-09-13 06:40:06 mind giving me a link? 2017-09-13 06:40:21 https://github.com/alpinelinux/aports/pull/2284 2017-09-13 06:40:31 i see it on the forum 2017-09-13 06:42:15 virt-what failed to build 2017-09-13 06:42:19 it seems ok. usually it is replacing { with \{, not [{] 2017-09-13 06:42:28 but I assume it is equivalent, so that should be fine 2017-09-13 06:42:48 https://git.savannah.gnu.org/cgit/automake.git/commit/?h=micro&id=13f00eb4493c217269b76614759e452d8302955e as ref 2017-09-13 06:44:14 ah nice ok 2017-09-13 06:44:19 then yes your fix is great. 2017-09-13 06:45:20 I see you removed some SPARC stuff, how come? 2017-09-13 06:45:28 obsolete code? 2017-09-13 06:45:29 the basisms 2017-09-13 06:45:41 did not spot it as bashisms 2017-09-13 06:45:58 all right 2017-09-13 06:46:03 the pr looks fine to me 2017-09-13 06:46:06 thats why I now added checkbashisms to check() 2017-09-13 06:51:12 foxkit: danieli: thanks for reviewing it. 2017-09-13 06:51:23 no push access though 2017-09-13 06:51:32 i just read over the changed files 2017-09-13 06:51:44 a build seems to fail though 2017-09-13 06:51:50 I'm not sure about the virt-what part since I'm not familiar with that package, sorry 2017-09-13 06:51:55 ^ 2017-09-13 06:52:04 danieli: yes, its still using the unfixed checkbashims it seems 2017-09-13 06:52:10 ah okay 2017-09-13 06:52:14 did you check locally in your build env? 2017-09-13 06:52:33 yes, after installing the fixed checkbashisms first 2017-09-13 06:52:42 nice, then it should be good to go 2017-09-13 06:54:49 >>> checkbashisms*: Create checkbashisms-2.0.0.2-r4.apk.... and then (7/8) Installing checkbashisms (2.0.0.2-r3) when building virt-what. 2017-09-13 06:55:06 from the travis log 2017-09-13 06:55:58 foxkit: can you use specific versions of a package in $depends? 2017-09-13 06:56:22 yes you can, in fact I just did it a minute ago in main/feh: 2017-09-13 06:56:41 +checkdepends="perl-utils>=5.26 perl-test-command" 2017-09-13 06:56:46 aah okay, thanks 2017-09-13 06:57:02 gotta make sure to note that down when starting the new docs 2017-09-13 06:57:21 yeah it takes 'atom' for dependencies 2017-09-13 06:57:44 https://wiki.adelielinux.org/wiki/APK_internals#.PKGINFO 2017-09-13 06:57:57 i actually have a tab open with that page 2017-09-13 06:57:59 so package[symbol]version 2017-09-13 06:58:10 where symbol is said on that page 2017-09-13 06:58:16 oh hah nice :) 2017-09-13 07:00:29 foxkit: does it work in travis, to use packages updated earlier i the same PR if I use versioned depends? 2017-09-13 07:01:01 HRio: unfortunately no. each package is built in a separate sandbox and packages are pulled in from the main repos. 2017-09-13 07:01:11 ^ 2017-09-13 07:07:19 foxkit: regarding the rest of virt-what, that package is contributed by me and only me that seems to do changes to it. I use it in apt-dater-host if virt-what is installed. 2017-09-13 07:07:38 ahh okay. 2017-09-13 07:28:12 2b879b3c5dcf:~$ nano APKBUILD 2017-09-13 07:28:12 Too many errors from stdin 2017-09-13 07:28:13 what the 2017-09-13 08:32:31 er, there is typo in main/tmux.. 2017-09-13 08:32:51 my PR is getting so damn large 2017-09-13 08:33:04 really need to coordinate the best way to handle it 2017-09-13 08:57:15 just squash some commits together 2017-09-13 09:26:54 Hello, everyone. 2017-09-13 09:27:17 I'm trying to use vala on pmos, but installed valac does not work: 2017-09-13 09:27:31 /usr/lib/glib-2.0/include/glibconfig.h:11:20: fatal error: limits.h: No such file or directory 2017-09-13 09:27:37 Any ideas how to fix that? 2017-09-13 09:32:27 pavelm3, what are you trying to do? 2017-09-13 09:32:44 fcolista: Compile hello world. Next step is compile some debug client on pmos. 2017-09-13 09:33:15 limits.h is included in the linux headers 2017-09-13 09:33:29 fcolista: Which package would that be? 2017-09-13 09:33:36 http://pkgs.alpinelinux.org/contents?file=limits.h&path=&name=&branch=&repo=&arch= 2017-09-13 09:34:07 fcolista: So I guess "musl-dev" is the right package? 2017-09-13 09:35:04 look what /usr/lib/glib-2.0/include/glibconfig.h is trying to find 2017-09-13 09:44:40 fcolista: Thanks... I added musl-dev and it now works. 2017-09-13 09:48:19 well done 2017-09-13 11:02:42 foxkit: i suppose the best thing to do would give you git push access :) 2017-09-13 11:14:20 then he will simply rebrand Alpine as Adelie. don't do that! 2017-09-13 11:14:29 j/k 2017-09-13 12:03:22 fcolista: thanks and bye. 2017-09-13 13:04:49 good morning my friends 2017-09-13 13:52:08 leo-unglaub: morning 2017-09-13 13:52:26 tmh1999: it looks like s390x builder is getting stuck on the tests 2017-09-13 13:52:32 test in pixman 2017-09-13 13:52:42 1670 1669 buildoze R 4912 0% 3 24% /home/buildozer/aports/main/pixman/src/pixman-0.34.0/test/.libs/lt-affine-test 2017-09-13 13:52:42 1520 1519 buildoze R 20120 0% 0 24% /home/buildozer/aports/main/pixman/src/pixman-0.34.0/test/.libs/lt-region-contains-test 2017-09-13 13:52:52 roger that 2017-09-13 13:53:09 the builder you are running is the one that I gave you ? 2017-09-13 13:53:10 If I could have someone look at the following GitHub PRs, it would be much appreciated. testing/nitro (2285), testing/agg (2279), and testing/libspatialite (2217) 2017-09-13 13:53:11 looks like there are some random numbers involved 2017-09-13 13:53:23 tmh1999: no on the builder we have for the official builds 2017-09-13 13:54:04 tmh1999: when i try reproduce in my dev env it gets stuck here: 2017-09-13 13:54:11 6579 6578 ncopa R 12316 0% 2 26% /home/ncopa/aports/main/pixman/src/pixman-0.34.0/test/.libs/lt-region-contains-test 2017-09-13 13:54:11 6762 6761 ncopa R 8732 0% 1 24% /home/ncopa/aports/main/pixman/src/pixman-0.34.0/test/.libs/lt-scaling-test 2017-09-13 13:54:14 ncopa : any chance you could allocate some time to move the builder to the vm I gave you? 2017-09-13 13:55:20 hum 2017-09-13 13:55:50 Filesystem Size Used Available Use% Mounted on 2017-09-13 13:55:50 /dev/dasdc1 119.6G 58.0G 55.4G 51% / 2017-09-13 13:55:50 I am checking in both vm and lxc env 2017-09-13 13:56:14 current builder uses 60G of 120 avail 2017-09-13 13:56:41 okay guess I will allocate more space 2017-09-13 13:58:09 the host machine I have only has 100GB so it might be sufficient atm 2017-09-13 13:58:49 taking down the vm for a moment 2017-09-13 13:58:53 ok 2017-09-13 13:59:48 i suppose we can decommision the old one, once we moved it over? 2017-09-13 14:00:45 I think it's all up to you. I don't have permission on your machines over there :D 2017-09-13 14:05:28 tmh1999: maybe we should discuss it in #alpine-infra 2017-09-13 15:27:47 does anybod know if libexecinfo actually works? 2017-09-13 15:29:07 i was able to build chromium with libexecinfo but it segfaults at initialization 2017-09-13 15:29:10 https://dpaste.de/gMSE 2017-09-13 15:48:53 i use libexecinfo in a few self built projects also libunwind, both seem to work for me 2017-09-13 15:49:31 (last time i used them was 1-2 months ago though) 2017-09-13 16:13:58 chromium does not seem to like it 2017-09-13 16:47:18 Hey Shiz :D 2017-09-13 17:14:31 alpine on z/OS directly, interesting 2017-09-13 17:16:21 giving foxkit push access wouldn't be an awful idea 2017-09-13 17:26:50 jo, quick question: Is s/o working on webkit2gtk 2.18 for alpine? - I'm currently doing the same for void. 2017-09-13 17:27:39 ncopa: would you bump community/emacs to 25.3 for me please? 2017-09-13 17:30:41 wWrAR07vpPZ1: have you submitted a patch? if not, i can 2017-09-13 17:35:27 no, i have not 2017-09-13 17:36:49 okay, i'll do it 2017-09-13 18:19:12 build testing 2017-09-13 18:20:27 thx 2017-09-13 18:24:55 it built fine, i'm sending the patch to aports 2017-09-13 18:29:16 there 2017-09-13 21:15:42 anyone mind merging the patch for emacs? 2017-09-13 22:32:17 ncopa, I am humbled by the suggestion. I certainly don't have an issue with that, however I would of course like to know what the policies are. if you like I can email my questions to you, or discuss them here or PM? 2017-09-13 22:33:02 probably discuss them here, so other core devs can answer. But since most of them are also EU-based, you won't get many answers at this time :) 2017-09-13 22:34:36 ah true. 2017-09-13 22:36:19 1) what is the process when there is a version bump or fix required for a package that has a Maintainer: that is not one's self? do you poke them first, send an email / message? is there a timeout (after x time, just do it anyway)? do they need to be Signed-off-by or Reviewed-by or Acked-by or similar? 2017-09-13 22:37:31 2) is there a 'staging' branch or similar where packages can built for all arches before being sent to master? I'm especially concerned about arm (which danieli is going to help with) and s390x, and I don't think there is even qemu for s390x that functions (at least, not last I checked). I do not want to submit packages to master that break on any arch. 2017-09-13 22:38:28 3) is there a QA process or review, where someone else is required to review a commit to master before it is pushed? i.e. it takes two core devs to commit? are there times where this is unnecessary (like security fix)? 2017-09-13 22:40:31 4) do we have clear guidelines about the movement of packages from testing to community? maximum time in testing looks to be 6 months; is there a minimum? if it passes build on all arches, and runs correctly on some selection of them, is that when it is moved? 2017-09-13 22:41:12 for instance, let us take simple package xpdf. let us say it isn't packaged, and I put it in testing. it builds on all builders and I read a book on x86 and armhf and ppc64le. is it ready for community at that time? 2017-09-13 22:42:39 5) I know that commit msg standard is "repo/pkg: what happened". but is there any other standard formatting to note? integration with redmine? does a commit "Fix #xxx" or "Fixes #xxx" or "Close #xxx" or "Closes #xxx"? 2017-09-13 22:43:05 if I think of any more, I'll write them here, but those are the questions I can think of for now... that's a lot already, heh 2017-09-13 23:01:28 I've gathered that AL's s390x builder runs on some VM or container somewhere 2017-09-13 23:02:07 foxkit: I was about to ask - what packages did your mega-PR update? 2017-09-13 23:02:18 I was gonna go through flagged packages 2017-09-13 23:02:40 some of them are probably already outdated by now 2017-09-13 23:26:34 apr-util, apr, gobject-introspection, glib, autoconf, ncurses 2017-09-13 23:26:57 danieli: ^ these packages were version bumped. everything else I left alone, focusing on modernisation of the build and adding test suite so future version bumps would be taken care of and tested properly 2017-09-13 23:27:09 ah I see, hmm 2017-09-13 23:27:19 I don't want to update flagged packages until your commits are merged in 2017-09-13 23:43:30 foxkit: ideas? 2017-09-14 00:16:56 danieli: https://code.foxkit.us/adelie/aports < you can check here if a package's tip commit is by me, if not it's clear to update. 2017-09-14 00:16:59 that is my main idea 2017-09-14 00:17:20 good idea, thanks foxkit 2017-09-14 00:17:52 when did the unmerged commits start? 2017-09-14 00:18:12 i see main, community and unmaintained have been changed recently 2017-09-14 00:28:21 https://git.merproject.org/mer-core/cor/tree/master does anyone know why CMakeLists.txt requires `VERSION` to be defined but its not dfined anywhere? 2017-09-14 01:12:39 foxkit: horizon seems interesting 2017-09-14 01:12:45 shame i associate the name with vmware 2017-09-14 03:02:09 danieli: I was saying just go in to the package and see if there is a different commit compared to the alpine git repo ;) should be pretty easy 2017-09-14 03:02:27 danieli: also, thanks! I didn't know VMware had a Horizon, but yes, we are hoping to make a better installer 2017-09-14 03:05:20 yeah, I'll do that - and the more you know :) 2017-09-14 03:05:24 a better installer would be really neat 2017-09-14 06:43:10 ncopa, krb 1.15.1 upgrade: 2017-09-14 06:43:10 https://dpaste.de/rgoJ/raw 2017-09-14 06:43:19 and the patch: 2017-09-14 06:43:19 https://dpaste.de/0mJV/raw 2017-09-14 06:45:02 do you need remove the --with-crypto-impl=openssl 2017-09-14 06:45:20 morning ncopa 2017-09-14 06:45:22 with this does not buidl 2017-09-14 06:45:33 void removed that part too 2017-09-14 06:45:40 ok 2017-09-14 06:45:53 and the libressl.patch is the one in thei repo 2017-09-14 06:46:10 do_configure() { 2017-09-14 06:46:10 WARN_CFLAGS= ./src/configure ${configure_args} --sbindir=/usr/bin \ 2017-09-14 06:46:10 --disable-rpath --with-system-et --without-system-verto --with-ldap \ 2017-09-14 06:46:10 --with-system-ss --with-system-db --enable-shared --without-tcl \ 2017-09-14 06:46:10 ac_cv_func_pthread_once=yes ac_cv_func_pthread_rwlock_init=yes \ 2017-09-14 06:46:11 acx_pthread_ok=yes ac_cv_func_regcomp=yes ac_cv_printf_positional=yes \ 2017-09-14 06:46:13 krb5_cv_attr_constructor_destructor=yes,yes 2017-09-14 06:46:22 their configure 2017-09-14 06:46:28 ok, lets go for that then 2017-09-14 06:46:42 need figure out what needs rebuilding 2017-09-14 06:46:42 of course is 1.15.1 2017-09-14 06:47:11 not many packages luckly 2017-09-14 06:47:17 i am working on upgrade chromium 2017-09-14 06:47:23 so you can skip chromium for now 2017-09-14 06:47:28 and chromium is a pain.... 2017-09-14 06:47:47 i can help out if you need anything, i'm familiar with large portions of the chromium code base 2017-09-14 06:47:52 and yes, it is a pain 2017-09-14 06:48:15 we should upgrade gnome 3.26 too 2017-09-14 06:48:24 danieli: oh nice 2017-09-14 06:48:30 ncopa, anyway I tested krb5 in a qemu box 2017-09-14 06:48:44 i'm gonna cross check with foxkit's commits and go through flagged packages 2017-09-14 06:48:56 i haven't get issue 2017-09-14 06:49:06 just fyi.. 2017-09-14 06:49:12 fcolista: very good 2017-09-14 06:49:34 danieli: i spent a day to make chromium 61 build 2017-09-14 06:49:45 it builds with libexecinfo library 2017-09-14 06:49:46 abi breaks for some libs 2017-09-14 06:49:47 https://abi-laboratory.pro/tracker/objects_report/krb5/1.14.2/1.15/report.html 2017-09-14 06:49:50 ncopa: jeez, poor thing 2017-09-14 06:49:56 but it segfault at startup 2017-09-14 06:50:03 hmm 2017-09-14 06:50:09 got the pkgbuild + patches, if any? 2017-09-14 06:50:26 i can dig around a little and see if i see something obvious 2017-09-14 06:50:44 our current, working, chromium 58 does not use libexecinfo, instead it patches out all the backtrace/libexecinfo stuff 2017-09-14 06:51:00 i think i need do the same 2017-09-14 06:51:09 i was able to build and run chromium 60 2017-09-14 06:51:32 however... every tab will "aw snap!" after a while 2017-09-14 06:51:36 so something is broken 2017-09-14 06:51:48 i suspect that it is the system libevent or similar 2017-09-14 06:52:10 it could also be some of the stats collectors 2017-09-14 06:53:43 this is the chromium 60 thing which crashes: http://tpaste.us/kq91 2017-09-14 06:55:00 this is the diff for chromium 61: http://tpaste.us/OzWd 2017-09-14 07:01:14 i think i will rebase the no-execinfo patch 2017-09-14 07:07:38 ncopa: so chrome 59 works, but not 60 or 61? 2017-09-14 07:13:12 chrome 58 works 2017-09-14 07:13:17 but not 60 or 61 2017-09-14 07:13:22 i have not tried to build 59 2017-09-14 07:13:26 I see 2017-09-14 07:13:35 60 and 61 needs a bunch of patches though 2017-09-14 07:13:42 60 brok the gn bootstrap 2017-09-14 07:13:58 61 fixed that breakage but introduced more breakages 2017-09-14 07:14:10 and introduced breakage with gcc 2017-09-14 07:14:24 apparently chrome CI only use clang nowdays 2017-09-14 07:14:31 so gcc isnt really supported anymore 2017-09-14 07:39:24 ncopa: by the way, does alpine have google api keys? 2017-09-14 07:39:28 for chromium 2017-09-14 07:43:41 danieli: https://github.com/alpinelinux/aports/blob/8b7e48dcaf6a2049edeffaa957db618e923b78ab/community/chromium/APKBUILD#L117 2017-09-14 07:43:41 https://github.com/alpinelinux/aports/blob/8b7e48dcaf6a2049edeffaa957db618e923b78ab/community/chromium/google-api.keys 2017-09-14 07:44:15 ah nice 2017-09-14 07:44:42 hiding in plain sight ;-) 2017-09-14 07:45:24 yeah, i just had irc open, should've just grepped aports 2017-09-14 07:50:03 i just wanted to prevent scrapers to get them. 2017-09-14 07:50:18 yeah, good point 2017-09-14 07:50:23 i guessed so when i saw it 2017-09-14 07:56:10 oh 2017-09-14 07:56:12 interesting 2017-09-14 07:56:12 rc-status 2017-09-14 07:56:13 Runlevel: default 2017-09-14 07:56:13 Segmentation fault 2017-09-14 07:56:43 what even 2017-09-14 07:57:11 edge? 2017-09-14 07:57:26 edge 2017-09-14 07:57:42 but 2017-09-14 07:58:00 it might be related to an USB disk that i'm trying to recover 2017-09-14 07:58:14 which have physical problem 2017-09-14 08:00:06 that's very graceful of it :P 2017-09-14 08:55:17 so uh 2017-09-14 08:55:24 this debian install.. doesn't like nvidia much 2017-09-14 08:55:38 nor do i. just got a sigsegv and complete freeze. god damn it. 2017-09-14 09:08:30 ok, i managed build chromium 61 that starts up 2017-09-14 09:08:57 but it still crashes in the tabs 2017-09-14 09:09:00 ncopa: woohoo. what have you changed? 2017-09-14 09:09:02 meh 2017-09-14 09:09:24 i rebased the no-execinfo.patch 2017-09-14 09:22:23 ncopa, i guess the void patches are of no use? 2017-09-14 09:22:38 seems they are still on 58 2017-09-14 09:34:56 they are behind yes 2017-09-14 09:35:58 i found a bunch of patches from fedora though 2017-09-14 09:36:11 fedora has musl? 2017-09-14 09:49:32 no 2017-09-14 09:49:39 but they have chromium package 2017-09-14 09:50:08 and they use gtk2 and gcc 2017-09-14 09:50:36 fcolista: heimdal does not use krb5. it is an alternative krb5 implementation 2017-09-14 09:51:33 ncopa, yes you're right 2017-09-14 09:52:29 ah fedora has cgit interface again. 2017-09-14 09:52:43 looks like samba can use krb5 1.15 2017-09-14 09:53:07 samba 4.7 that is 2017-09-14 09:53:10 i'd like that 2017-09-14 09:53:12 latest samba version, 4.6.7, does not build btw 2017-09-14 09:53:29 due to ldb version we have in aport that is not supported 2017-09-14 09:53:32 the latest one 2017-09-14 09:54:10 commit 97544a0209db2fcfef7ba6807f7a210b21e480f5 2017-09-14 09:54:10 Author: Natanael Copa 2017-09-14 09:54:10 Date: Mon Jul 17 08:51:28 2017 +0000 2017-09-14 09:54:10 2017-09-14 09:54:10 samba 4.6.* does not build with 1.2: 2017-09-14 09:54:10 main/ldb: downgrade to 1.1.31 2017-09-14 09:54:10 2017-09-14 09:54:11 /usr/include/ldb_module.h:68:2: error: #error "Samba < 4.7 is not 2017-09-14 09:54:11 compatible with this version of ldb due to assumptions around read 2017-09-14 09:54:12 locks" 2017-09-14 09:54:50 Checking for system pyldb-util (>=1.1.29 <=1.1.99 !=1.1.30 !=1.1.31) : not found 2017-09-14 09:54:50 we should upgrade ldb and samba 4.7 2017-09-14 09:54:50 ERROR: System library pyldb-util of version 1.1.29 not found, and bundling disabled 2017-09-14 09:54:50 >>> ERROR: samba: all failed 2017-09-14 09:55:01 apk upgrade -U -a 2017-09-14 09:55:01 that would be cool 2017-09-14 09:56:18 ncopa, what do you mean they use gcc? 2017-09-14 09:56:41 clandmeter: google does not really support gcc anymore. they use llvm/clang 2017-09-14 09:56:54 so chromium 61 fails to build with gcc 2017-09-14 09:57:04 and fedora has patches to fix that 2017-09-14 10:01:06 sounds like even more fun then before 2017-09-14 10:05:18 yup 2017-09-14 10:06:14 they basically have forked the entire OS 2017-09-14 10:06:32 basically every existing library except glibc 2017-09-14 10:06:41 :| 2017-09-14 10:06:44 and added to their third_party dir 2017-09-14 10:07:02 fedora has some patches to use system zlib and others 2017-09-14 10:07:14 and when you dont use their 3rdparty crap you get broken tabs everywhere 2017-09-14 10:07:24 i dont know if that is the problem 2017-09-14 10:07:30 ncopa: who forked the entire OS? 2017-09-14 10:07:36 chromium 2017-09-14 10:07:37 <^7heo> ncopa: people. it always is. 2017-09-14 10:07:50 <^7heo> (what is the problem) 2017-09-14 10:08:04 chromium tabs crashes 2017-09-14 10:08:12 since chromium 60 2017-09-14 10:08:16 and i have no clue why 2017-09-14 10:08:18 as I said lately in some other channel: "google is all about open-source, but only branded by Google. if they use something and it's not google-ish, they'll fork it ultimately. they can improve some stuff, it happens, but bastardizing for their own means is also their forte, which even happens to their own products..." 2017-09-14 10:08:26 soon they will add their own kernel as well 2017-09-14 10:08:46 they already do 2017-09-14 10:08:59 you have to run chromeOS in docker on alpine :| 2017-09-14 10:09:00 they have their own kernel 2017-09-14 10:09:18 but i was thinking the same 2017-09-14 10:09:37 the only thing missing in chromium/third_party is qemu + linux kernel 2017-09-14 10:10:57 or just introduce Fuchsia's kernel :v 2017-09-14 10:10:59 qemu with zircon support... 2017-09-14 10:11:27 there was lots of #if defined (OS_FUCHSIA) introduced 2017-09-14 10:11:43 which does not support execinfo 2017-09-14 10:11:55 huh 2017-09-14 10:12:00 yeah, there's a thing or two about magenta too 2017-09-14 10:12:04 (fuchsia kernel) 2017-09-14 10:13:22 should soon say Zircon 2017-09-14 10:16:19 interesting. i didnt know they were working on their own kernel 2017-09-14 10:16:36 fuchsia is the os, magenta the kernel 2017-09-14 10:17:18 <^7heo> Looks like some people are fan of Pratchett 2017-09-14 10:17:20 they recently renamed magenta to zircon 2017-09-14 10:17:29 i see 2017-09-14 10:17:36 i haven't paid much attention 2017-09-14 10:19:04 i dont care that much, i just wonder how that "micro kernel" thing is going to work out. 2017-09-14 10:19:42 ^ 2017-09-14 10:20:33 apparently they do a thing or two with musl in "zircon" 2017-09-14 10:20:50 nvm it's in third_party/, typical google 2017-09-14 10:21:31 there has been talk about google switching to musl 2017-09-14 10:21:41 but i didnt hear anything recently anymore 2017-09-14 10:23:08 yeah, i've heard some chatter 2017-09-14 10:24:07 real musl or musl-forked-by google, like the one in fuchsia? 2017-09-14 10:29:45 ncopa, do we really want mit-krb5 with samba 4.7? 2017-09-14 10:29:52 he feature set is not on par with the Heimdal build but the most important 2017-09-14 10:29:52 things, like forest and external trusts, are working. Samba uses the KDC binary 2017-09-14 10:29:52 provided by MIT Kerberos. 2017-09-14 10:30:04 Missing features: 2017-09-14 10:30:05 Missing features, compared to Heimdal, are: 2017-09-14 10:30:05 * PKINIT support 2017-09-14 10:30:05 * S4U2SELF/S4U2PROXY support 2017-09-14 11:45:14 heimdal itself is not good enough. theyhave a bundled/patched version of heimdal in samba 2017-09-14 11:45:21 that is what i wanted get rid of 2017-09-14 11:46:06 ah ok 2017-09-14 12:29:32 wondering if we might have, somehow, gnome 3.26 working without systemd.. 2017-09-14 12:31:28 kaniini. if I'm not wrong you had success with Gnome 3 2017-09-14 12:32:15 https://wiki.gentoo.org/wiki/GNOME/GNOME_Without_systemd 2017-09-14 12:33:08 https://github.com/dantrell/gentoo-project-gnome-without-systemd 2017-09-14 13:09:05 does anybody know the status of porting rust to other arches? 2017-09-14 13:16:51 ncopa it should work for arm and aarch64 fairly well 2017-09-14 13:18:04 hmm there might be some issues with musl still though for arm 2017-09-14 13:18:07 https://forge.rust-lang.org/platform-support.html 2017-09-14 13:18:18 looking at armv7-unknown-linux-musleabihf 2017-09-14 13:19:04 aarch64 seems fine though aarch64-unknown-linux-gnu but that seems eabi independant which is strange 2017-09-14 13:20:01 and x86? 2017-09-14 13:21:10 looks like there is a rust 1.20 available 2017-09-14 13:21:17 we still are on 1.17 2017-09-14 13:21:28 looks like the patches needs to be rebased 2017-09-14 13:22:09 jirutka: hi. do you know what the status is for rust? looks like rust 1.20 is out 2017-09-14 13:22:59 jirutka: do you think it is realistic to upgrade rust to 1.20 and add x86, armhf and aarch64 support for rust before 1 oct? 2017-09-14 13:23:40 ppc64le would be nice too 2017-09-14 13:25:50 hmm it seems this doc is out of date because x86 musl works 2017-09-14 13:26:42 ncopa: regarding ppc64le. Rust builds depends on llvm-libunwind and it is currently unsupported on ppc64le 2017-09-14 13:28:46 ok 2017-09-14 13:28:55 :-/ 2017-09-14 13:31:29 Shiz: do you think you can help with llvm5? 2017-09-14 13:31:43 i suppose we need an llvm5 package in parallel with llvm4? 2017-09-14 13:32:17 ncopa: leitao talked with a guy that usually does package ports for ppc64le and he will work on llvm-libunwind port soon 2017-09-14 13:32:28 nice! 2017-09-14 13:32:43 so i suppose for alpine 3.8 we can have rust for ppc64le 2017-09-14 13:33:06 rdutra, well, we still need to see if libunwind could be ported in time. 2017-09-14 13:33:22 ncopa, but we are going to try to have it, yes. 2017-09-14 13:34:09 ncopa, rdutra: Are you compiling rust with LLVM? 2017-09-14 13:34:54 we are compiling rust with rust :) 2017-09-14 13:35:13 but it seems that llvm-libunwind is a dependency yes 2017-09-14 13:35:28 ncopa I think he means are you using vanilla LLVM now instead of rebuilding the rust LLVM fork. 2017-09-14 13:35:29 ncopa: it is a bit "weird" because there is already a rust for ppc64le 2017-09-14 13:35:42 i wonder how they did that 2017-09-14 13:35:47 since rust 1.18 this should be possible 2017-09-14 13:36:02 before 1.18 they were using llvm 3.9 2017-09-14 13:36:04 maybe llvm-libunwind is no longer a hard dep? 2017-09-14 13:37:13 shiz: are you here :D 2017-09-14 13:37:48 ncopa: well, in rust webpage there is a powerpc64le-unknown-linux-gnu .tar.gz under rust stable 1.2.0 2017-09-14 13:38:02 ncopa: maybe it is not.. 2017-09-14 13:40:51 does rust use LLVM as the backend? 2017-09-14 13:41:09 i dont know 2017-09-14 13:41:34 apparently rust uses something from llvm 2017-09-14 13:41:39 but exactly what i dont know 2017-09-14 13:41:48 I think they use the LLVM IR as the backend 2017-09-14 13:42:03 I think rdutra can check if llvm libunwind is really a hard requirement for rust. 2017-09-14 13:42:12 it does not seem to be on other distros I am looking at 2017-09-14 13:42:21 leitao: I can try to bootstrap rust 2017-09-14 13:42:25 rdutra, nice 2017-09-14 13:42:29 leitao: like we did in go 2017-09-14 13:43:10 martell, do you have llvm-libunwind on aarch64?> 2017-09-14 13:46:28 leitao: I do not 2017-09-14 13:48:44 rust uses llvm as the backend much like swift does 2017-09-14 14:05:03 martell: Shiz is on vacation 2017-09-14 14:05:14 ahh thanks skarnet 2017-09-14 14:06:48 I was wondering what to do about llvm 5.0 2017-09-14 14:07:09 I believe it is possible to build it for alpine without any musl patches anymore 2017-09-14 14:07:31 And you should be able to have a standalone toolchain with compiler-rt and lld 2017-09-14 14:07:45 libc++ llvm-libunwind etc 2017-09-14 14:08:23 Looking at the tech talks coming up for the LLVM meetup we should also be able to build the kernel with clang also. 2017-09-14 14:08:57 See: http://llvm.org/devmtg/2017-10/#talk21 2017-09-14 14:13:36 ncopa what are you thoughts on using clang instead of gcc for bootstrapping if we can make lld link the kernel before 6.0 ? 2017-09-14 14:14:36 rather I mean llvm toolchain vs gcc + binutils 2017-09-14 14:44:16 martell: not for alpien v3.7 2017-09-14 14:44:22 but we talked about it for 3.8 2017-09-14 14:44:29 but i want rust fixed first 2017-09-14 14:44:46 ok, interesting observation 2017-09-14 14:44:59 does not seem like chromium crashes with --no-sandbox 2017-09-14 16:49:12 Could someone point me in the right direction. I'm trying to build an alpine image that is a LiveCD that has preconfigured packages, users and settings on boot. No actual setup to disk it has it all predefined and runs entirely in RAM 2017-09-14 19:09:22 <_ikke_> thegenious: You can use lbu / apkoverlay 2017-09-14 19:16:18 ncopa: strange, what env is it running in? 2017-09-14 23:58:22 I don't have the juice at hand to build chromium, does anyone have the 61 apk? 2017-09-15 04:19:10 anyone tried the go tests lately? it looks like the -src subpkg bug is fixed upstream. I'm going to try it, watch it break, then quietly sweep it away and pretend I never looked 2017-09-15 04:20:29 foxkit: toss me the results, i'll have a look if it breaks bad 2017-09-15 04:22:42 can anyone explain why it wants to build openbsd and darwin targets for the 80386? 2017-09-15 04:22:50 that's just bizarre.. 2017-09-15 04:23:04 wat 2017-09-15 04:23:19 Installed Go for darwin/386 in /usr/src/aports/community/go/src/go 2017-09-15 04:23:21 Installed commands in /usr/src/aports/community/go/src/go/bin 2017-09-15 04:23:42 I am 100% sure I did not type this on my macbook... 2017-09-15 04:23:53 it doesn't even have abuild installed, and I don't think it'd work on OS X anyway.. 2017-09-15 04:24:35 I am also 99.9% sure you cannot run Darwin on an 80386 2017-09-15 04:25:04 the initial Tiger release (10.4.3i) required SSE2 which is Pentium 4 or higher, and anything past Mavericks (10.9) requires 64-bit host 2017-09-15 04:25:24 that is very bizarre 2017-09-15 04:27:14 windows/386!!! 2017-09-15 04:27:23 can I run it on Windows 3.0 then? 2017-09-15 04:28:01 https://electricthrift.files.wordpress.com/2013/04/w386_front.jpg 2017-09-15 04:34:01 it's definitely breaking bad 2017-09-15 04:40:39 danieli: https://bpaste.net/show/199857592a55 test starts at l 6245 2017-09-15 04:51:07 thanks foxkit 2017-09-15 06:57:41 foxkit: it's because the go toolchain can natively cross-compile 2017-09-15 06:58:29 Xe: but I don't want to ship a Darwin 32-bit x86 Go toolchain for my 64-bit PowerPC distro 2017-09-15 06:58:33 this just seems bizarre 2017-09-15 06:58:37 why is it /doing/ it 2017-09-15 06:58:45 like, I understand the option; that could even be useful to people 2017-09-15 06:58:51 but I don't want it 2017-09-15 06:58:58 what is the abuild exactly? 2017-09-15 06:59:04 it's community/go 2017-09-15 06:59:12 I think I see the reason it is happening 2017-09-15 06:59:12 okay, i'm gonna do a sanity check 2017-09-15 06:59:17 but I don't understand the point... 2017-09-15 06:59:34 that's a weird loop, yeah, I think I see where it is doing it but WHY 2017-09-15 06:59:37 there is no justification... 2017-09-15 06:59:40 OH 2017-09-15 06:59:57 okay, it looks like this was doing the old way of making a cross-compiling toolchain 2017-09-15 07:00:03 feel free to rip that out entirely 2017-09-15 07:00:30 it won't make a difference to the 1 person who is actually targeting Windows 3.0 or whatever using alpine? 2017-09-15 07:01:01 go doesn't support windows 3.0, but yes it shouldn't affect that 2017-09-15 07:01:11 in fact lemme hack at it, i have a macbook in front of me 2017-09-15 07:06:15 foxkit: oh, that's to create cross-compiler subpackages 2017-09-15 07:06:26 something that's also not needed after go 1.5 2017-09-15 07:06:48 Xe: but but windows/386 2017-09-15 07:06:55 that *means* something! 2017-09-15 07:07:16 Xe: https://electricthrift.files.wordpress.com/2013/04/w386_front.jpg see 2017-09-15 07:09:04 foxkit: https://github.com/golang/go/tree/master/src/cmd/link/internal 2017-09-15 07:09:14 cmd/link is the linker go uses 2017-09-15 07:09:47 Inferno OS? o.O 2017-09-15 07:10:00 wow, this is legit Lucent code 2017-09-15 07:10:15 I ... I may /know/ the daughter of the author of this code. 2017-09-15 07:10:34 this is way too trippy for me I'm just gonna trust you, you're the Go expert 2017-09-15 07:10:49 for a long time, go's compiler was literally the plan 9 compiler suite 2017-09-15 07:11:02 modified ofc, but in go 1.5 they ported the compiler to go 2017-09-15 07:11:20 in the process they gained the ability to cross-compile without having to set up a cross-compiler 2017-09-15 07:14:45 danieli: i have packaged my builddir for chromium 2017-09-15 07:14:48 its 6G 2017-09-15 07:15:46 hi ncopa! before I bed, did you get a chance to see what I wrote earlier? just curious, if you are preparing reply I understand 2017-09-15 07:15:59 no i didnt 2017-09-15 07:16:52 go tests? 2017-09-15 07:17:00 https://github.com/alpinelinux/aports/pull/2312 it's conflicting because i forgot to `git pull` my aports repo, but a fix to that is coming 2017-09-15 07:17:43 ncopa: https://bpaste.net/raw/2bae4d816322 2017-09-15 07:17:48 much earlier than that, hehe 2017-09-15 07:19:54 thanks 2017-09-15 07:20:00 i suppose i reply in email 2017-09-15 07:20:05 if thats ok 2017-09-15 07:20:14 ok to cc alpine-devel? 2017-09-15 07:20:34 ncopa: yes definitely 2017-09-15 07:21:26 foxkit: for testing go i'd suggest running `src/race.bash` in the go tree if you haven't found that already 2017-09-15 07:23:12 Xe: the tests are run by 'run.bash' 2017-09-15 07:23:21 Xe: and a lot of them fail 2017-09-15 07:23:33 Xe: I have changed it slightly, let me push what I have WIP 2017-09-15 07:23:33 ACTION tests 2017-09-15 07:25:12 i'm testing gnome 3.26 2017-09-15 07:25:35 WARNING: CPU: 0 PID: 3031 at /home/buildozer/aports/main/linux-hardened/src/linux-4.9/drivers/gpu/drm/drm_irq.c:1254 drm_wait_one_vblank+0x1a9/0x222 [drm] 2017-09-15 07:25:41 segfaults 2017-09-15 07:26:07 https://dpaste.de/tBXg 2017-09-15 07:28:11 ncopa: i guessed it'd be huge 2017-09-15 07:28:19 fcolista: likely needs paxmarking 2017-09-15 07:28:24 ah 2017-09-15 07:28:26 i don't run -hardened on my laptop 2017-09-15 07:28:30 good point kaniini 2017-09-15 07:29:02 those are the commits i've in queus 2017-09-15 07:29:04 https://dpaste.de/CJ1H 2017-09-15 07:29:06 *queue 2017-09-15 07:29:41 i think ncopa liked have gnome 3.26 for al 3.7 2017-09-15 07:30:16 fcolista: https://code.foxkit.us/adelie/aports/commit/e004b7d1db95ae88c3d2654580a304061504913d https://code.foxkit.us/adelie/aports/commit/4eb5d526b1ac9928e4c402e853f9923b7e2fab20 2017-09-15 07:30:41 fcolista: if you are already modernising libXdamage and libXcomposite, these are additional fixes we needed at Adélie, not sure if you have them all locally 2017-09-15 07:30:55 there are actually about 80 more of these thingies 2017-09-15 07:30:57 lol 2017-09-15 07:31:08 foxkit, I don't actually 2017-09-15 07:31:23 good that you mention that 2017-09-15 07:32:18 foxkit has a commit or two (tee hee) that he wants to merge into aports 2017-09-15 07:32:31 they :) 2017-09-15 07:32:34 and yeah.. 2017-09-15 07:32:48 This branch is 435 commits ahead, 115 commits behind alpinelinux:master. 2017-09-15 07:32:57 a commit or 435 2017-09-15 07:33:26 not really; about 30-40 of those are adelie-specific configuration stuffs that don't really make sense for Alpine 2017-09-15 07:33:36 like enabling jpeg2k support in gdk-pixbuf 2017-09-15 07:33:46 this is mainly for photographers I know 2017-09-15 07:34:08 kaniini, i'm not sure what i need to paxmark 2017-09-15 07:34:19 if gnome-session or gnome-shell 2017-09-15 07:35:17 bah 2017-09-15 07:35:18 grsec: denied resource overstep by requesting 2048 for RLIMIT_NOFILE against limit 1024 for /usr/bin/Xorg[X:2721] uid/euid:1000/0 gid/egid:1000/1000, parent /usr/bin/xinit[xinit:2720] uid/euid:1000/1000 gid/egid:1 2017-09-15 07:35:24 this too segfaults 2017-09-15 07:35:38 having gnome3 + hardened kernel it's quite mad 2017-09-15 07:36:42 gnome-shell/gnome-shell-3.22.3.ebuild: pax-mark E "${ED}usr/bin/gnome-shell"{,-extension-prefs} 2017-09-15 07:36:44 gnome-shell/gnome-shell-3.22.3.ebuild: pax-mark m "${ED}usr/bin/gnome-shell"{,-extension-prefs} 2017-09-15 07:36:48 this from Gentoo's build 2017-09-15 07:37:00 only paxmark I see in any gnome3 package 2017-09-15 07:37:09 but that doesn't mean it's right, only that's what they ship 2017-09-15 07:37:33 it's a start, at least 2017-09-15 08:00:47 umh 2017-09-15 08:00:51 this is a bug 2017-09-15 08:00:51 drivers/gpu/drm/drm_irq.c:1254 drm_wait_one_vblank+0x130/0x15d [drm] 2017-09-15 08:01:02 does the same with linux-vanilla 2017-09-15 08:01:11 that definitely looks like a bug 2017-09-15 08:01:17 before gnome is launched 2017-09-15 08:01:22 bah 2017-09-15 08:01:46 i bet it something related to power management of intel module.. 2017-09-15 08:01:47 i915 2017-09-15 08:03:58 fcolista: tried drm intel nightly? 2017-09-15 08:05:03 nope 2017-09-15 08:06:09 umh... 2017-09-15 08:08:35 seems i've found the issue 2017-09-15 08:08:50 15.275911] [drm] forcing SVIDEO-1 connector OFF 2017-09-15 08:09:04 I had to disable SVIDEO-1 2017-09-15 08:09:13 by passing video=SVIDEO-1:d on boot 2017-09-15 08:09:57 \o/ 2017-09-15 08:10:06 got gnome 3.26 running! 2017-09-15 08:10:12 h neat 2017-09-15 08:10:14 ah* 2017-09-15 08:12:53 no I want to try the grsec kernel 2017-09-15 08:12:56 *now 2017-09-15 08:18:30 fcolista: GMA X3100 series? 2017-09-15 08:18:38 i915 2017-09-15 08:22:35 fcolista: yeah, but this is known bug with GMA X3100 and X2700 series 2017-09-15 08:23:38 fcolista: I believe intel said that they were 'deprecating' the xf86-video-intel in favour of raw modesetting, you might try the modesetting driver instead of intel and see if that solves the issue. I believe mesa i965_dri still works with it.. 2017-09-15 08:23:47 it's been almost a year since I dealt with intel gpus, heh 2017-09-15 08:24:14 foxkit, thx! I removed xf86-video-intel 2017-09-15 08:24:33 what is the modesetting driver? 2017-09-15 08:24:43 i've also mesa-intel 2017-09-15 08:24:57 sorry 2017-09-15 08:24:59 mesa-dri-intel 2017-09-15 08:29:29 stupid question: check() should be after the build() and before package() right? 2017-09-15 08:30:14 yes 2017-09-15 08:30:17 and not stupid :) 2017-09-15 08:30:25 fcolista: it's built in to X server, there is no package 2017-09-15 08:30:37 foxkit, ah ok! 2017-09-15 08:30:45 that's why is raw.. 2017-09-15 08:30:56 so I only need to remove the *-intel packages 2017-09-15 08:31:09 mesa-dri-intel is still needed for 3D accel 2017-09-15 08:31:11 (in my case, just mesa-dri-intel) 2017-09-15 08:31:21 xf86-video-intel is all you need to remove, it's the X driver 2017-09-15 08:31:31 ok 2017-09-15 08:31:49 thanks for this insighful info 2017-09-15 08:32:50 glad to help! 2017-09-15 08:37:50 function order matters in an apkbuild? 2017-09-15 08:38:39 clandmeter: it in theory shouldn't? 2017-09-15 08:39:16 its a shell script, so it shouldnt matter i believe. 2017-09-15 08:39:35 maybe cosmetics can be a concern. 2017-09-15 08:42:52 it really shouldn't matter, but cosmetics is a very minor one, i guess 2017-09-15 08:42:58 order the functions in the order of the process 2017-09-15 08:44:23 no doesn't matter 2017-09-15 08:44:40 it just help to follow the correct process 2017-09-15 08:44:46 first build, then check, then package 2017-09-15 08:44:57 but it does not affect the functionality of course 2017-09-15 08:45:00 cereal, then milk, THEN bowl 2017-09-15 08:45:01 it's cosmetic :) 2017-09-15 08:45:13 yeah, a sort of :) 2017-09-15 08:51:35 i've this warning but I'm unsure if (or I should) fix: 2017-09-15 08:51:36 >>> WARNING: gnome-desktop-dev*: Could not find any provider for pc:gsettings-desktop-schemas 2017-09-15 08:51:49 gsettings-desktop-schemas actually provides th .pc 2017-09-15 08:51:54 *the .pc 2017-09-15 08:52:37 i think i should create the dev() function in gnome-desktop and add a specific depends="gsettings-desktop-schemas" 2017-09-15 08:53:01 ./gsettings-desktop-schemas/usr/share/pkgconfig/gsettings-desktop-schemas.pc 2017-09-15 08:54:06 what is strange is that gsettings-desktop-schemas is compiled with arch="all" even if it's noarch 2017-09-15 09:01:53 ncopa: 2017-09-15 09:04:37 hi Gottox 2017-09-15 09:04:52 whoops, misstapped, but hi ncopa ;) 2017-09-15 09:05:12 fcolista: i think abuild expecit it to be in /usr/lib/pkgconfig 2017-09-15 09:05:23 instead of /usr/share/pkgconfig 2017-09-15 09:05:31 aah 2017-09-15 09:05:35 that's why! 2017-09-15 09:05:41 ok, i need to move it 2017-09-15 09:05:53 do you guys have any plan to ship gnome with musl? - on void it builds, but at least 3.24 segfaults. 2017-09-15 09:05:53 should be a configure option for it, hopefully 2017-09-15 09:06:09 Gottox: looks like we are working on it 2017-09-15 09:06:19 PKG_CONFIG_LIBDIR 2017-09-15 09:06:42 but currently im trying to figure out why chromium sandbox crashes 2017-09-15 09:07:03 i don't have a juicy build host locally available 2017-09-15 09:07:15 that's why we still at chromium-58, iirc 2017-09-15 09:07:21 Gottox: correct 2017-09-15 09:07:25 danieli: i can let you download my build, its 6G 2017-09-15 09:07:31 ncopa: that works for me 2017-09-15 09:07:37 what arch, x86_64? 2017-09-15 09:07:41 yes 2017-09-15 09:07:46 good, same 2017-09-15 09:09:31 the debug build failed for some reason 2017-09-15 09:09:42 is_debug=true 2017-09-15 09:09:50 so its build with is_debug=false 2017-09-15 09:12:47 danieli: https://dev.alpinelinux.org/~ncopa/alpine/chromium-debug.tar.gz 2017-09-15 09:12:59 cd aports/community && rm -r chromium 2017-09-15 09:13:00 i didn't know about dev.a.o, cool 2017-09-15 09:13:03 thanks 2017-09-15 09:13:10 tar -zxf chromium-debug.tar.gz 2017-09-15 09:13:25 you need 6GB free disk space 2017-09-15 09:13:36 worst case i'll put it on my SAN 2017-09-15 09:13:41 cd chromium 2017-09-15 09:13:43 rm -r pkg 2017-09-15 09:13:44 i have more than enough on this box though 2017-09-15 09:13:53 abuild deps rootpkg index 2017-09-15 09:14:03 should build you the package 2017-09-15 09:14:44 you can modify stuff and rebuild binary: 2017-09-15 09:15:02 cd src/chromium-61*/. 2017-09-15 09:15:27 2017-09-15 09:16:09 ninja -C out/Release chrome chrome_sandbox chromedriver 2017-09-15 09:16:40 it works with --no-sandbox 2017-09-15 09:16:50 yeah, i'll do some debugging 2017-09-15 09:17:13 the sandboxed child process gets killed 2017-09-15 09:17:23 +++ killed by SIGSYS +++ 2017-09-15 09:17:40 apparently it only generates core dump with non-threading programs 2017-09-15 09:17:48 and no core dump when its threaded 2017-09-15 09:18:10 strace does not tell which syscall it is. kernel juust kills it 2017-09-15 09:18:53 i'll check it out 2017-09-15 09:18:55 thank you 2017-09-15 09:23:24 hum 2017-09-15 09:23:31 looks like its possible to have kernel to log it 2017-09-15 09:23:39 should be, yes 2017-09-15 09:23:43 need enable CONFIG_AUDIT 2017-09-15 09:23:46 its disabled 2017-09-15 09:33:01 ok, it was relatively easy to enable it, not tons of CONFIG_SOMETHING_AUDIT questions 2017-09-15 09:38:42 i wonder if we should enable it in our default config? 2017-09-15 09:39:12 looks like there is a boot option: audit=1|0 2017-09-15 09:39:38 we probably want have it disabled by default at runtime 2017-09-15 09:39:55 it's used for selinux too 2017-09-15 09:39:58 for perfomance reasons 2017-09-15 09:40:02 agreed 2017-09-15 09:40:05 it should be disabled by default 2017-09-15 09:40:32 question is if we want enable it in kernel config 2017-09-15 09:40:37 but it would be still nice to have it compiled in, I think 2017-09-15 09:40:40 looks useful for this purpose at least 2017-09-15 09:40:50 przemoc: thats what im thinking 2017-09-15 09:40:52 could leave the option there and compile it in 2017-09-15 09:40:59 won't be a lot larger iirc, and it'll be useful 2017-09-15 09:41:19 i dunno if audit=0 is the default or not 2017-09-15 09:42:01 i finally have hope to find the syscall 2017-09-15 09:42:34 file .pc shouldn't be in -dev ? 2017-09-15 09:42:42 it should 2017-09-15 09:42:58 PKG_CONFIG_LIBDIR is not honerd 2017-09-15 09:43:03 *honored 2017-09-15 09:43:12 not surprised 2017-09-15 09:43:19 makefile is bugged.. 2017-09-15 09:43:45 i'm going to add gsettings-desktop-schemas-dev 2017-09-15 09:43:47 technically, /usr/share/pkgconfig is correct and /usr/lib/ is wrong 2017-09-15 09:43:48 subpackage 2017-09-15 09:44:08 if the .pc is the only file in the subpackage, then its not worth it 2017-09-15 09:44:33 actually there are header files 2017-09-15 09:44:40 then -dev is needed 2017-09-15 09:44:58 im gonna reboot 2017-09-15 09:45:16 yes..and this means several packages mus be rebuild... 2017-09-15 09:53:55 Sep 15 10:53:07 ncopa-desktop kern.warn kernel: [ 126.556124] audit_printk_skb: 411 callbacks suppressed 2017-09-15 09:53:55 Sep 15 10:53:07 ncopa-desktop kern.notice kernel: [ 126.556128] audit: type=1701 audit(1505469187.186:51): auid=4294967295 uid=1000 gid=1000 ses=4294967295 pid=4376 comm="TaskSchedulerRe" exe="/usr/lib/chromium/chrome" sig=31 2017-09-15 09:59:50 i'm thinking it might have to do with seccomp 2017-09-15 10:03:48 relevant? https://github.com/lluixhi/musl-extras/issues/10 2017-09-15 10:03:58 @ ncopa 2017-09-15 10:04:05 foxkit, I'm going merge your two patches for libxdamage and libxcomposite if you are right 2017-09-15 10:04:15 *if you are ok 2017-09-15 10:05:25 im relatively sure it has to do with seccomp 2017-09-15 10:05:34 i think musl does a syscall that is not whitelisted 2017-09-15 10:05:47 and the seccomp sandbox kills it 2017-09-15 10:06:08 that sounds very likely 2017-09-15 10:07:55 yes, that looks relevant 2017-09-15 10:07:57 https://github.com/lluixhi/musl-extras/issues/10#issuecomment-328272349 2017-09-15 10:08:03 ncopa, sounds similar what i fixed in the beginning with dalias. i think thats the musl-sandbox patch 2017-09-15 10:08:59 i did some debugging, then some googling, and found that 2017-09-15 10:09:20 i bet that is the fix 2017-09-15 10:09:32 judging by my common sense, that seems likely 2017-09-15 10:11:07 clandmeter: yes its similar to musl-sandbox.patch 2017-09-15 10:14:07 i just discovered noname-linux 2017-09-15 10:14:13 another musl based distro 2017-09-15 10:16:18 i've just looked over the repo 2017-09-15 10:17:15 2. emoji wcwidth patch for musl 2017-09-15 10:17:16 LOL 2017-09-15 10:17:37 looks like nobody ever cared to upstream any of the sane musl patches. 2017-09-15 10:18:26 :-( 2017-09-15 10:18:27 looks like it 2017-09-15 10:18:33 still crashes 2017-09-15 10:18:35 #justiceformusl 2017-09-15 10:23:22 ok i need to figure how to configure auditd to only log seccomp issues 2017-09-15 10:27:44 ncopa: AUDIT_SECCOMP 2017-09-15 10:27:52 iirc anyway 2017-09-15 10:28:10 i think thats enabled already in kernel 2017-09-15 10:28:24 i just need the auditctl ... seccomp something 2017-09-15 10:29:16 i want figure out what sends the SIGSYS 2017-09-15 10:29:52 ncopa: `dmesg | grep type=1326`? 2017-09-15 10:30:22 nothing 2017-09-15 10:30:27 whats 1326? 2017-09-15 10:32:15 it varies, some times type=SYSCALL, some times type=1326 2017-09-15 10:32:17 1326* 2017-09-15 10:32:41 it does not log all syscalls 2017-09-15 10:32:58 and i dont really want flood it with logging all syscalls 2017-09-15 10:33:17 i want it only log when seccomp ends up killing the process with SIGSYS 2017-09-15 10:33:23 which does not generate coredump 2017-09-15 10:40:21 ncopa: try `-a entry,always -S kill -k teste_kill` in audit.rules 2017-09-15 10:40:42 the `-k $name` part is arbitrary, it's used for grepping the logs 2017-09-15 10:46:09 hum 2017-09-15 10:46:29 maybe i didnt restart chromium after the last rebuild 2017-09-15 10:46:41 looks like it no longer crashes 2017-09-15 10:46:58 it was probabaly that patch 2017-09-15 10:47:43 type=SYSCALL msg=audit(1505472361.193:7840534): arch=c000003e syscall=62 success=yes exit=0 a0=1adb a1=f a2=0 a3=21eac00fdcd3c5 items=0 ppid=4066 pid=6644 auid=4294967295 uid=1000 gid=1000 euid=1000 suid=1000 fsuid=1000 egid=1000 sgid=1000 fsgid=1000 tty=(none) ses=4294967295 comm="TaskSchedulerSi" exe="/usr/lib/chromium/chrome" key="teste_kill" 2017-09-15 10:48:40 hum, its the app that sends the kill 2017-09-15 10:48:55 chrome 61 works fine with the patch? 2017-09-15 10:49:04 looks like that yes 2017-09-15 10:49:09 neat-o 2017-09-15 10:49:24 seccomp usually complains a lot at the chrom(e|ium) sandbox 2017-09-15 10:53:25 i'll reboot again and test with the pax'ed kernel 2017-09-15 11:10:26 ncopa: what did you figure out? 2017-09-15 11:23:36 chromium runs like a charm now 2017-09-15 11:23:46 i must have failed restarting chrome properly earlier 2017-09-15 11:23:56 im gonna push the patch 2017-09-15 11:24:02 that's great! 2017-09-15 11:24:07 chrome 62 here we come 2017-09-15 11:37:11 ncopa: or it's some audit presence aftermatch 2017-09-15 11:37:15 did you retest w/o it? 2017-09-15 11:39:29 audit should be passive 2017-09-15 11:42:34 i rebooted to kernel without CONFIG_AUDIT 2017-09-15 11:42:39 and it still runs 2017-09-15 11:42:48 should != is, so it's good to be double sure. if you have CONFIG_AUDITSYSCALLS turned on, it's slightly invasive according to Andy's comment quoted in https://lwn.net/Articles/600568/, but not sure if ncopa turned it on too 2017-09-15 11:42:56 if it works, then great 2017-09-15 11:42:59 it works 2017-09-15 11:43:11 and i am super happy :) 2017-09-15 11:43:20 its a perfect friday :) 2017-09-15 11:43:24 great! 2017-09-15 11:43:35 danieli: thank you very much! 2017-09-15 11:43:42 <3 2017-09-15 11:43:50 if you're into chrome, ncopa, could you also add those deps I mentioned once here, so headless mode can be run w/o itch? 2017-09-15 11:44:01 whats worse is that it fails to download from google :-/ 2017-09-15 11:44:09 ncopa: why? 2017-09-15 11:44:15 elaborate? 2017-09-15 11:44:37 >>> chromium: Cleaning temporary build dirs... 2017-09-15 11:44:37 % Total % Received % Xferd Average Speed Time Time Time Current 2017-09-15 11:44:37 Dload Upload Total Spent Left Speed 2017-09-15 11:44:37 3 262k 3 9299 0 0 9299 0 0:00:28 --:--:-- 0:00:28 34062 2017-09-15 11:44:37 curl: (23) Failed writing body (0 != 9299) 2017-09-15 11:44:50 bah 2017-09-15 11:44:53 disk full 2017-09-15 11:44:53 oh the tarball 2017-09-15 11:45:02 oh, well shit 2017-09-15 11:45:03 explains it 2017-09-15 11:49:58 I'll repeat my message, just in case, so you don't have to look for it 2017-09-15 11:50:00 2017-09-05 12:28 < przemoc> [edge] to be able to start chromium, which already pulled 64 packages beside chromium, in headless mode I still needed to install eudev-libs (FATAL:udev_loader.cc(38)] Check fail ed: false.), ttf-opensans (FATAL:platform_font_linux.cc(63)] Check failed: typeface. Could not find any font: sans, sans). 2017-09-15 11:52:04 as I said back then, I didn't check it with other fonts, simply used the most obvious result from `apk search sans` 2017-09-15 12:15:56 przemoc: did you have a PR? 2017-09-15 12:17:01 przemoc: how can i test headless? 2017-09-15 12:17:42 i suppose i should fix it while at it 2017-09-15 12:20:03 ncopa: --headless simply 2017-09-15 12:20:16 usually used with --remote-debugging-port 2017-09-15 12:20:46 what is it used for? 2017-09-15 12:21:06 extracting metadata, automated tests, etc 2017-09-15 12:21:14 screenshots 2017-09-15 12:24:18 ok i can reproduce the issue 2017-09-15 12:24:35 installing ttf-dejavu also works 2017-09-15 12:24:49 that's the better fix then, i'd say 2017-09-15 12:24:51 opensans is a better alternative? 2017-09-15 12:25:00 agreed, if it works 2017-09-15 12:28:44 heh 2017-09-15 12:29:01 ttf-opensans exists in both testing and unmaintained 2017-09-15 12:29:18 strange 2017-09-15 12:29:26 does it even need maintaining? 2017-09-15 12:33:55 fcolista: looks like you added back ttf-opensans to testing 2017-09-15 12:34:15 git log unmaintained/ttf-opensans testing/ttf-opensans 2017-09-15 12:34:53 the original opensans had some config 2017-09-15 12:35:22 ncopa, i didnt' noticed that it was on unmaintained actually 2017-09-15 12:35:36 i figured that 2017-09-15 12:35:43 i wonder if we could clean it up a bit 2017-09-15 12:36:20 sure 2017-09-15 12:36:42 I've several APKBUILD in queue I'm working on for gnome 3.26 2017-09-15 12:36:48 I can add this to the list too 2017-09-15 12:37:11 not sure where ttf-opensans original package comes from 2017-09-15 12:37:20 apparently host it on dev.a.o 2017-09-15 12:37:31 but from where was it originally taken 2017-09-15 12:37:43 looks like the one in testing found it from some frugalware mirror 2017-09-15 12:37:50 hash it and google the hash 2017-09-15 12:40:19 or just look what fedora does: http://pkgs.fedoraproject.org/cgit/rpms/open-sans-fonts.git/tree/ 2017-09-15 12:40:47 fcolista: i think i will just move it to community for now 2017-09-15 12:40:55 we can clean it up afterwards 2017-09-15 12:41:00 for what is needed? 2017-09-15 12:41:08 chromium --headless 2017-09-15 12:41:12 ah ok 2017-09-15 12:41:16 i want push chromium 2017-09-15 12:41:20 sure 2017-09-15 12:41:25 might as well perfect it a bit before pushing it 2017-09-15 12:41:28 making common features work 2017-09-15 12:41:55 ? 2017-09-15 12:42:11 people are gonna want to use alpine for chromium --headless use 2017-09-15 13:02:17 danieli, did you mean that headless chromium does not work on alpine? 2017-09-15 13:02:41 andypost: 59 is the latest chromium that is in alpine repos, we're working on 62 2017-09-15 13:03:23 it works, its just the dependencies that is wrong 2017-09-15 13:03:29 well yeah 2017-09-15 13:04:06 danieli, 59 is headless https://developers.google.com/web/updates/2017/04/headless-chrome 2017-09-15 13:04:29 *in alpine* 2017-09-15 13:05:07 ncopa, please care https://github.com/alpinelinux/aports/pull/2308 2017-09-15 13:07:23 andypost: pushed 2017-09-15 13:07:47 i just updated the commit message a bit 2017-09-15 13:10:16 ncopa, thanx a lot, btw incredible amount of cves fixed 2017-09-15 13:10:27 i saw 2017-09-15 13:13:06 ncopa: sorry, wasn't around. no, I didn't send PR, because you were not convinced back then that ttf-opensans should be used and as I am not X user AL, I couldn't tell what would be the best font-related added dep. 2017-09-15 13:13:32 im going for ttf-opensans 2017-09-15 13:14:54 http://www.tcpdump.org/tcpdump-changes.txt impressive number of CVEs indeed 2017-09-15 13:18:03 chromium build on x86 is still broken 2017-09-15 13:18:06 hump 2017-09-15 13:18:14 that was the original thing i wanted to fix 2017-09-15 13:21:40 ncopa: what fails? 2017-09-15 13:22:13 gn bootstrap 2017-09-15 13:22:51 http://tpaste.us/Lvr4 2017-09-15 13:22:59 <^7heo> There's a new alpine every 6 months right? 2017-09-15 13:23:46 correct 2017-09-15 13:23:57 next is scheduled for november 2017-09-15 13:24:12 <^7heo> So you guys could change your release number to match firefox's :D 2017-09-15 13:27:14 AL 4 is reserved for ground-breaking improvements/changes, I presume. e.g. busybox replaced by toybox by default, gcc replaced by clang by default, openrc & co replaced by s6 & co. :) 2017-09-15 13:27:51 dreaming far i see 2017-09-15 13:28:04 <^7heo> why would one restrict their dreams? 2017-09-15 13:28:08 <^7heo> it's the only freedom we've got left. 2017-09-15 13:28:42 ncopa: FYI, you've reviewed a number of my recent PRs. I've just closed three (testing/libspatialite, testing/nitro, and testing/laszip) in favor of a single, combined PR https://github.com/alpinelinux/aports/pull/2314 for testing/pdal (as this was all really just leading up to the PDAL aport anyway). Appreciate the review from you or another maintainer when you get the chance. 2017-09-15 13:29:29 chambbj: can you please ping me next week 2017-09-15 13:29:34 AL4, simply 4 next generations 2017-09-15 13:29:38 good tagline is important 2017-09-15 13:30:02 Sure, no problem! Thx again. 2017-09-15 13:31:00 ah, there should be one more big change beside those 3 already mentioned. not sure what it could be. 2017-09-15 13:31:12 <^7heo> przemoc: you make it sound like Alpine is gonna be 4 for 2 decades or so 2017-09-15 13:31:53 apk tools rewrite 2017-09-15 13:32:09 right! 2017-09-15 13:32:50 I forgot about it. 2017-09-15 13:33:01 isn't it kind of on-going thing? 2017-09-15 13:33:29 <^7heo> I think it is yes 2017-09-15 13:37:43 chromium pushed 2017-09-15 13:38:56 nice! 2017-09-15 13:39:02 great! 2017-09-15 13:45:10 shiz: http://dup.pw/aports/21e1af7f 2017-09-15 13:46:23 shiz: do you have that linux-4.1.y-rpi-20170501.patch stored some place? 2017-09-15 13:47:50 shiz is on vacation IIRC 2017-09-15 13:48:18 correct ^ 2017-09-15 15:00:10 Is connman still being maintained? 2017-09-15 19:37:54 ncopa: is there anything craftyguy or I could do to help getting this PR in faster? https://lists.alpinelinux.org/alpine-aports/4884.html 2017-09-15 19:38:21 ncopa: also are patches prefered via e-mail, or via github PRs (because PRs will trigger travis builds)? 2017-09-15 19:38:53 *wrong terminology. that of course is a mail patch 2017-09-15 20:18:07 fcolista: I am quite okay with the merging. almost that entire repo should be hopefully merged someday 2017-09-16 09:25:59 martell: got a bad connection or something? 2017-09-16 09:26:01 you join and leave a lot 2017-09-16 09:34:41 XD 2017-09-16 09:34:58 just left on me 2017-09-16 09:34:59 hrmph 2017-09-16 09:36:29 Weechat can filter those messages. there is a similar plugin for irssi, but it doesn't work that well. 2017-09-16 09:36:59 I am aware, but I don't wish to remove them 2017-09-16 09:55:36 i'm gonna make and test an alpine build for the pocket chip 2017-09-16 09:55:39 could probs distribute it 2017-09-16 18:35:19 Hi, how do I provide updated APKBUILDS when there are new package updates available? I read the relevant wiki page, and submitted a patch on the aports mailing list: http://lists.alpinelinux.org/alpine-aports/4884.html 2017-09-16 18:36:12 <_ikke_> `That should be enough 2017-09-16 18:36:16 I could potentially help with updating many other packages that are out of date (there seem to be no shortage in the repos), but I want to make sure the process in the wiki of submitting the patch on the mailing list is still correct since it seems like this one patch I sent has had no action/responses 2017-09-16 18:37:39 <_ikke_> another option would be to create a pull request on github 2017-09-16 18:37:45 <_ikke_> both options are valid though 2017-09-16 18:39:31 Ah, I didn't know there was a repo on github. 2017-09-16 18:40:26 Whichever method is easiest for y'all to review/merge, I'll go with 2017-09-17 00:33:58 /usr/lib/gcc/pentium4-foxkit-linux-musl/6.4.0/../../../../pentium4-foxkit-linux-musl/bin/ld: abuild-tar.o: file class ELFCLASS64 incompatible with ELFCLASS32 2017-09-17 06:59:51 well 2017-09-17 07:00:04 alpine now has the distinction of handling ppc32, an arch that yall don't even support, better than gentoo does 2017-09-17 07:00:15 I'm not sure if that shows how well made alpine is, or how badly made gentoo is 2017-09-17 07:00:21 either way, this works great for me! 2017-09-17 07:01:50 cool! 2017-09-17 07:02:00 did you test it on ppc32? 2017-09-17 07:03:13 did I test what? 2017-09-17 07:03:22 I'm using my aports branch with all the test suites yes 2017-09-17 07:04:58 ah I see 2017-09-17 13:09:52 for py devs, http://www.nbu.gov.sk/skcsirt-sa-20170909-pypi/ 2017-09-17 13:13:04 nice examples of infoware showing signs of malwarism 2017-09-17 13:14:09 vkrishn: are you a bot? 2017-09-17 13:25:42 danieli, let me think...... why say so? 2017-09-17 13:28:00 just had to ask, i've seen you post seemingly unrelated things once in a while 2017-09-17 13:31:59 example ? 2017-09-17 17:00:53 where is the alpine installer, in git? 2017-09-17 17:11:49 <_ikke_> alpine installer? 2017-09-17 17:13:00 > for py devs, http://www.nbu.gov.sk/skcsirt-sa-20170909-pypi/ 2017-09-17 17:13:00 well, there is no any verification at all. no surprise here unfortunately 2017-09-17 17:13:52 <_ikke_> scadu: you are too late :-) 2017-09-17 17:15:04 :( 2017-09-17 17:16:55 _ikke_: setup-alpine 2017-09-17 17:20:15 <_ikke_> alpine-conf 2017-09-17 17:20:20 <_ikke_> https://git.alpinelinux.org/cgit/alpine-conf/tree/ 2017-09-17 17:20:44 aaah okay, thank you 2017-09-18 08:05:08 there was a bugfix in the bitlbee facebook plugin... i just created a PR for aports about it. else its not working anymore 2017-09-18 08:05:53 if you need it immediately (without time to wait for the PR to be merged and built), you can just build it yourself 2017-09-18 08:15:15 i know i know... i just mean for the other users... :) 2017-09-18 08:16:08 yeah, true, but it'll probably be merged soon enough 2017-09-18 08:16:23 i pushed an mtr version bump earlier, removed a patch 2017-09-18 08:32:38 im sure about this :) 2017-09-18 08:32:50 and maybe im the only one using this package anyway ;-) 2017-09-18 08:53:03 16 Applications, 4 Desktop, 22 Development, 24 GNOME, 21 Graphics, 11 Networking, 9 Security, 27 System, 95 X11, 6 XFCE 4, 10 Misc, 7 I need an adult/ncopa 2017-09-18 08:53:09 made topic branches for all the 400 commits 2017-09-18 08:53:29 it turns out there are only 252 to actually merge 2017-09-18 08:53:59 the others are not suitable for upstream due to patching or still being tested for correctness 2017-09-18 09:25:02 ugh, git log / diff is not working very well, some of these commits are already merged 2017-09-18 09:25:16 <_ikke_> foxkit: ? 2017-09-18 09:25:39 _ikke_: so, I forked aports.git, I've been doing periodic merges 2017-09-18 09:25:54 _ikke_: git log master..upstream/master now includes some commits I already merged upstream 2017-09-18 09:26:37 <_ikke_> well, according to git these commits are not reachable from upstream/master 2017-09-18 09:26:40 I probably need to do some horrible rebase crap that will corrupt the repo and make my life miserable 2017-09-18 09:26:48 <_ikke_> foxkit: Is this repo public? 2017-09-18 09:26:52 yes it is 2017-09-18 09:27:04 <_ikke_> What is the url? 2017-09-18 09:27:05 https://code.foxkit.us/adelie/aports with a mirror at https://github.com/AdelieLinux/aports 2017-09-18 09:27:11 <_ikke_> thanks 2017-09-18 09:27:22 you will need to append '.git' to either of those URLs to pull with actual git 2017-09-18 09:27:42 <_ikke_> right 2017-09-18 09:27:47 <_ikke_> git itself is smart enough to try that though 2017-09-18 09:28:55 i only see 2 branches? 2017-09-18 09:31:50 for an example that shows up in the log but I know is merged, 4bdfdd7f12517eceaf55c99210de500debdd0463 2017-09-18 09:32:00 <_ikke_> foxkit: I see only commits that upstream has but not in alpine/master 2017-09-18 09:32:17 <_ikke_> foxkit: that's what your command shows 2017-09-18 09:32:29 clandmeter: I haven't pushed the new branches yet, I'm trying to *create* them 2017-09-18 09:32:38 _ikke_: oh sorry, I meant other way: git log upstream/master..master 2017-09-18 09:33:02 _ikke_: so this shows 4bdfdd7f12517eceaf55c99210de500debdd0463 for me, even though I know upstream/master has this commit 2017-09-18 09:33:19 I see it in git log upstream/master 2017-09-18 09:33:41 <_ikke_> foxkit: for me it returns nothing 2017-09-18 09:33:49 weird 2017-09-18 09:34:02 wonder if maybe it is a git setting that I don't know about 2017-09-18 09:34:02 <_ikke_> git log --oneline adelie/master..master | wc -l -> 0 2017-09-18 09:34:18 <_ikke_> foxkit: git rev-parse master 2017-09-18 09:34:19 oh. for me, 'upstream' remote is alpine 2017-09-18 09:34:28 and 'origin' remote is adelie 2017-09-18 09:35:00 9c15fcd8c574e9d12a8d83229deaab9e2f7b30ac 2017-09-18 09:35:09 <_ikke_> Well, from what I see, origin/master has commits that upstream/master does not have 2017-09-18 09:35:41 yes, a lot of them; but some of the ones that show up in the log command are ones that it *does* have 2017-09-18 09:35:56 maybe I just do not know how to use git properly .-. 2017-09-18 09:36:18 <_ikke_> http://tpaste.us/MMqX 2017-09-18 09:36:35 there are some changes that don't make sense to push upstream to alpine (like adding jasper to gdk-pixbuf, or adding mmx support to openssl) 2017-09-18 09:36:45 so this will never be /clean/ 2017-09-18 09:36:56 <_ikke_> right, which is not an issue 2017-09-18 09:37:30 <_ikke_> but I do not see anything in alpines master that your master does not have 2017-09-18 09:37:39 but I bring commits out of master that I want to put upstream into for-adelie branch, then offer that as a PR, and then usually n copa or f abled merges them 2017-09-18 09:38:12 _ikke_: right, my problem is the other way. I want to see commits that my master has, that alpine master does not have, and it is showing me commits that I know alpine has. 2017-09-18 09:38:25 <_ikke_> right 2017-09-18 09:38:26 such as the lutok tests commit ( sha1 4bdfdd7f12517eceaf55c99210de500debdd0463 ) 2017-09-18 09:39:22 I guess what I really want is "filter commits from X branch that would not cause a difference if applied to Y branch" 2017-09-18 09:40:17 git log --cherry almost looked promising, but it yields the same output :/ 2017-09-18 09:40:54 <_ikke_> foxkit: Did you send that commit as a patch? 2017-09-18 09:41:02 yes 2017-09-18 09:42:23 git log --pretty=fuller shows Timo committed it, almost exactly seven days after the authordate 2017-09-18 09:42:36 <_ikke_> right, that indeed means that they are two separate commits 2017-09-18 09:42:48 that is annoying 2017-09-18 09:42:50 <_ikke_> So git log will always show them as different 2017-09-18 09:42:50 :/ 2017-09-18 09:43:00 <_ikke_> git cherry should do what you want 2017-09-18 09:43:10 <_ikke_> it looks at the patch-id (hash of a normalized diff) 2017-09-18 09:43:34 <_ikke_> Unless they modified the patch 2017-09-18 09:45:53 that is indeed better, it eliminates about 100 commits for me (451 vs 558) 2017-09-18 09:45:59 thank you! 2017-09-18 09:46:35 <_ikke_> yw :-) 2017-09-18 10:00:02 is rgdoliveira here? 2017-09-18 10:02:36 nvm, ill leave a msg on github. 2017-09-18 10:36:48 https://github.com/alpinelinux/aports/pulls it's way too late for me to continue, but here are six while I sleep :) 2017-09-18 10:54:29 foxkit, dont bump pkgrel if resulting apk doesnt change. 2017-09-18 20:49:55 clandmeter: both Shiz and ncopa said pkgrel has to be bumped when adding tests or else the builders won't run the tests. 2017-09-18 20:50:03 has that changed now? 2017-09-18 20:52:17 No it didn't 2017-09-18 20:53:26 There was another commit which disabled tests 2017-09-18 20:53:30 Iirc 2017-09-18 20:58:45 foxkit, the screen commit 2017-09-18 21:08:41 <^7heo> does someone here know if adb works on alpine? 2017-09-18 22:00:12 ^7heo: it does 2017-09-18 22:00:21 at least when started in a chroot on Arch and other systems 2017-09-18 22:00:36 (I know because of that project with phones) 2017-09-18 22:21:05 clandmeter: I will fix it when I push other fixes to that branch, thanks for pointing it out. (originally that commit changed other config too, hence we revbumped it for *us*, but when I was changing the commit for alpine merge I missed it sorry) 2017-09-18 22:40:04 <^7heo> ollieparanoid[m]: thanks 2017-09-18 22:59:26 these PRs always make me nervous. hardened doesn't work right for me in virtualbox (random lockups and other general weirdness) so I can never test on a hardened box, and it makes me feel dumb when things fail on travis because it uses hardened x.x 2017-09-18 23:08:18 travis doesnt use hardened 2017-09-18 23:08:22 it uses the ubuntu kernel 2017-09-18 23:08:24 lol 2017-09-18 23:21:59 kaniini: that's odd, because I can't reproduce the test failures on my vanilla box, but I *can* reproduce it on the hardened box if I manage to get abuild to finish before it locks up 2017-09-19 06:09:15 ncopa: what was the filename for the RPi patch? 2017-09-19 07:36:45 danieli: linux-4.1.y-rpi-20170501.patch 2017-09-19 07:36:57 thank you 2017-09-19 07:37:00 also, good morning 2017-09-19 07:37:35 good morning :) 2017-09-19 07:46:24 if nobody has it somewhere, it's lost 2017-09-19 07:46:54 if it wasn't written by someone *for* AL, it might be possible to figure the source out 2017-09-19 07:56:47 danieli, its a specific alpine patch made by shiz, our distfiles got erased so that's the reason its lost. 2017-09-19 07:57:19 and shiz is on holiday,. he is probably the only person to have a copy. 2017-09-19 07:57:58 ah i see.. hmm. 2017-09-19 07:58:07 clandmeter: what did the patch do? 2017-09-19 07:58:33 i think its a diff between vanilla and rpi kernel. 2017-09-19 07:59:31 # rpi patches are git diff against latest release tag from: 2017-09-19 07:59:33 yep 2017-09-19 09:17:08 ncopa: re bug #7855 - whenever perl is updated, urxvt will need a rebuild too 2017-09-19 09:24:05 i realized that 2017-09-19 09:24:10 dunno how to detect it though 2017-09-19 09:24:22 i mean, i wonder why i didnt notice that when updating perl 2017-09-19 09:24:25 me neither l0l 2017-09-19 09:24:30 oh that 2017-09-19 09:24:44 well, it's fixed, but will need extra care in the future when updating perl 2017-09-19 09:37:11 Alpine needs something like revdep-rebuild >.> 2017-09-19 09:58:19 ncopa: i removed one of your patches on mtr, it's in patchwork 2017-09-19 10:23:47 foxkit: i do something like: apk search -r -q --origin so:libperl.so | sh ../rebuild-since $githash_of_perlupdate^ "rebuild against perl 5.26" 2017-09-19 11:04:40 foxkit: thanks for splitting your PRs into logic chunks 2017-09-19 11:43:26 the day Alpine gets revdep-rebuild, I'll be switching distros. oh dear god, no more Gentoo for me ;P 2017-09-19 12:06:25 foxkit: as far as i've seen, people prepend the repo to the package, like "main/git: commit message" 2017-09-19 12:06:35 also might be a good idea to watch #alpine-commits 2017-09-19 12:09:29 ncopa: https://up.shiz.me/priv/linux-4.1.y-rpi-20170501.patch 2017-09-19 12:10:04 yessss 2017-09-19 12:43:34 Shiz: thank you very much 2017-09-19 12:50:08 i cant install gst-libav no more in edge. i get unsatisfiable constraints about libXrender.so.1 and libXrender.so.1 2017-09-19 12:50:14 anything i can do about it? 2017-09-19 12:51:14 <_ikke_> is libxrender installed as a dependency? 2017-09-19 12:51:31 i was just about to suggest the libxrender package 2017-09-19 12:51:46 just had to check - and correctly - it provides libxrender.so.1[.3.0] 2017-09-19 12:51:52 <_ikke_> https://pkgs.alpinelinux.org/contents?file=libXrender.so.1&path=&name=&branch=&repo=&arch= 2017-09-19 12:52:18 yes i know but i cant find it on the mirror 2017-09-19 12:52:20 maybe thats the problem 2017-09-19 12:52:40 apk search libxrender doesnt find anything 2017-09-19 12:52:56 <_ikke_> Right, I don't see it on dl-4 either 2017-09-19 12:53:02 which mirror? 2017-09-19 12:53:04 that's strange 2017-09-19 12:53:11 <_ikke_> neither on rsync 2017-09-19 12:53:14 wat 2017-09-19 12:54:37 on the http://nl.alpinelinux.org/alpine neither 2017-09-19 12:55:00 <_ikke_> if it's not on rsync, it won't be on other mirrors either 2017-09-19 12:55:09 ^ 2017-09-19 12:59:58 how this could happen? ;-) 2017-09-19 13:00:22 <_ikke_> xsteadfastx: not sure, clandmeter or ncopa should be able to check 2017-09-19 13:01:08 hey, what happened 2017-09-19 13:01:28 <_ikke_> libxrender seems to be missing from the mirrors 2017-09-19 13:01:32 including rsync 2017-09-19 13:01:46 check the buildserver, foxkit updated the apkbuild for that package recently 2017-09-19 13:01:47 <_ikke_> in edge at least 2017-09-19 13:03:51 it got built though: http://build.alpinelinux.org/buildlogs/build-edge-x86_64/main/libxrender/libxrender-0.9.10-r2.log 2017-09-19 13:04:04 strange 2017-09-19 13:30:04 i think it happens because it moves on and build next repo (community) if main fails 2017-09-19 13:30:12 then it will rsync all of it 2017-09-19 13:30:17 or something like that 2017-09-19 13:30:57 ok... so i just have to wait 2017-09-19 13:31:10 i think unbreaking the builder should fix it 2017-09-19 13:31:18 i suppose i just disable tests in wayland again 2017-09-19 13:31:24 if only i could unfuck the builder 2017-09-19 14:26:33 ncopa: what is the `MIRRORS.txt` on the rsync root used for? 2017-09-19 14:26:55 s/on the rsync root/file on the rsync root/ 2017-09-19 14:27:12 not sure if its used for anything nowdays 2017-09-19 14:27:36 okay, which mirror lists are effectively in use? 2017-09-19 14:27:59 in use by what? 2017-09-19 14:28:14 in general - setup-apkrepos, other things 2017-09-19 14:28:55 setup-apkrepos uses the mirror list from alpine-mirrors package 2017-09-19 14:29:31 i think alpinelinux.org/downloads uses some mirror list too 2017-09-19 14:29:36 but i dont remember where 2017-09-19 14:30:07 we used to have a php service on wiki.a.o that would use a mirror list and give you a random mirror redirect 2017-09-19 14:30:27 yeah, i wrote a python thing to do just that 2017-09-19 14:30:32 infra/mirrors 2017-09-19 14:31:02 and there's a https://wiki.alpinelinux.org/cgi-bin/dl.cgi - whatever it does, seems to redirect 2017-09-19 14:34:20 that one was it 2017-09-19 14:37:44 Shiz: do you have grsec-4.1.39-3.1-201509201149-alpine.patch 2017-09-19 14:38:04 https://up.shiz.me/priv/grsec-4.1.39-3.1-201509201149-alpine.patch 2017-09-19 14:38:59 thank you 2017-09-19 14:43:40 ncopa: https://github.com/alpinelinux/aports/pull/2284 can you look at the checkbashisms part, its currently totally broken in master. 2017-09-19 14:44:48 HRIo: ok i'll look at it now 2017-09-19 14:44:49 thanks 2017-09-19 14:45:02 thanks 2017-09-19 14:46:21 HRIo: virt-what still fails 2017-09-19 14:46:35 yes, it pulls the unfixed 2017-09-19 14:47:11 checkbashisms*: Create checkbashisms-2.0.0.2-r4.apk.... and then (7/8) Installing checkbashisms (2.0.0.2-r3) when building virt-what. 2017-09-19 14:47:38 seems to be the way travis works? 2017-09-19 14:47:49 or rather our travis scripts 2017-09-19 14:48:43 when I build checkbashisms and install that first, virt-what is OK to build (check phase) 2017-09-19 14:49:22 ok it was my repositories that didnt pull in main 2017-09-19 14:50:03 pushed 2017-09-19 14:50:04 thanks 2017-09-19 14:50:38 and thank you! ;-) 2017-09-19 14:53:14 a pity that more pkgs that does not depend on bash, is not using checkbashisms, in check(). Currenly its only virt-what :-) 2017-09-19 14:59:29 looks like ppc64le didnt work: http://build.alpinelinux.org/buildlogs/build-edge-ppc64le/testing/virt-what/virt-what-1.18-r1.log 2017-09-19 14:59:55 running checkbashisms in check was a good idea 2017-09-19 15:07:28 mm, will try to improve my other packages as well 2017-09-19 15:09:02 ppc64le: (7/8) Installing checkbashisms (2.0.0.2-r4) but still the perl error... very strange indeed, maybe more lines in checkbashims that need fixing then 2017-09-19 15:11:04 ncopa: I will have a look in checkbashisms again 2017-09-19 15:20:14 ncopa: https://git.alpinelinux.org/cgit/aports/commit/?id=366c48a4e2b2508b8a886efec0ef8efda508f139 2017-09-19 15:20:25 we have two -4 ? 2017-09-19 15:22:05 one from A. Wilcox and one from me... 2017-09-19 15:22:42 foxkit: ^ 2017-09-19 15:23:09 there is not functional diff in https://git.alpinelinux.org/cgit/aports/commit/?id=366c48a4e2b2508b8a886efec0ef8efda508f139 but version is bumped 2017-09-19 15:24:58 ncopa: virt-what fails on ppc64le as checkbashisms is not rebuilt with the fix yet. 2017-09-19 15:25:07 http://build.alpinelinux.org/buildlogs/build-edge-ppc64le/main/checkbashisms/ 2017-09-19 15:26:48 ah, that explains 2017-09-19 15:28:12 ncopa: sent a patch regarding main/alpine-mirrors, please apply 2017-09-19 15:28:32 need to run now, will merge it tomorrow (or later tonight) 2017-09-19 15:28:39 no worries 2017-09-19 18:46:30 for diskless we have an update-kernel script and a similar one if using uboot (update-u-boot), but could not find a script for upgrading xen hypervisor when running diskless. Somebody here that knows? 2017-09-19 22:15:53 foxkit, why did you set interpreter-version=38? 2017-09-19 22:39:20 danieli: I would be interested in an alpine build for Chip. Planned to do it myself, but never found the time. 2017-09-19 22:39:40 tdtrask: i'm gonna do some tricks, then test it out when my chip arrives 2017-09-19 22:39:51 i'm gonna have a go at an alpine mips build 2017-09-19 22:40:13 ACTION has had a chip for months 2017-09-19 22:40:25 done nothing but boot it with official image :( 2017-09-19 22:40:33 i just recently cared to actually order one 2017-09-19 22:40:38 tdtrask: you can use the alpine rootfs :) 2017-09-19 22:42:39 clandmeter: ? in what package? that doesn't sound familiar 2017-09-19 22:42:50 freetype 2017-09-19 22:43:58 foxkit, 3651e41726f8baf204c9d5cef25418d8baab0629 2017-09-19 22:45:46 ooo very sorry. that should not have been submitted. I was a little.. overwhelmed by all the diffs, I must have missed that one ._. 2017-09-19 22:46:22 why did you choose it over mode 40? 2017-09-19 22:46:46 i mean even by accident 2017-09-19 22:47:56 v40 is cleartype and v38 is infinality 2017-09-19 22:48:46 back on the gentoo days I did an irl experiment with people looking at different things with the different renderers and the study showed that everyone that I gathered (granted, only five people) preferred v38 over v40, they found it easier to read and more legible 2017-09-19 22:49:10 did you read https://www.freetype.org/freetype2/docs/subpixel-hinting.html ? 2017-09-19 22:49:42 im currently not using alpine on the desktop so i have personal preference. 2017-09-19 22:50:04 but what i read here it makes me favour 40 instead of 38 2017-09-19 22:55:41 err, i have "no" personal preference... 2017-09-19 22:59:33 I guess I can run another experiment :) 2017-09-19 23:01:25 ps i got an email asking if i could revert it. 2017-09-19 23:03:34 foxkit: minor slip at https://git.alpinelinux.org/cgit/aports/tree/main/xorg-server/APKBUILD#n139 2017-09-19 23:03:35 yeah, you can go ahead and revert it. the only change is the more explicit configure (which can catch some errors like if the depends are updated and one is removed). 2017-09-19 23:04:04 ncopa, any reason zfs is behind? and do you want to take over maintainership? 2017-09-19 23:04:41 dsabogal: wow, this is unacceptable. the sheer amount of commits and merges I had to make caused the quality to slip I guess. I feel real bad now. 2017-09-19 23:04:52 there were over 250 commits I had to pour over... :/ 2017-09-19 23:05:03 lol 2017-09-19 23:05:07 well, really, 400, but I filtered it down to 250 2017-09-19 23:05:18 taking over the world are we? :) 2017-09-19 23:05:30 not intentionally! 2017-09-19 23:05:37 alpine is the new skynet 2017-09-19 23:06:38 <^7heo> foxkit: just put some precommit hook matching patterns 2017-09-19 23:07:10 the X packages just seemed especially neglected, which is understandable because there is not so many people using alpine on desktop. so that's why I focused a lot of effort on making them all good and correct. and obviously I failed to vet the merge properly :/ 2017-09-19 23:07:20 <^7heo> that way you don't have to stress 2017-09-19 23:07:34 ^7heo: that's a good idea! 2017-09-19 23:07:39 thanks 2017-09-19 23:07:56 <^7heo> you're welcome 2017-09-19 23:08:44 <^7heo> it helps for any code too 2017-09-19 23:08:48 we should spank the commiter as wel. 2017-09-19 23:08:55 now now 2017-09-19 23:09:02 don't mix work and pleasure 2017-09-19 23:09:27 haha 2017-09-19 23:09:34 <^7heo> right. then it could become fun 2017-09-19 23:09:53 <^7heo> and we sure do not want that 2017-09-19 23:10:03 well, else it wouldn't be work, DUH 2017-09-19 23:10:16 <^7heo> precisely 2017-09-19 23:12:43 bed time. gn ppl. 2017-09-19 23:14:01 i think i'm starting to get sick, so i'll go to sleep as well 2017-09-19 23:14:02 clandmeter: night 2017-09-19 23:14:21 aw, feel better soon 2017-09-19 23:15:20 thanks <3 2017-09-20 02:29:18 going to do one-package-per-branch PRs from this point on 2017-09-20 02:29:28 smaller but more frequent 2017-09-20 02:29:33 don't ever want to have 300 in queue ever again 2017-09-20 03:01:51 clandmeter: clamav does not build now because libmilter is in community/ and clamav is in main/. do we want to move libmilter to main or remove it? 2017-09-20 07:05:05 morning 2017-09-20 07:06:41 foxkit: i appreciate that you follow up on x11 stuff. minor slips happens, and its not worse than we can fix them 2017-09-20 07:09:19 and i like that you grouped the pull requests 2017-09-20 07:16:17 good morning climbers 2017-09-20 08:39:22 foxkit, i moved it to main. 2017-09-20 08:40:22 morning 2017-09-20 09:18:10 i'm patching secfixes for openjpeg 2017-09-20 09:18:30 CV1-2017-14039 patch applies but return error in compilation 2017-09-20 09:18:30 https://github.com/uclouvain/openjpeg/commit/c535531f03369623b9b833ef41952c62257b507e 2017-09-20 09:19:20 https://dpaste.de/51ur 2017-09-20 09:28:45 god damn that dpaste font is annoying 2017-09-20 09:29:21 :) 2017-09-20 09:31:04 how true 2017-09-20 09:32:01 fcolista, that patch is probably only valid for master 2017-09-20 09:32:35 i'm guessing so too 2017-09-20 09:32:38 needs to be backported 2017-09-20 09:32:39 i should diff master and 2.2.0 stable 2017-09-20 09:32:42 ? 2017-09-20 09:32:51 i think the previous commit could fix it if it applies 2017-09-20 09:32:55 how can we get openjpeg patched ? 2017-09-20 09:32:56 i think you have to read the code and see that the patch does 2017-09-20 09:33:19 as in you may need to understand the code 2017-09-20 09:33:33 alternatively, you could see if some other distro has a backport of the patch 2017-09-20 09:33:36 debian for example 2017-09-20 09:33:59 https://security-tracker.debian.org/tracker/CVE-2017-14039 2017-09-20 09:34:14 status vulnerable, unfixed 2017-09-20 09:34:32 right 2017-09-20 09:46:38 so youre not the only one having this problem :) 2017-09-20 09:46:45 you could also ask upstream for help 2017-09-20 09:54:52 ncopa, what is the reason zfs is behind? 2017-09-20 09:55:07 nobody has taken time to work on it 2017-09-20 09:55:32 ok 2017-09-20 09:55:38 im maintainer so i guess i should take the blame. 2017-09-20 09:55:53 i guess so 🙃 2017-09-20 09:55:55 most core devs are maintainer for a ton of packages 2017-09-20 09:56:19 ncopa, but you are a user ;-) 2017-09-20 09:56:29 i know... 2017-09-20 09:56:33 want me to check out zfs version bumping or libjpeg? 2017-09-20 09:56:47 tbh, i didnt notice it was outdated 2017-09-20 09:56:51 the fix in git is just 24 additions 2017-09-20 09:57:03 i just wanted to know if there was a specific reason. 2017-09-20 09:57:08 danieli: i think zfs may be tricky 2017-09-20 09:57:12 oh 2017-09-20 09:57:20 0.6.5.9-r1 / 0.7.1 2017-09-20 09:57:22 ncopa: i think so too 2017-09-20 09:57:33 it always has :) 2017-09-20 09:57:41 or may need some special attention due to ROP from PaX 2017-09-20 09:57:43 but not as bad as chromium :) 2017-09-20 09:57:50 true to both 2017-09-20 09:58:06 chromium can be a real pain in the ass, but it's massive lol 2017-09-20 10:09:21 open PRs on github reduced to < 100 2017-09-20 10:09:27 i need to go out for a while 2017-09-20 10:09:36 later 2017-09-20 10:11:02 nice, all our patches are in 0.7.0 :) 2017-09-20 10:11:23 ncopa, good idea to ask upstream 2017-09-20 10:11:28 i've done that. 2017-09-20 10:11:33 clandmeter: zfs merged them in? 2017-09-20 10:11:46 yes 2017-09-20 10:11:49 neat! 2017-09-20 10:11:56 i think they should release 2.2.1 (imho) 2017-09-20 10:12:18 wat 2017-09-20 10:12:28 we had 9 patches for alpine/musl 2017-09-20 10:13:09 anyone created an uefi iso to boot off recent apple laptops? 2017-09-20 10:13:24 i haven't, but you should be able to do that yourself 2017-09-20 10:13:39 people have had success using refit and apple's own boot manager 2017-09-20 10:14:54 i made one long time ago 2017-09-20 10:15:09 but not specific for mac 2017-09-20 10:15:11 in a galaxy far, far away? :) 2017-09-20 10:15:30 when playing with aarch64 2017-09-20 10:15:50 bah I need a dualboot betwenn alpine and win10 2017-09-20 10:15:56 two different hdds 2017-09-20 10:16:04 chainloading next 2017-09-20 10:16:08 shouldn't be too hard 2017-09-20 10:16:10 indeed 2017-09-20 10:16:21 i run windows in VMs l0l 2017-09-20 10:16:28 me too 2017-09-20 10:16:43 but i have a laptop with win10 too 2017-09-20 10:16:45 for other reasons 2017-09-20 10:16:53 its a mac? 2017-09-20 10:16:56 no 2017-09-20 10:17:01 dell 2017-09-20 10:17:10 why not just use the bios to select boot device? 2017-09-20 10:17:13 i don't do much work with windows anymore, so i don't need a windows-only machine. i just keep my windows stuff in VMs for now 2017-09-20 10:17:20 that's what I do now clandmeter 2017-09-20 10:17:30 F12 and choose the hdd :) 2017-09-20 10:19:19 crap, that pax-rap patch doesnt seem too simple... 2017-09-20 10:20:22 rap rip rop 2017-09-20 10:20:34 i should ask snoop to help me out :) 2017-09-20 10:20:46 paxizzle fo shizzle 2017-09-20 10:21:30 yes, it's not too hard to do unless you are on a Mac trying to create a bootable AL usb stick. I have no idea how to use a Mac 2017-09-20 10:22:31 i've done some development on macs, fairly low level.. but so long ago a lot is obsolete now :/ 2017-09-20 10:22:40 if i had a mac i'd help :) 2017-09-20 10:24:20 I found a terminal, but all the util-linux stuff works differently. I can't do frisk -l for ex 2017-09-20 10:24:54 gummiboot doesnt work anymore? 2017-09-20 10:26:01 how the hell am I going to get gummiboot onto a usb stick on a Mac :) I haven't been managed to format the stick yet. 2017-09-20 10:26:35 i dont own a mac so i cant help you. i remember ncopa made an usb stick with gummiboot to install alpine. 2017-09-20 10:27:17 yeah, there is a page here that shows where to put it https://wiki.alpinelinux.org/wiki/Create_UEFI_boot_USB 2017-09-20 10:28:05 ncopa, did you upstream that rap stuff for zfs? i cannot find any reference to it. 2017-09-20 10:28:14 ncopa uses the apple boot manager to boot alpine 2017-09-20 10:29:21 but you still need uefi media to install right? 2017-09-20 10:29:25 i see two issues on zfsonlinux/zfs that are relevant to rap 2017-09-20 10:29:38 https://github.com/zfsonlinux/zfs/issues/5911 https://github.com/zfsonlinux/zfs/issues/5436 haven't read through them, ctrl+f for "rap" 2017-09-20 10:29:43 "apple boot manager", as if that is supposed to mean something ;) 2017-09-20 10:30:29 the one where you hold option on startup 2017-09-20 10:30:48 i think that's the one 2017-09-20 10:30:53 Startup Manager yes 2017-09-20 10:34:22 ok, I have everything I need except no idea how to set boot flag in osx 2017-09-20 10:38:05 hmm, what the full path of the boot loader files? 2017-09-20 10:38:36 https://wiki.alpinelinux.org/wiki/Create_UEFI_boot_USB#Create_configuration_files_for_boot_loader <- ? 2017-09-20 10:42:36 ok, trying /boot 2017-09-20 12:29:08 let's see if zfs builds 2017-09-20 12:30:15 clandmeter: it built just like that 2017-09-20 12:30:25 vanilla anyway 2017-09-20 12:30:30 yes it should 2017-09-20 12:30:46 and grsec also 2017-09-20 12:31:12 but on grsec it will not work i believe. 2017-09-20 12:31:40 the spl patch is still ok, just the zfs patch needs updating. 2017-09-20 12:31:54 yeah, gathered as much 2017-09-20 12:42:26 hm, somehow terminus-font broke with the latest X11 library upgrades 2017-09-20 12:54:03 screenshots? 2017-09-20 12:54:12 or does it just fallback? 2017-09-20 12:56:14 not sure, seems to fallback to a different font 2017-09-20 12:56:21 here is a screenshot of urxvt http://abload.de/img/terminus-fontf3sz1.png 2017-09-20 12:56:36 clandmeter or foxkit: any idea? ^ 2017-09-20 12:58:05 to reproduce this add `URxvt.font: xft:Terminus:pixelsize=16` to your .Xresources 2017-09-20 12:58:23 font also looks broken in dmenu and dwm, might be an libxft issue? 2017-09-20 12:58:25 nmeum, you are on edge right? 2017-09-20 12:58:28 yep 2017-09-20 12:58:40 X packages were indeed recently updated 2017-09-20 12:59:14 i bumped freetype, not sure its related. 2017-09-20 12:59:38 clandmeter: bumped it to what version? 2017-09-20 12:59:41 2.7.x? 2017-09-20 12:59:47 patch bump 2017-09-20 13:00:03 i think its now 2.8.1 2017-09-20 13:00:10 nmeum: see if "xos4 Terminus" works for font 2017-09-20 13:00:18 I downgraded freetype to freetype-2.8-r1 didn't seem to fix it 2017-09-20 13:00:24 e.g. xft:xos4 Terminus:pixelsize=16 2017-09-20 13:00:31 i.e.* damn latin 2017-09-20 13:00:48 danieli: that fixes it 2017-09-20 13:00:55 https://lists.nongnu.org/archive/html/freetype/2017-01/msg00020.html 2017-09-20 13:01:05 you can use fontconfig to define an alias for it 2017-09-20 13:01:20 had a feeling that was it :) 2017-09-20 13:01:53 oh the short code for font issue? 2017-09-20 13:02:06 the freetype devs renaimg stuff 2017-09-20 13:02:15 i removed 2 patches 2017-09-20 13:02:18 maybe we should add that alias to /etc/fonts/conf.avail/75-yes-terminus.conf 2017-09-20 13:02:25 I will read that ML thread in a second 2017-09-20 13:02:40 seems as they were upstreamed 2017-09-20 13:03:00 correct 2017-09-20 13:03:02 i remember the fontcache needs updating. 2017-09-20 13:21:14 http://bugs.alpinelinux.org/issues/7446 2017-09-20 13:28:34 clandmeter: aal alright 2017-09-20 13:29:29 aah* 2017-09-20 13:29:42 but i dont think i changed anything related to PCF names. 2017-09-20 13:29:58 it was changed upstream 2017-09-20 13:30:01 in freetype 2017-09-20 13:30:26 yes but thats 2.7 2017-09-20 13:30:31 we are on 2.8 iirc 2017-09-20 13:30:36 did they revert it? 2017-09-20 13:30:39 i haven't kept up to date 2017-09-20 13:30:41 and yes, 2.8 2017-09-20 13:31:21 and we have a local patch for PCF 2017-09-20 13:31:31 but i have no way to test. 2017-09-20 13:31:35 ah okay 2017-09-20 13:31:39 uh, how come no way to test? 2017-09-20 13:31:50 im on windows 2017-09-20 13:31:54 oh right 2017-09-20 13:32:24 and i dont have time atm to setup a new env to test. 2017-09-20 13:32:34 i'm on debian 9.1 2017-09-20 14:00:10 ncopa and @jirutka: suggested changes made to https://github.com/alpinelinux/aports/pull/2314. Thx for the review! 2017-09-20 14:01:05 clandmeter/danieli: i can test 2017-09-20 14:01:09 what do you need me to test? 2017-09-20 14:04:04 chambbj: thanks! 2017-09-20 14:04:19 chambbj: do you think you can squash the commits? 2017-09-20 14:04:24 or do you want me to do it for you 2017-09-20 14:04:58 i mean, the fixes into the corresponding "testing/: blah" commit 2017-09-20 14:07:03 clandmeter danieli: re: freetype what would be you preferred way to fix this? because this will likely break the setup of a few people in the future 2017-09-20 14:07:59 i dont have clear understanding of the problem at hand 2017-09-20 14:09:23 freetype fucked around with short/long font names, after upgrading freetype "Terminus" became "xos4 Terminus" 2017-09-20 14:09:32 freetype upstream changed some font names 2017-09-20 14:09:34 as far as I understand, they reverted this in later versions? 2017-09-20 14:09:37 or? 2017-09-20 14:09:45 how does other distros handle it? 2017-09-20 14:09:46 I don't think so 2017-09-20 14:09:55 ncopa: it breaks users' configs 2017-09-20 14:10:07 the best fix feels like adding aliases to fontconfig 2017-09-20 14:10:10 ncopa: no problem - i'll squash 2017-09-20 14:10:27 chambbj: i already did 2017-09-20 14:10:31 clandmeter: does `fc-cache -f` fix it? 2017-09-20 14:10:48 >i remember the fontcache needs updating. 2017-09-20 14:10:52 danieli: nope 2017-09-20 14:11:01 chambbj: does this look ok? http://tpaste.us/Ynjg 2017-09-20 14:11:08 i see. so it simply breaks users' configs 2017-09-20 14:11:17 yep 2017-09-20 14:11:47 https://wiki.archlinux.org/index.php/font_configuration#FreeType_Breaking_Bitmap_Fonts 2017-09-20 14:12:12 lunch. brb 2017-09-20 14:12:25 ncopa: looks great 2017-09-20 14:12:26 tyt 2017-09-20 14:16:50 ncopa: apparently, freetype2 already reverted to short names again 2017-09-20 14:16:53 https://git.savannah.gnu.org/cgit/freetype/freetype2.git/commit/?id=7ccca6aec167c2c30c569765ece808f0eee023a6 2017-09-20 14:18:29 I know I'm being lazy, but suppose you wanted to repackage the kernel. would just having the kernel binary and its respective /lib/modules/ entry be enough to set up a working system? 2017-09-20 14:19:26 danieli: in that case we should just backport this commit imho 2017-09-20 14:19:35 still working on that secure booting environment and it kind of requires repackaging the kernel 2017-09-20 14:20:21 there's this property, pcf:no-long-family-names=1/0 2017-09-20 14:22:36 used as, for example, `FREETYPE_PROPERTIES="pcf:no-long-family-names=1" fc-cache -rvf` 2017-09-20 14:22:53 upstream apparently reverted it because it was 'too radical' anyway 2017-09-20 14:31:34 we already have a similar patch https://git.alpinelinux.org/cgit/aports/tree/main/freetype/0004-Enable-long-PCF-family-names.patch 2017-09-20 14:31:39 which didnt change 2017-09-20 14:40:58 so we just stick to the long names for now? 2017-09-20 14:41:15 try `FREETYPE_PROPERTIES="pcf:no-long-family-names=0" fc-cache -rvf` 2017-09-20 14:41:22 wait wait 2017-09-20 14:41:25 "no" - use =1 2017-09-20 14:42:30 yeah that works, the question is just what the default behaviour should be 2017-09-20 14:43:20 imo the original behavior that doesn't break shit 2017-09-20 15:07:16 i think the question should be, why did you bump into this issue? 2017-09-20 15:08:12 this patch is already 2 months in aports 2017-09-20 15:25:34 what are the difference between long and short names? 2017-09-20 15:25:46 i mean, the benefit with long names 2017-09-20 15:26:07 why was https://git.alpinelinux.org/cgit/aports/tree/main/freetype/0004-Enable-long-PCF-family-names.patch added in the first place 2017-09-20 17:16:30 fcolista: looks like backporting the fix is gonna take a bit of effort, we should probably wait for 2.2.1 2017-09-20 17:16:36 re: openjpeg 2017-09-20 17:26:44 https://svn.apache.org/viewvc/httpd/httpd/branches/2.4.x/server/core.c?r1=1805223&r2=1807754&pathrev=1807754&view=patch 2017-09-20 17:26:48 optionsbleed patch 2017-09-20 17:44:09 ncopa: an apache2 optionsbleed patch is in patchwork 2017-09-20 21:05:55 samba 4.6.8 requires ldb-dev version (>=1.1.29 <=1.1.99 !=1.1.30 !=1.1.31) 2017-09-20 21:07:17 looks like i'll have to update ldb-dev before samba 2017-09-20 21:08:58 wait, damn it 2017-09-20 21:09:44 _ikke_: how can i list available versions of a package on the repo? 2017-09-20 21:09:59 apk policy only lists 1.1.31 which samba won't accept 2017-09-20 21:10:08 the new ldb version is 1.2.2 2017-09-20 21:11:00 <_ikke_> danieli: apk version? 2017-09-20 21:15:52 _ikke_: i think i'm doing it wrong, i grepped for "ldb" and found nothing 2017-09-20 21:33:30 <_ikke_> danieli: did you submit the update to 3.4.1 for zabbix already? 2017-09-20 21:43:00 _ikke_: waiting in patchwork, no push access 2017-09-20 21:43:21 <_ikke_> danieli: right 2017-09-20 21:43:53 there, on my laptop 2017-09-20 21:44:37 <_ikke_> I have a patch that should be applied on top of yourss 2017-09-20 21:45:25 _ikke_: all right, well, mine isn't applied yet 2017-09-20 21:45:47 i sent it to alpine-aports, it's waiting in patchwork 2017-09-20 21:48:36 <_ikke_> https://patchwork.alpinelinux.org/patch/3561/ 2017-09-20 21:50:27 yep 2017-09-20 21:50:50 looks like something along my chain added newlines, i should check that up and keep patches small 2017-09-20 21:51:22 <_ikke_> I added missing dependencies 2017-09-20 21:51:34 wonderful 2017-09-21 06:35:02 there 2017-09-21 06:35:16 alpine's httpd is now not vulnerable to optionsbleed 2017-09-21 07:08:51 apache? 2017-09-21 07:08:53 morning 2017-09-21 07:17:36 yes apache 2017-09-21 07:17:39 good morning ncopa :) 2017-09-21 07:18:03 i backported the optionsbleed patch 2017-09-21 07:45:51 thanks 2017-09-21 07:47:14 good morning 2017-09-21 07:47:19 people still use apache? 2017-09-21 07:47:30 unfortunately :) 2017-09-21 07:48:05 :) 2017-09-21 08:10:52 >>> apache2*: Stripping binaries 2017-09-21 08:10:52 getfattr: ./usr/lib/apache2/stBonmop: No such file or directory 2017-09-21 08:10:52 strip: './usr/lib/apache2/stBonmop': No such file 2017-09-21 08:10:58 from the build logs of apache2 on aarch64 2017-09-21 09:00:33 php7 fails due to tests failing 2017-09-21 11:37:02 systemd based lxc containers now work under OpenRC https://github.com/alpinelinux/aports/pull/2345 2017-09-21 11:43:25 cool 2017-09-21 11:46:45 BitL0G1c: you tested it, right 2017-09-21 11:46:46 ? 2017-09-21 11:46:55 if so, kick-ass 2017-09-21 11:47:04 yes I use in manjaro openrc 2017-09-21 11:47:09 neat! 2017-09-21 11:47:11 ncopa: ^ :) 2017-09-21 11:47:31 I run centos & ubuntu lxc containers as well as alpine 2017-09-21 11:51:14 manjaro-openrc has morphed into artix linux (still arch with openrc) 2017-09-21 11:51:34 long ago? i haven't kept in the loop 2017-09-21 11:51:42 i'mma tell a manjaro using friend of mine 2017-09-21 11:52:38 I used these instructions danieli https://artix-linux.github.io/migrate/ 2017-09-21 11:59:01 i started work on compiling alpine for mips 2017-09-21 13:32:30 ncopa: just curious, why isn't alpine packaged for mips? 2017-09-21 13:32:52 danieli: nobody has made the effort 2017-09-21 13:32:57 i see 2017-09-21 13:33:00 and we dont have the hardware 2017-09-21 13:33:03 mips can make it very interesting for embedded devices 2017-09-21 13:33:04 gotcha 2017-09-21 13:56:03 ncopa: if i can get us some hardware through contacts, do you think people would be willing to put the work into it? 2017-09-21 13:56:30 i personally have not the capacity to do the work 2017-09-21 13:57:09 that i know, just trying to probe for interest 2017-09-21 14:03:46 clandmeter: how hard would it be to set up a mips builder? 2017-09-21 14:22:34 danieli, not sure. it depends what hard means. you will need the hardware (capable to keep up with other infra) boostrap toolchain (fabled did most of others archs) build world (takes a lot of time but is not that hard). And it also introduces more maintenance which kind of sucks. 2017-09-21 14:22:59 yeah, maintenance is the big deal here i'd say 2017-09-21 14:23:13 there are people who could help with toolchains 2017-09-21 14:24:22 sure, but its probably the smallest part costing time (if you have the knowledge) 2017-09-21 14:24:45 the question is, the ppl who want it, if they will maintain it. 2017-09-21 14:25:34 i remember kaniini talked about it. but till now no progress i believe. 2017-09-21 14:25:51 talked about mips? 2017-09-21 14:26:06 i think he is also interested in it. 2017-09-21 14:31:08 it's used in lots of routers/gateways/modems 2017-09-21 14:31:13 i think there is probably a quite large group that is interested in replacing their router fw with alpine. but im not sure that group is as eager to support the development :) 2017-09-21 14:31:25 yeah, that's a good thought 2017-09-21 14:34:41 you should also consider that a "regular" alpine install will most probably not fit on most routers without external storage. 2017-09-21 14:35:27 yeah of course, would need a couple builds 2017-09-21 14:35:42 a lot of network devices only have 8MB or so flash storage 2017-09-21 14:35:53 yes 2017-09-21 14:35:58 some even less 2017-09-21 14:36:05 indeed 2017-09-21 14:36:12 how will you solve that? 2017-09-21 14:37:27 the best solution would probably be to trim out unnecessary features and do board-specific builds 2017-09-21 14:37:37 ACTION glances at openwrt 2017-09-21 14:37:41 dont get me wrong, i would also love it. but its a lot of work to create something similar like openwrt with alpine's tools. 2017-09-21 14:37:49 yeah, that is true 2017-09-21 14:38:33 i would love to see apk on openwrt 2017-09-21 14:39:14 luci is cool too 2017-09-21 14:39:32 i think i need to have a chat with kaniini and synchronize thoughts 2017-09-21 14:41:27 on the other hand, you could just buy an apu2 and have real hardware with similar design. 2017-09-21 14:45:06 but it removes the fun of hacking :) 2017-09-21 14:45:20 that is true 2017-09-21 15:38:57 hello guys, i want to start as packager, any help? 2017-09-21 15:40:04 downloaded aports, tried with rarcrack package from AUR but it's buggy :D 2017-09-21 15:40:43 <_ikke_> What is the problem? 2017-09-21 15:42:30 in AUR package is an fix for ... look https://aur.archlinux.org/packages/rarcrack/ 2017-09-21 15:43:04 there is a patch to fix one of 3 supported filetypes, but one another need too be fixed :D 2017-09-21 15:43:14 uzerus: why don't you upstream it? 2017-09-21 15:43:46 dead package? 2017-09-21 15:44:09 danieli: cos it's over my skills, yea a little deadly package 2017-09-21 15:44:27 it needs to be rewriten 2017-09-21 15:44:50 well, patches are applied to be.. patches. they're meant to fix something 2017-09-21 15:45:01 if you want to change functionality of a package, you should do that upstream 2017-09-21 15:45:32 danieli: just learning scripting 2017-09-21 15:45:55 so.. what exactly do you want to do, in context of alpine? how is the rarcrack package you are building not working? 2017-09-21 15:48:48 forget about alpine, just the wiki is sometimes a little unclear to me, maybe cos of my too little skills 2017-09-21 15:49:04 the wiki is messy and is going to be reworked 2017-09-21 15:50:38 _ikke_: is it all right to package a `$pkgname-beta`? 2017-09-21 15:50:50 or how is that done? 2017-09-21 15:51:25 <_ikke_> Haven't seen that done before 2017-09-21 15:51:33 is it "bad"? 2017-09-21 15:53:04 <_ikke_> no clue 2017-09-21 15:53:09 <_ikke_> You could add it to testing first 2017-09-21 15:53:37 asking a theoretical question for now, basically providing two versions of a package as separate packages, a stable and an unstable 2017-09-21 15:53:43 just wondering how that would be done 2017-09-21 15:55:25 $pkgname and $pkgname-dev ? 2017-09-21 15:55:45 <_ikke_> -dev is something different 2017-09-21 15:55:53 or git, svn? 2017-09-21 15:56:32 or just $pkg-unstable 2017-09-21 15:58:13 i'm not talking about the naming scheme 2017-09-21 15:58:37 asking about the best way to handle different versions of a package - different name, different repo or something else 2017-09-21 16:00:13 danieli, $pkgname-$pkgver_bX-$pkgrel 2017-09-21 16:00:46 I see 2017-09-21 16:00:48 atleast this is what ive seen in aports 2017-09-21 16:23:43 i fixed kismet 2017-09-21 16:23:57 can i move it from unmaintained to testing? 2017-09-21 16:30:22 i am stuck, how to compile a package from source? configured APKBUILD unpacked and configured... how to compile pkg without install it? 2017-09-21 16:30:33 uzerus: `abuild -r` 2017-09-21 16:30:39 that'll package it 2017-09-21 16:34:04 make[2]: Entering directory '/home/test/stress/src/stress-1.0.4/src' 2017-09-21 16:34:09 test -z "/usr/local/bin" || .././install-sh -c -d "/usr/local/bin" 2017-09-21 16:34:14 make[2]: Nothing to be done for 'install-data-am'. 2017-09-21 16:34:19 /usr/bin/install -c 'stress' '/usr/local/bin/stress' 2017-09-21 16:34:24 install: can't create '/usr/local/bin/stress': Permission denied 2017-09-21 16:34:39 what is wrong with install.sh .... :/ 2017-09-21 16:34:58 looks like something in your apkbuild is wrong 2017-09-21 16:35:00 pastebin it 2017-09-21 16:35:15 its probably missing make DESTDIR=$pkgdir or similar 2017-09-21 16:35:36 ^ looks like it 2017-09-21 16:35:45 uzerus: it's trying to install it on your machine, not in the $pkgdir 2017-09-21 16:35:56 try `make DESTDIR="$pkgdir" install || return 1` 2017-09-21 16:39:57 kaniini: ping ding, tell me when you're around, we need to brain-sync about mips 2017-09-21 17:43:41 remind me, why was gdb not updated to 8? 2017-09-21 17:43:47 i'm making an issue on it in redmine 2017-09-21 17:54:59 can some1 build for me an package? its "stress" named 2017-09-21 17:55:59 https://jpst.it/158Rm <-- APKGBUILD /incomplete, probably need to do something with install="" , just want to learn something 2017-09-21 17:56:15 need it for https://aur.archlinux.org/packages/s-tui/ 2017-09-21 17:57:49 it contains src="[...]archive.org[...]" cos it's no longer maintained 2017-09-21 17:58:49 i'll check it out in a sec 2017-09-21 17:59:23 uzerus: link me the upstream project URL? 2017-09-21 18:00:00 http://people.seas.harvard.edu/~apw/stress/ 2017-09-21 18:00:07 okay, yeah, that one 2017-09-21 18:01:25 uzerus: did you try building the one in unmaintained? 2017-09-21 18:02:13 its just clone of this file, yeah i tried 2017-09-21 18:02:16 wait a sec 2017-09-21 18:02:58 uzerus: it built fine for me 2017-09-21 18:03:08 test -z "/usr/local/bin" || .././install-sh -c -d "/usr/local/bin" --- from abuild -R 2017-09-21 18:03:21 not -R 2017-09-21 18:03:23 just abuild -r 2017-09-21 18:03:40 >>> ERROR: stress*: package failed 2017-09-21 18:03:56 due to what? 2017-09-21 18:03:58 pastebin it 2017-09-21 18:04:31 2b879b3c5dcf:~$ cd aports/unmaintained/stress 2017-09-21 18:04:32 2b879b3c5dcf:~/aports/unmaintained/stress$ abuild -r 2017-09-21 18:04:34 i did this, and it works 2017-09-21 18:04:54 https://pastebin.com/m2Xkr4Xy 2017-09-21 18:05:22 ah, its from my own build 2017-09-21 18:05:30 w8 2017-09-21 18:05:43 just use the apkbuild from unmaintained/stress 2017-09-21 18:05:47 way easier 2017-09-21 18:06:16 yeah, works... 2017-09-21 18:07:21 how can i maintain it, need it as dependency 2017-09-21 18:07:39 in commnity repo in future, now testing 2017-09-21 18:10:50 just move this APKGBUILD to aports/testing/stress? 2017-09-21 18:14:55 danieli: in compiling time it asks for !check in $options , it's ok to leave that or define it in APKBUILD 2017-09-21 18:15:44 i'm a tad busy, will reply when i have time 2017-09-21 19:17:15 uzerus: if the package does not have tests, it should have !check 2017-09-21 19:18:13 tests for what? 2017-09-21 19:18:40 look at unmaintained/stress 2017-09-21 19:20:55 uzerus: you do not know if the package you plan to maintain has a testsuite to run or not ? 2017-09-21 19:24:35 kaniini: kk 2017-09-21 19:25:45 kaniini: I wonder what's is the process of introducing new packages to repo already abandoned by upstream, like in this case 2017-09-21 19:27:44 scadu: in few days i want to add new package https://aur.archlinux.org/packages/s-tui/ to alpine, "stress" is a dependency 2017-09-21 19:28:21 its a little usefull tool 2017-09-21 19:28:30 k goodnight 2017-09-21 19:35:44 scadu: if a maintainer wishes to maintain a package not maintained upstream, then he becomes responsible for any bugfixes himself 2017-09-21 19:44:03 kaniini: sounds like a lot of fun 2017-09-21 19:50:57 kaniini: thoughts on alpine on mips? 2017-09-21 20:46:48 danieli: i have some edgerouter devices which have quad-core octeon in them 2017-09-21 20:46:54 danieli: and upgraded memory (upgraded to 8gb) 2017-09-21 20:48:40 kaniini: yeah, i have a bunch of devices it'd be nice to run alpine mips on 2017-09-21 20:48:58 kaniini: what do you think of the suitability and sustainability of an arm build of alpine? 2017-09-21 20:49:20 we already ship arm 2017-09-21 20:49:23 do you mean mips ? 2017-09-21 20:49:24 lol 2017-09-21 20:49:40 uuuuh 2017-09-21 20:49:42 brainfart 2017-09-21 20:49:43 i meant mips 2017-09-21 20:55:12 i mean, it will add some maintenance overhead - and how would we deal with board-specific builds? not at all? 2017-09-21 22:57:35 hi 2017-09-21 22:59:59 anyone in here 2017-09-21 23:00:55 most people are EU-based, so you'll have better luck during EU daytime :) 2017-09-21 23:01:16 oh 2017-09-21 23:02:20 im from kurdistan :D 2017-09-21 23:02:29 2:02 AM 2017-09-21 23:03:13 go to sleep, come back tomorrow ;) 2017-09-21 23:04:17 why you are in EU and you are not sleep, i think you watch porn 2017-09-21 23:04:32 thats not good thing 2017-09-21 23:04:47 because that's obviously the only possible reason to be awake at night 2017-09-21 23:06:18 but if i will come tomorrow if question anything i can get the answer? 2017-09-21 23:06:24 sorry my english so bad 2017-09-21 23:07:14 if you have Alpine Linux-related questions, yes, it is likely that you will get answers tomorrow 2017-09-21 23:07:49 where are you from? :D 2017-09-21 23:08:06 I'm from France, it's 1 AM here, and I'll soon be asleep 2017-09-21 23:09:14 wow good country superpower counrty list :D 2017-09-21 23:09:55 but its bad country frace genocide algerian 2017-09-22 00:19:35 danieli: i think mips has DTBs now 2017-09-22 00:57:25 Hi! If the edge build servers didn't produce updated APKs for certain architectures (e.g. armhf) for recent apkbuild updates, and http://build.alpinelinux.org/ has shown the armhf edge build server on a single package for several hours now, who should be alerted? 2017-09-22 01:43:13 craftyguy: i took care of it 2017-09-22 01:43:39 soon the stuck builder problem won't be 2017-09-22 03:40:20 danieli: I already started, but I had some issues 2017-09-22 03:40:43 danieli: http://lists.alpinelinux.org/alpine-devel/5813.html 2017-09-22 03:42:45 danieli: our use cases are different (I would personally like to provide a way forward for nekochanites and the like, which is an SGI community - the O2 and Octane are still capable computers, but Irix is quite obsolscent now unfortunately) 2017-09-22 03:43:05 but the CPU arch is the same, and combined we can probably get it done 2017-09-22 03:43:45 I have a Creator CI20 which is mips32 LE, and an Indy which is mips64 BE, and a Netgear router running mips32 BE, so I can test *all* the things! 2017-09-22 03:44:16 the netgear has openwrt on it already, so I can test static-linked packages on it already, and since I just use it as repeater, nothing too much lost if I end up breaking it 2017-09-22 04:14:23 kaniini: thanks! 2017-09-22 04:35:34 foxkit: i'm not really interested in mips32 port atm, mips64 is the bigger 'win' target imo 2017-09-22 04:36:11 a lot of routers are mips32 these days, but yeah, mips64 is still pretty good server iron. 2017-09-22 04:36:25 I started with mips64 myself 2017-09-22 05:40:10 https://github.com/alpinelinux/aports/commit/f7adf1f7133fe013da0f42b0c6befda9181b812c was commited but i got https://puu.sh/xFRSd/fc4daf56ee.png in my email 2017-09-22 05:52:15 that email looks non-legit because of all the broken english 2017-09-22 05:52:19 I guess it was just written in a hurry 2017-09-22 05:52:53 is that email template something that lives in an open repo? I'd be happy to fix it up 2017-09-22 05:54:17 Xe: it looks like that email was sent less than one hour after that was committed, so it probably 'crossed in the post' to use an outdated idiom 2017-09-22 05:54:26 I don't know what the modern version of that idiom is 2017-09-22 05:54:55 sweet 2017-09-22 05:55:07 race condition 2017-09-22 06:07:30 not necessarily? if this automated process is using, say, the ARM repo, the arm builders are slow/buggy/crashy it seems like 2017-09-22 06:07:47 so it is possible the ARM builders had not built it yet 2017-09-22 06:07:50 resulting in that flag 2017-09-22 06:22:54 <^7heo> foxkit: I'm not sure this lives in an open repo but I believe you can submit your corrections on the ml and someone (tm) will eventually fix it 2017-09-22 06:25:47 <^7heo> also fwiw your explanation for that behavhor is the only one that makes sense to me; so it must be that ;) 2017-09-22 06:29:19 <^7heo> but it's impossible to confirm without a more precise datetime for the mail 2017-09-22 07:20:31 anitya only updates once per day afaik 2017-09-22 07:52:23 kaniini: yeah, mips has had DYBs since Q3 2014 or so 2017-09-22 07:53:23 foxkit: what is a nekochanite? 2017-09-22 07:53:46 nekochan.org 2017-09-22 07:53:50 and yes, mips32 is more widespread, mips64 is, well, used more for servers and other devices 2017-09-22 07:53:56 kaniini: i see, thanks 2017-09-22 07:54:34 just a sec 2017-09-22 07:54:37 what am i looking at...? 2017-09-22 07:55:07 i've been around weebs online for years on a chat site, so i know what a neko is and what -chan means, but that just looks like a gif to me 2017-09-22 07:58:11 oh 2017-09-22 07:58:14 that used to be a website 2017-09-22 07:58:20 for people who ran SGI IRIX 2017-09-22 07:58:25 i think i've seen it before but i have no recollection of -- 2017-09-22 07:58:27 ah okay 2017-09-22 08:19:36 foxkit: so basically, you're interested in running alpine/adelié on older hardware? 2017-09-22 08:21:50 s/older/non-mainstream but still alive/ but yes 2017-09-22 08:23:27 older 2017-09-22 08:23:35 SGI no longer make mips kit 2017-09-22 08:25:34 I used 'alive' in the sense of 'there's a community of people still using it' 2017-09-22 08:25:59 it's still 'not new' 2017-09-22 08:26:15 actually SGI no longer exist at all 2017-09-22 08:26:22 they got acquired by HP 2017-09-22 08:26:22 go figure 2017-09-22 08:28:39 fate: chapter 11 bankruptcy 2017-09-22 08:28:40 damn 2017-09-22 08:29:41 oh, sorry. HPE, which is basically the new Compaq 2017-09-22 08:29:41 right, SGI -> rackable (renames itself to SGI) -> HP 2017-09-22 08:31:05 oh right, true 2017-09-22 08:49:36 cool, an alpine mask on freenode 2017-09-22 08:53:52 kaniini: foxkit: willing to put some work into alpine on mips? 2017-09-22 08:56:29 danieli: http://forums.nekochan.net/ 2017-09-22 08:56:50 danieli: not exclusively older, but older is a target, yes. 2017-09-22 08:56:57 well yeah 2017-09-22 08:57:15 it never ceases to amaze me how surprised people are that I think instead of throwing computers out you can just put linux on them 2017-09-22 08:57:51 danieli: I definitely want to see a MIPS port happen. I don't know how much free time I'll have in the next week or so however 2017-09-22 08:58:02 I don't think it's the surprising part 2017-09-22 08:58:10 time is a constant :) 2017-09-22 08:58:25 the surprising part is that you insist it's a good ROI 2017-09-22 08:58:54 because most people would rather throw the old computer out and buy another one and save time 2017-09-22 08:59:05 hmm, you can do everything a new computer can do, but save your money and use a few-years-old one instead :P 2017-09-22 08:59:10 i doubt it's a good ROI but it could be a target once it's running on more modern hardware 2017-09-22 08:59:16 (and I'm not saying I agree with that attitude) 2017-09-22 09:00:18 danieli: idk, my first 'off-brand' port of alpine was back in the uclibc days, to ppc 2017-09-22 09:00:38 danieli: a 2002 Power Macintosh ran a moderately-trafficked rails portal using it, for about a year 2017-09-22 09:00:56 saw somewhere around 200-300 req/s 2017-09-22 09:01:21 that could've been ran on newer hardware, saving the operators a lot of effort 2017-09-22 09:01:27 it's fun and all but i'm not so sure it's especially useful 2017-09-22 09:02:11 it sure saved me a lot of money. rails is typically extremely memory hungry, and boards that took 8 GB RAM back then were not easy to come by. or I could have paid 6000$ to dell 2017-09-22 09:02:22 or paid 0$ and kept the old server going another year 2017-09-22 09:03:08 mm, good point, but it's not a good analogy today 2017-09-22 09:03:13 it probably wouldn't have been worth it if it was any serious amount of effort, but you have to realise that I was doing cross-toolchains in high school, this is all second nature to me, I had it up and running in two days. 2017-09-22 09:04:04 if only I was as good with 3D graphics as I am with toolchains :( 2017-09-22 09:05:26 aaaanyway, that's way OT (or is it, since it was an alpine box XD); the point is, yes, I want to see Alpine on MIPS and I have the hardware and the technical knowledge to likely make it a reality. I just need help figuring out some of the bits of GCC that keep churning (as noted in the ML post) 2017-09-22 09:05:43 I'm willing to do the work, I'm just concerned about the time 2017-09-22 09:05:50 I could get a hold of people to help with toolchains and cross compiling 2017-09-22 09:06:10 i just don't see why having older hardware as a primary target is a good ROI, as skarnet said 2017-09-22 09:06:20 getting it working eventually could be nice though 2017-09-22 09:06:38 that's not really what skarnet said 2017-09-22 09:06:46 also, lowest-common-denominator 2017-09-22 09:06:55 I SAID WHAT I SAID 2017-09-22 09:07:13 you get it working on the SGI O2, you're pretty much guaranteed to work on any mips64 box newer than that 2017-09-22 09:07:36 though you may want a custom build for loongson since it has some weird ISA extensions 2017-09-22 09:07:56 yeah, loongson has some weird shit 2017-09-22 09:08:22 but if you /start/ targeting the newest possible computer, then you end up having to put in more effort to make it work on older things, which leads you to maybe thinking it isn't worth it, because it's even /more/ effort 2017-09-22 09:08:36 if you /start/ with older things, there's /no/ effort to target them, and very minimal - if any - effort to target newer things 2017-09-22 09:09:08 performance is an issue too 2017-09-22 09:09:37 a good analogy here might be: if you write assembly that only works on skylake, then someone asks you to port it to ivy bridge, you're gonna be in some soup. if you just start targeting sandy bridge and add a routine that "if skylake extensions are available then use this", you're set for life 2017-09-22 09:10:00 and you've also got a more extensible system when the next gen comes out 2017-09-22 09:10:09 but idk. 2017-09-22 09:10:37 danieli: I'm not saying we build /on/ an SGI O2. those are very expensive anyway. it's just simple gcc switches to target MIPS III or MIPS IV 2017-09-22 09:12:04 danieli: alpine's x86 triplet is i486-alpine-linux-musl, but I doubt it's built on a 486, you know? :P 2017-09-22 09:12:18 yeah, point taken 2017-09-22 09:12:54 that would be kind of hilarious though, get one of those 4-way compaq servers from 1993 and try to build a whole linux distro on it... come back in two years 2017-09-22 09:13:58 danieli: my exact problem is that mips-linux targeted gcc seems to require linux-headers and there's no way to get a bootstrap cross linux-headers that I can figure out using abuild 2017-09-22 09:14:28 i'm gonna test, one sec 2017-09-22 09:14:36 danieli: this isn't a problem on, say, gentoo, where crossdev brings in sys-kernel/hardened-headers before gcc-stage2 2017-09-22 09:14:54 danieli: so I think my issue is more that I have a lack of experience with abuild, more than an actual issue in gcc 2017-09-22 09:15:04 which is why I posted on the ML and not the gcc bz :P 2017-09-22 09:16:20 I have noticed the ML is kinda comatose lately :/ 2017-09-22 09:16:51 IRC and github are still active and lively, so I'm not worried about alpine as a project, but it is making me think twice before sending mail 2017-09-22 10:22:14 just compiled the 4.9.50 vanilla kernel for ar71xx (a router i have here) 2017-09-22 10:42:33 https://p.duniel.no/8MFi 2017-09-22 11:22:04 hi. how can i mount a raid array (fstab) after raid is created on xen-alpine 3.6.2? 2017-09-22 11:22:30 hi. how can i mount a raid array (fstab) after raid is created on xen-alpine 3.6.2? 2017-09-22 11:22:37 were you the one who posted that in #alpine-linux not long ago? 2017-09-22 11:23:07 yes 2017-09-22 11:52:19 i have rebased the zfs patches but it crashes men modprobe 2017-09-22 11:57:43 hmm 2017-09-22 12:08:25 > i have rebased the zfs patches but it crashes men modprobe 2017-09-22 12:08:26 ncopa is there any guide or anything regarding the ZFS on Alpine? It has some problems after 3.6 release and I didn't check later. Does it work ootb nowadays? 2017-09-22 12:09:30 scadu: not really 2017-09-22 12:09:39 its somehwat complicated to install it 2017-09-22 12:09:41 well 2017-09-22 12:09:55 unless you have it only for a data disk 2017-09-22 12:10:24 problem with hardened kernel is ROP 2017-09-22 12:11:34 really old pointers? they decay with age and hardened doesn't like that? 2017-09-22 12:12:05 problem is function pointer typecases 2017-09-22 12:12:18 (which to my understanding is UB anyways) 2017-09-22 12:12:59 with ROP it will go boom if function pointer types does not match 2017-09-22 12:19:55 you mean RAP 2017-09-22 12:20:00 :P 2017-09-22 12:20:21 ROP is the exploit technique RAP aims to prevent 2017-09-22 12:20:24 probs PaX RAP 2017-09-22 12:20:25 ^ 2017-09-22 12:20:27 RIP RAP ROP, ofc... 2017-09-22 12:20:32 ROP is evil 2017-09-22 12:20:42 i mean RAP ofc 2017-09-22 12:20:58 and i think i found it 2017-09-22 12:21:19 some of the things in current patch went upstream 2017-09-22 12:21:30 so i thought they added suport for RAP in zfs upstream 2017-09-22 12:21:50 and i did find some "comaptility: grsecurity .." commits 2017-09-22 12:21:56 but no... 2017-09-22 12:31:59 damn you, zfs 2017-09-22 12:37:07 what would be nice would be a gcc flag that would error out on functino pointer typecasts 2017-09-22 12:38:11 actually, the only kernel module i have fixed RAP for is zfs 2017-09-22 12:38:22 i think vbox modules are broken too 2017-09-22 12:38:32 that wouldn't surprise me 2017-09-22 12:38:39 i havent checked xtables-addons 2017-09-22 14:26:55 Shiz: do you have time to help with rebasing the rust patches to newer rust? 2017-09-22 14:27:48 yea 2017-09-22 15:09:22 Shiz: this is as far as i got https://p.duniel.no/K9ep 2017-09-22 15:09:53 do double check them 2017-09-22 16:49:31 if a subpackage of a package gets removed, do i create a separate package for that subpackage? 2017-09-22 16:49:36 in this case, erlang-gs from erlang 2017-09-22 16:51:14 do i just remove it completely? 2017-09-22 20:28:42 had an oopsie with git, didn't realize i had an erroneous argument. ignore the aports mail. 2017-09-22 21:27:09 looks like py2-twisted is broken. i'm gonna make a py-automat package and make it python agnostic. 2017-09-22 21:34:17 ncopa: what is the process for renaming a wrongly named package? 2017-09-22 21:34:24 i have a new package that needs the name 2017-09-22 21:34:35 the alpine package is py-graphviz, should be py-pygraphviz 2017-09-22 21:34:44 clumsy name but such is life 2017-09-23 02:51:48 Hi! It seems like the armhf edge build server is stuck again, it has been at testing/filebeat for a couple hours now, which seems a tiny bit unrealistic. 2017-09-23 08:07:18 filebeat has been blocking builders for a while already :/ 2017-09-23 08:27:42 good news everyone 2017-09-23 08:27:52 the grub package in main works on big-endian powerpc with no changes! 2017-09-23 08:28:13 now if I can just manage to write a grub.cfg file and put a kernel on that doesn't immediately make the firmware crash..... 2017-09-23 08:28:49 cooool 2017-09-23 08:42:35 https://i.imgur.com/yfEEkK8.jpg 2017-09-23 08:43:09 failed to execute /init 2017-09-23 08:43:14 yeah yeah BUT 2017-09-23 08:43:16 it booted 2017-09-23 08:43:18 well, that means the kernel loads 2017-09-23 08:43:18 yeah 2017-09-23 08:43:27 compile openrc!!!1 2017-09-23 08:43:35 and busybox 2017-09-23 08:43:46 I've already compiled over 1700 packages lol 2017-09-23 08:43:47 or just busybox and init=/bin/sh 2017-09-23 08:43:49 including those 2017-09-23 08:43:50 lol holy shit 2017-09-23 08:44:04 is this ppc.. what, exactly? 2017-09-23 08:44:10 ppc32be? 2017-09-23 08:44:11 this is a PowerPC G4 iMac 2017-09-23 08:44:13 yes 2017-09-23 08:44:15 32be 2017-09-23 08:44:19 guessed as much 2017-09-23 08:44:33 a friend of mine in the UK has 5x PPC G5 iMac 2017-09-23 08:44:37 they're just laying around 2017-09-23 08:44:55 but if you've built that much software, why aren't you trying it out? like, setting up init and all 2017-09-23 08:45:11 I have a quad power mac G5 tower that is the ppc32 builder machine; bootstrapped using debian, but it's running native using debian's bootloader and kernel 2017-09-23 08:45:16 it even has X11 =) 2017-09-23 08:45:25 danieli: I *am* trying it out. I dunno why /init failed to exec lol 2017-09-23 08:45:34 it should have loaded a live CD environment 2017-09-23 08:45:46 ah okay, oopsies 2017-09-23 08:46:04 ... I wonder if it's because /lib is empty 2017-09-23 08:46:06 :| 2017-09-23 08:46:21 LOL 2017-09-23 08:46:22 well shit 2017-09-23 09:16:29 https://i.imgur.com/EN3VrIf.jpg 2017-09-23 09:16:47 haha, didn't enable the right controller driver, so it sees no block devices... 2017-09-23 09:16:49 no boot media found 2017-09-23 09:16:54 lol attempted to kill init 2017-09-23 09:17:16 and cool - i2c 2017-09-23 09:22:54 need sleep, it's 4:22 AM here, but that's some major progress. tomorrow I reconfigure the kernel and hopefully enable the right controller... 2017-09-23 09:23:32 yeah, go to sleep - see you later, and have a good one :) 2017-09-23 09:23:50 thanks, you too! :3 2017-09-23 09:25:31 thankies <3 just bumping versions and some other stuff when i'm done 2017-09-23 09:31:14 ncopa: considered having an #alpine-issues? 2017-09-23 09:31:22 redmine can hook nicely 2017-09-23 09:37:28 LOL I like how htop on my machine displays the wrong usernames due to uids in docker containers 2017-09-23 10:26:52 i wonder what the heck this is supposed to mean (syslog-ng): https://p.duniel.no/emtd 2017-09-23 10:27:29 <_ikke_> submodule sounds like git 2017-09-23 10:27:32 <_ikke_> not sure though 2017-09-23 10:27:47 yeah, it is a git thing 2017-09-23 10:27:51 but it isn't in the github release tgz 2017-09-23 10:28:08 <_ikke_> No, they do not include submodules 2017-09-23 10:28:21 aah, explains it 2017-09-23 10:28:28 i should really start using github a bit more 2017-09-23 10:28:36 <_ikke_> s/github/git :P 2017-09-23 10:28:57 i've used git enough, but mostly for pragmatic proprietary projects 2017-09-23 10:29:35 <^7heo> the thing with github is that is it the ONLY actually good git management interface out there. 2017-09-23 10:29:42 <^7heo> everything else sucks to some degree. 2017-09-23 10:29:56 gitlab and bitbucket aren't too bad imo, but they're kinda.. clunky 2017-09-23 10:44:31 gitlab is okay but a beast to run yourself 2017-09-23 10:44:36 it needs so many packages and resources 2017-09-23 10:44:40 i quite like gitea though 2017-09-23 10:44:41 yeah, that's true 2017-09-23 10:45:00 i don't think i've heard of gitea 2017-09-23 10:45:02 similar to gogs? 2017-09-23 10:45:06 it's a gogs fork :P 2017-09-23 10:45:18 what edge does it have over gogs? 2017-09-23 10:47:12 more active development, afaik 2017-09-23 10:47:56 aha 2017-09-23 10:50:58 <^7heo> oh btw 2017-09-23 10:51:03 <^7heo> I have made a js mod on gogs 2017-09-23 10:51:09 <^7heo> to allow inline comments in the diffs 2017-09-23 10:51:28 <^7heo> currently it works with an external api to store the data, but I'll change that to use the "conversation" thread. 2017-09-23 10:51:39 <^7heo> mosez: would you integrate that? 2017-09-23 10:51:43 <^7heo> mosez: even if it's js only? 2017-09-23 12:12:06 ncopa: is the s390x builder broken? 2017-09-23 16:14:40 Hello, I need a python3 version of both py-dbus and py-gobject. I have hacked the current py-dbus apkbuild to support python3 but how should I go about adding "proper" dual-python support? Both libraries use the gnu autotools (./configure) 2017-09-23 16:17:46 z3ntu: https://wiki.alpinelinux.org/wiki/APKBUILD_examples:Python#Multiversion_package_without_executables 2017-09-23 16:19:15 danieli: Yes I know of that page but it's only for packages with a setup.py which aren't these two packages. Arch Linux uses two seperate build directories for such packages but I don't know if that's "acceptable" in Alpine 2017-09-23 16:19:42 ah, gotcha.. well, separate apkbuilds for each package, and $submoduledir 2017-09-23 16:19:50 iirc that's the variable 2017-09-23 16:21:35 z3ntu: it's acceptable 2017-09-23 16:21:50 no separate apkbuilds 2017-09-23 16:22:07 i meant for py-dbus and -gobject respectively 2017-09-23 16:58:04 / 24 2017-09-23 17:23:01 hmm looks like the armhf edge build server is still stuck on filebeats package. Who can kick it? 2017-09-23 17:24:51 <_ikke_> Might be that there is more that is stuck 2017-09-23 17:25:39 <_ikke_> clandmeter, ncopa, kaniini: Anyone available to look at the build servers? 2017-09-23 17:26:17 it means the filebeats package is broken 2017-09-23 17:26:18 should fix it 2017-09-23 17:32:56 _ikke_: there /are/ more that are stuck 2017-09-23 17:33:02 the s390x builder is dead, or so it seems 2017-09-23 17:33:07 and a handful of packages are stuck 2017-09-23 17:33:09 i'll check it out 2017-09-23 17:33:14 (the packages themselves) 2017-09-23 17:56:56 the log for filebeats shows that it can't find ld, which seems odd 2017-09-23 18:23:23 hello there, i want to install arch as main distro from alpine- after change from swap partition to ext on /dev/sda2 i installed arch on it. now can i delete alpine or i must change syslinux settings to boot on sda2 and after that delete alpine from main? 2017-09-23 18:26:20 i am a little new in linux and do not know how some things are done 2017-09-23 18:27:55 <_ikke_> I would first make sure you can boot to arch, before deleting your alpine install 2017-09-23 18:28:28 how to do that without any rescue usb/cd ? :D 2017-09-23 18:29:50 <_ikke_> uzerus: I don't know syslinux that well, but you can check if you can dualboot 2017-09-23 18:30:33 sure, thanks @_ikke_ 2017-09-23 19:00:12 Aport patches to alpine-aports@lists.alpinelinux.org, right? 2017-09-23 19:03:51 z3ntu: correct 2017-09-23 19:04:03 git patches go there, get added to patchwork.alpinelinux.org for review 2017-09-23 19:17:42 danieli: Could you take a look at http://patchwork.alpinelinux.org/patch/3593/ ? I made it as good as possible but it might be a bit hacky / don't follow conventions etc 2017-09-23 19:19:45 z3ntu: I'll take a look but I don't have write access to pretty much anything 2017-09-23 19:19:58 I got it to my mailbox btw ;) 2017-09-23 19:47:13 <_ikke_> Any way to provide optional / recommended dependencies (in case where you need one dep from a set of deps) 2017-09-23 19:55:12 Looks like build-edge-armhf is stuck again with another failed package. Why do failed packages break the build servers? Shouldn't the server just move on to the next package if one fails, and maybe flag that package somehow as 'failed' ? 2017-09-23 19:55:25 they should, that's the logical thing to do 2017-09-23 19:55:58 <_ikke_> I think they are working on an update to fix these kind of issues 2017-09-23 19:56:09 who, exactly? 2017-09-23 19:56:13 is this mostly custom made stuff? 2017-09-23 19:56:19 <_ikke_> yes 2017-09-23 19:56:54 i see. 2017-09-23 19:57:02 i'm not too familiar with alpine's build system internals 2017-09-23 19:57:22 me neither, hence the question I had :P 2017-09-23 19:57:33 <_ikke_> me neither, but they use a lua version of apk / abuild 2017-09-23 19:57:45 <_ikke_> https://git.alpinelinux.org/cgit/lua-aports/ 2017-09-23 19:58:05 well if there's anything we can do to help with fixing this somehow, don't be shy! 2017-09-23 19:58:56 technically i did the work to enable that already, it is just a matter of the buildserver script being updated to use rootbld 2017-09-23 19:59:05 buildrepo already can use rootbld 2017-09-23 19:59:28 and nmeum did the actual work on rootbld itself 2017-09-23 20:59:55 z3ntu,this seems close to postgresql build which does same thing for multi python 2017-09-23 21:00:09 z3ntu, why did you remove pkgrel? 2017-09-23 21:00:44 pkgrel should be incremented unless the pkgver is incremented, in which it should be reset to 0 2017-09-23 21:00:52 to trigger a rebuild of the package 2017-09-23 21:03:09 foxkit: I'll also update it to 1.2.4 then as that version is out, or is there a reason it's still at 1.2.0? 2017-09-23 21:03:33 check if the newer version builds. if it does, it was probably just neglected 2017-09-23 21:06:07 i'm tired as all hell, good night 2017-09-23 21:06:46 <_ikke_> nite 2017-09-23 21:09:56 danieli: (even if you only see it tomorrow) 1.2.2 & 1.2.4 give me ">>> ERROR: py-dbus: all failed"... I love error messages :P 2017-09-23 21:10:44 error messages are fun 2017-09-23 21:20:08 Log is in the next message 2017-09-23 21:20:12 ACTION sent a long message: z3ntu_2017-09-23_21:20:11.txt  2017-09-23 21:23:11 huh `update_config_sub` breaks the build apparently... 2017-09-23 21:27:26 Wait, this channel is on matrix somewhere? 2017-09-23 21:27:50 <_ikke_> craftyguy: not yet 2017-09-23 21:28:02 <_ikke_> craftyguy: they are looking into matrix though 2017-09-23 21:28:06 z3ntu's message seems to indicate there's a bridge? 2017-09-23 21:28:33 <_ikke_> I'm not aware of any 2017-09-23 21:28:36 Yes I'm using Matrix 2017-09-23 21:28:44 you can access irc via matrix 2017-09-23 21:28:44 You can access every freenode channel from matrix 2017-09-23 21:28:58 oh! I didn't know that.. what's the trick? 2017-09-23 21:30:14 https://matrix.org/blog/2015/06/22/the-matrix-org-irc-bridge-now-bridges-all-of-freenode/ 2017-09-23 21:30:54 With Riot you can go to "Room directory", switch the dropdown to Freenode and enter your #channel 2017-09-23 21:31:18 Thanks! 2017-09-23 21:35:18 Got py-dbus 1.2.4 built and (at least) `python3 -c "import dbus"` (and same with python2) works 2017-09-23 23:04:44 Has anyone know about making CMake work properly with Backtrace? 2017-09-23 23:58:56 https://i.imgur.com/gQV3y2e.jpg 2017-09-24 00:01:30 aw man sweet 2017-09-24 04:15:34 we are planning to eventually make the matrix side of the channel the authoritative side, but aren't quite there yet 2017-09-24 04:36:15 kaniini: how do I take you up on your offer to move connman to community & have me maintain it? Anything I need to do? 2017-09-24 04:37:57 Oh, and it seems that build-edge-armhf is stuck on a failed package. Half of the connman arch packages haven't been able to build for several days now because of build systems getting stuck 2017-09-24 06:35:27 craftyguy: yes, just submit a PR doing it 2017-09-24 07:14:04 rootbld seems to have some issues 2017-09-24 07:14:29 it doesn't respect $sysconfdir, just yanking /etc/abuild.conf directly; while that's probably not the worst thing it could do, it seems like it might cause some issues/headaches later 2017-09-24 07:16:45 additionally, it completely ignores the hostname of the current machine, just calling it "build-edge-$CARCH". which might be a silly thing to complain about, but what if it isn't even edge being built? what if you want or need to know, for whatever reason, *which* builder built this package? 2017-09-24 07:18:37 foxkit, patches are welcome :) 2017-09-24 07:18:47 but yeah, the hostname thingy was accidentally left in 2017-09-24 07:18:51 needs to be fixed 2017-09-24 07:19:42 I like to talk about abuild before jumping in and fixing it, don't want to break it 2017-09-24 07:51:29 can someone explain to me why some vars in abuild.conf are exported and others are not? is there a reason, or is it just style-related? 2017-09-24 09:59:02 i am back from the dead 2017-09-24 13:40:22 <_ikke_> Any reason why openrc might loose track of a process? I noticed twice already today that it ran a second instance of a process, and the old process is not stopped anymore 2017-09-24 14:09:50 because it's not a proper supervision suite. 2017-09-24 14:12:02 <_ikke_> I think because the default config does not write a pid-file 2017-09-24 14:31:17 I'm pointing at the moon and you're looking at the finger 2017-09-24 14:33:33 <_ikke_> I understood what you said 2017-09-24 14:33:45 <_ikke_> But even what openrc normally does, it did not do 2017-09-24 14:53:51 _ikke_: is having two versions of a package in the repos acceptable? 2017-09-24 14:54:10 like python2 and python3 2017-09-24 14:54:45 in this case, prosody (0.9.x), i want to create "prosody-0.10" (beta) as the pkgname 2017-09-24 15:18:42 the `nodejs-current` package is flagged by someone, says "Hi, this package doesn't work anymore on Alpine x64! :O When I try to run node or npm I always get this error: Error relocating /usr/bin/node: uv_fs_copyfile: symbol not found" 2017-09-24 15:18:56 i cannot reproduce this 2017-09-24 16:32:59 kaniini: submitted PR on mailinglist to move connman, thanks for the help :) 2017-09-24 16:33:56 do we add a contributor field if we just update package versions or move them? 2017-09-24 16:34:12 <_ikke_> I wouldn't bother 2017-09-24 20:02:00 _ikke_: hey :) are you available atm? 2017-09-24 20:03:26 <_ikke_> Bloo[m]: I am 2017-09-24 20:03:47 _ikke_: Id really appreciate it if you could help me out with `cor`? 2017-09-24 20:04:19 Im really stumped with one error Im getting 2017-09-24 20:04:22 cant get around it at all 2017-09-24 20:05:16 http://paste.debian.net/987333/ It seems as though `backtrace` is undefined for some reason... Regardless of having the correct packages for it. I've also included `FindBacktrace.cmake` in my Cmake files but no luck 2017-09-24 20:05:34 woah, yikes... Sorry for that lengthy link preview wtf lol 2017-09-24 20:06:59 <_ikke_> Bloo[m]: what is your apkbuild? 2017-09-24 20:07:36 _ikke_: https://hastebin.com/bolebofoyu.bash 2017-09-24 20:07:58 <_ikke_> why is this running sudo? 2017-09-24 20:08:59 <_ikke_> Bloo[m]: What package should provide backtrace? 2017-09-24 20:09:39 sigh 2017-09-24 20:09:45 i'm going to just patch backtrace into our musl 2017-09-24 20:09:46 i no longer care 2017-09-24 20:10:04 <_ikke_> kaniini: Is this a common issue? 2017-09-24 20:10:17 yes 2017-09-24 20:10:26 <_ikke_> ah ok 2017-09-24 20:10:28 and from what i see musl 1.1.17 will never be released 2017-09-24 20:10:35 and musl 1.2 is vapourware too 2017-09-24 20:10:37 so 2017-09-24 20:10:38 dunno 2017-09-24 20:11:19 because hyperbole always helps 2017-09-24 20:11:28 i sure wish i was wrong about that, but our patch counts are increasing and largely they are unreleased upstream fixes that have been sitting for months, so it is starting to look like a uclibc situation again 2017-09-24 20:11:35 _ikke_: Oh thats just pmbootstrap, that shouldnt effect anything afaik as its all in a chroot. As for which package shjould rovide backtrace `libexecinfo-dev` should do 2017-09-24 20:11:39 and it does 2017-09-24 20:11:56 because the `FindBacktrace.cmake` file I put in as a CMake module actually finds it 2017-09-24 20:12:00 yes, it does. i am thinking about patching the linker script to pull it in 2017-09-24 20:12:01 but for whatever reason the code doesnt seem to want to know 2017-09-24 20:12:30 <_ikke_> kaniini: :-( 2017-09-24 20:12:34 Bloo: because the code assumes libexecinfo will just be pulled in by the libc linker script 2017-09-24 20:13:02 this assumption being so common that we used to do exactly this with uclibc 2017-09-24 20:13:28 ohhh, so (forgive me for my ignorance) what should I do to fix this? 2017-09-24 20:14:10 just hack LDFLAGS for now 2017-09-24 20:14:36 i've made a mental note about possibly conditionally pulling libexecinfo in with the linker script 2017-09-24 20:16:05 <_ikke_> kaniini: is there any sane alternative libc available then? 2017-09-24 20:16:19 not right now, which is annoying 2017-09-24 20:16:26 we are pretty much stuck with what we have 2017-09-24 20:16:32 <_ikke_> hmmf 2017-09-24 20:16:50 kaniini: sorry I still dont follow, what would I put in LDFLAGS? 2017-09-24 20:16:51 i have heard that glibc 3 intends to import lots of the good stuff from musl as well as make it possible to disable legacy stuff 2017-09-24 20:17:01 Bloo: -lexecinfo 2017-09-24 20:17:03 Like a fish out of water with this, just started out wiht porting 2017-09-24 20:17:09 aha 2017-09-24 20:17:11 so maybe glibc 3 is a possible alternative if that ever ships 2017-09-24 20:17:32 lol let's see how this plays out 2017-09-24 20:17:42 but musl is not what i would define as sane either 2017-09-24 20:17:53 different problems 2017-09-24 20:18:44 it's all crap 2017-09-24 20:18:48 kaniini: so that would be (on the buld command) cmake -DLDFLAGS="-lexecinfo" 2017-09-24 20:18:53 gonna test it now 2017-09-24 20:18:55 cheers 2017-09-24 20:18:55 no 2017-09-24 20:18:59 make LDFLAGS="-lexecinfo" 2017-09-24 20:19:02 ohh 2017-09-24 20:20:00 skarnet: all i know is that we have to layer a bunch of stuff on top of musl to make it palatable, and then even things are still broken, and if you bend musl to be workable then dalias gets all emo about it 2017-09-24 20:20:05 um, normally -l stuff goes into LIBS 2017-09-24 20:20:06 Using `cmake` doesnt seem to like that, just says `2a40fefd97746188b274/-LDFLAGS=-lexecinfo" does not exist.` 2017-09-24 20:20:15 you shouldn't have to pass -lstuff into LDFLAGS, that's not what it's for 2017-09-24 20:20:28 skarnet: cmake doesn't care about LIBS 2017-09-24 20:20:42 oh, cmake 2017-09-24 20:20:58 for some reason anyway... I did try this a while back coming to tink of it,. Could never get it to work without it appending the -LDFLAGS to the file name :( 2017-09-24 20:21:05 folder* 2017-09-24 20:21:18 i said make, not cmake 2017-09-24 20:21:33 ohh, didnt know you could use make instead 2017-09-24 20:21:44 you use cmake to make the makefiles 2017-09-24 20:21:50 and then you use make to build it 2017-09-24 20:21:51 ah 2017-09-24 20:22:03 you do the LDFLAGS with the make invocation 2017-09-24 20:25:28 kaniini: right so running cmake to make the makefiles and then running cmake with the LDFLAGS you gave me sadly brings me the same error 2017-09-24 20:27:34 try with LIBS maybe instead 2017-09-24 20:27:44 but i am pretty sure cmake does not care about LIBS 2017-09-24 20:27:47 also maybe try 2017-09-24 20:27:59 env LIBS=-lexecinfo cmake ... 2017-09-24 20:29:04 kaniini: damn... Still no dice 2017-09-24 20:36:49 ACtually, Ive heard of a fix for this called `FindBacktrace.cmake` but it didnt work 2017-09-24 20:38:22 that requires modifying the CMakeLists.txt 2017-09-24 20:38:40 which is its own nightmare 2017-09-24 20:42:43 kaniini: true, did try it but nothing came of it sadly... Not usre what else to do at this point 2017-09-24 20:44:41 have you asked upstream to properly integrate findbacktrace.cmake ? 2017-09-24 20:45:05 Ive contacted one of the devs yeah, awaiting a response 2017-09-24 20:45:21 although Il contact a few more of the maintainers 2017-09-25 06:42:42 hi all 2017-09-25 07:09:38 morning climbers 2017-09-25 07:10:52 hi clandmeter 2017-09-25 07:15:38 ^7heo: there is an open issue to add inline comments generally 2017-09-25 08:11:27 <^7heo> mosez: yeah well I did it. 2017-09-25 08:11:30 <^7heo> mosez: about client side. 2017-09-25 08:11:38 <^7heo> s/about/aber/ 2017-09-25 08:11:42 <^7heo> s/aber/but/ 2017-09-25 08:43:44 morning 2017-09-25 08:44:04 <^7heo> moin 2017-09-25 08:44:28 <^7heo> ls 2017-09-25 08:44:31 <^7heo> oops sorry 2017-09-25 10:46:02 ^7heo: someone's tired :) 2017-09-25 12:00:24 <^7heo> danieli: indeed. 2017-09-25 13:32:38 hey friends, is there a way to tell the busybox ntpd to sync the time no matter what the difference is? 2017-09-25 13:37:45 hi 2017-09-25 13:37:50 does not look like that 2017-09-25 13:38:03 the sntpc may do that for you 2017-09-25 13:39:09 i am currently looking in the code, maybe it is just a disabled feature 2017-09-25 13:40:19 danieli: erlang uses openjdk7. do you think we could/should useopenjdk8 instead? 2017-09-25 13:40:30 since openjdk7 is no longer uspported upstream 2017-09-25 13:53:44 https://pkgs.alpinelinux.org does timeout 2017-09-25 13:53:56 do you have the same problem? 2017-09-25 13:56:38 works fine for me 2017-09-25 14:00:41 we had a small power issue 5 min ago. all should be ok now. 2017-09-25 14:01:00 algitbot, thanks 2017-09-25 14:01:08 <_ikke_> lol 2017-09-25 14:17:11 ncopa: sure 2017-09-25 14:17:14 thanks for the heads up 2017-09-25 14:20:44 openjdk 9 is out btw 2017-09-25 14:33:48 ncopa: oh you fixed it, neat 2017-09-25 14:45:38 danieli: look slike the abi changed too 2017-09-25 15:53:41 just curious, why do APKBUILDs seem to favor tarballs compressed with xz (.tar.xz) rather than gzip (.tar.gz)? 2017-09-25 15:53:44 is there a reason? 2017-09-25 15:54:15 the speed tradeoff? 2017-09-25 15:54:44 i think i just answered my own question, nevermind. speed and compression efficiency. 2017-09-25 15:59:35 better compression 2017-09-25 15:59:51 yeah, decompressing is faster with xz 2017-09-25 16:00:32 don't know if gz decompressors utilise multiple threads; xz does 2017-09-25 16:01:18 i think GNU gzip is multithread 2017-09-25 16:01:32 whiel busybox gzip is not 2017-09-25 16:01:57 danieli: i think the imagemagick update breaks abi 2017-09-25 16:02:04 and for some reason awesome does not build 2017-09-25 16:02:32 ncopa: hmm.. 2017-09-25 16:02:42 god damn imagemagick 2017-09-25 16:02:48 i also removed the --without-threads while at it 2017-09-25 16:04:59 http://tpaste.us/VbWe 2017-09-25 16:05:15 https://abi-laboratory.pro/tracker/timeline/imagemagick/ 2017-09-25 16:07:08 ncopa: see if 7.0.7-4 unbreaks it 2017-09-25 16:07:22 it doesnt 2017-09-25 16:08:42 is --disable-shared still gone? 2017-09-25 16:09:11 yes, there are no --disable-shared 2017-09-25 16:09:20 in the APKBUILD 2017-09-25 16:13:01 i wonder if i should enable rsvg while at it 2017-09-25 16:13:14 i'm compiling, gonna test 2017-09-25 16:13:22 imagemagick always breaks abi 2017-09-25 16:21:53 ncopa: apparently they renamed libMagickCore-7.Q16HDRI.so.3 to .so.4 2017-09-25 16:21:57 same with libMagickWand 2017-09-25 16:22:15 due to abi breakage 2017-09-25 16:25:54 this smells like a libtool issue 2017-09-25 16:26:02 i suspect so 2017-09-25 16:27:15 yep 2017-09-25 16:27:24 really seems like it 2017-09-25 16:27:42 i wonder if we could use slibtool 2017-09-25 16:27:53 and force slibtool for imagemagick 2017-09-25 16:30:31 readelf -Ws `find / -name "png.so" 2>&1 | head -n 1` 2017-09-25 16:34:36 ncopa: i did some work on that over the summer, and it has promise, but i need to work with midipix to squash a few more bugs 2017-09-25 16:35:12 i was thining using slibtool as hard dep for imagemagick for now 2017-09-25 16:36:22 hum, imagemagick uses libltdl 2017-09-25 16:36:30 configure: error: libltdl is required for modules build 2017-09-25 16:36:30 See `config.log' for more details 2017-09-25 16:37:57 hm 2017-09-25 16:40:12 i wonder if we can do something like: 2017-09-25 16:40:20 #define lt_dlopen dlopen 2017-09-25 16:40:31 #define lt_dlclose dlclose 2017-09-25 16:40:33 etc... 2017-09-25 16:40:51 i see no reason to why we should actually use libltdl 2017-09-25 16:43:26 afk for a bit, gotta help my grandpa sell some ice cream and fill up the ice cream truck 2017-09-25 18:05:07 back 2017-09-25 18:06:15 gone 2017-09-25 18:06:18 :) 2017-09-25 18:06:31 imagemagick should be gone 2017-09-25 18:06:32 :@ 2017-09-25 18:07:56 <_ikke_> hai 2017-09-25 18:09:58 <_ikke_> Why does ash seem to allow ${foo#bar}, but when I build a package, it does not seem to work 2017-09-25 18:14:04 ash is your default shell, right? and 'sh' points to 'ash'? 2017-09-25 18:14:30 <_ikke_> My default shell is bash :P 2017-09-25 18:17:11 <^7heo> some days reading #alpine makes me cry. 2017-09-25 18:17:33 <_ikke_> ^7heo: You object my choice of default shell? 2017-09-25 18:17:40 <^7heo> _ikke_: do you at least write your bash scripts using #!/bin/bash? 2017-09-25 18:17:54 <^7heo> (or, anything else than #!/bin/sh actually) 2017-09-25 18:18:10 <^7heo> _ikke_: yeah I do; but it's your machine. 2017-09-25 18:18:10 <_ikke_> ^7heo: I aim towards writing posix complient scripts 2017-09-25 18:18:15 <^7heo> compliant* 2017-09-25 18:18:20 <^7heo> must be hard, using bash... 2017-09-25 18:18:20 <_ikke_> ^7heo: well, on my machine I actually use zsh 2017-09-25 18:18:30 <^7heo> yeah same here, zsh for interactive shells 2017-09-25 18:18:30 <_ikke_> ^7heo: probably 2017-09-25 18:18:36 <^7heo> otherwise, ash. 2017-09-25 18:18:45 <^7heo> because it's the best POSIX shell I found so far. 2017-09-25 18:19:07 there are a couple nice shells 2017-09-25 18:19:11 <_ikke_> Is $(a#b} not POSIX complient? 2017-09-25 18:19:11 oops 2017-09-25 18:19:13 <^7heo> danieli: :D 2017-09-25 18:19:20 <_ikke_> danieli: nice effect though 2017-09-25 18:19:26 <^7heo> danieli: nice whitespace program; but it doesn't execute in my IRC client. 2017-09-25 18:19:29 a bag of potato chips hit space, I think 2017-09-25 18:19:33 it might have been RTL somehow 2017-09-25 18:19:44 <^7heo> danieli: easy: stop keeping your potato chips in bags. 2017-09-25 18:19:45 I have no idea what I pressed to do that 2017-09-25 18:19:48 uhhh 2017-09-25 18:19:54 no. 2017-09-25 18:20:05 <^7heo> _ikke_: that specific expression isn't gonna work anywhere... 2017-09-25 18:20:23 ${parameter#word}' Remove Smallest Prefix Pattern. The word is expanded to produce a pattern. The parameter expansion then results in parameter, with the smallest portion of the prefix matched by the pattern deleted. 2017-09-25 18:20:36 from the ash man 2017-09-25 18:20:38 <^7heo> danieli: from opengroups? 2017-09-25 18:20:48 <^7heo> danieli: yeah ok, but ash isn't exactly POSIX 2017-09-25 18:20:48 <_ikke_> ash seems to take it 2017-09-25 18:20:55 <^7heo> yeah ash takes more than POSIX 2017-09-25 18:21:05 <^7heo> a little more, the stuff useful and simple to implement. 2017-09-25 18:21:09 <_ikke_> But I still wonder why my package fails to build 2017-09-25 18:21:19 <^7heo> _ikke_: I can take a look in 1h if you want. 2017-09-25 18:21:40 <_ikke_> provides="$(pkgname#alpine-)=$pkgver" 2017-09-25 18:22:01 <^7heo> _ikke_: use {} 2017-09-25 18:22:09 yeah 2017-09-25 18:22:10 <^7heo> otherwise you're asking for a subshell. 2017-09-25 18:22:12 it's specifically curly braces 2017-09-25 18:22:29 <^7heo> _ikke_: that's why I said earlier: < ^7heo> _ikke_: that specific expression isn't gonna work anywhere... 2017-09-25 18:22:36 <^7heo> <@_ikke_> Is $(a#b}..... 2017-09-25 18:22:41 <^7heo> $(} 2017-09-25 18:22:46 <^7heo> Nice smiley but that's about it. 2017-09-25 18:23:02 <_ikke_> agh, I'm being thick 2017-09-25 18:23:06 <^7heo> :D 2017-09-25 18:23:10 <^7heo> It's just monday dude 2017-09-25 18:23:12 thicc 2017-09-25 18:23:15 <^7heo> change your coffee intake 2017-09-25 18:23:21 ^7heo: any day can be an excuse 2017-09-25 18:23:23 <^7heo> more or less, but whatever you're doing atm, it's not working ;) 2017-09-25 18:23:34 <^7heo> danieli: on monday, you're supposed to be a little more fresh ;) 2017-09-25 18:23:47 <_ikke_> http://assets.amuniversal.com/a2274d506d5901301d7d001dd8b71c47 2017-09-25 18:23:49 no, you're supposed to hate life 2017-09-25 18:24:01 <^7heo> danieli: please explain how those two assertions are mutually exclusive. 2017-09-25 18:24:07 <^7heo> danieli: because I fail to see how. 2017-09-25 18:24:31 wat 2017-09-25 18:24:33 <^7heo> _ikke_: pretty much 2017-09-25 18:24:55 btw - https://elvish.io/ and https://fishshell.com/ 2017-09-25 18:25:23 <^7heo> danieli: please tell me how "to be a little more fresh" and "you're supposed to hate life" are mutually exclusive. 2017-09-25 18:25:35 <^7heo> danieli: because AFAICT those are correlated. 2017-09-25 18:25:43 <^7heo> but maybe that's just me. 2017-09-25 18:26:36 not being fresh because the weekend is over 2017-09-25 18:27:10 <^7heo> that does not compute. 2017-09-25 18:27:23 i wasn't being entirely serious 2017-09-25 18:27:38 <^7heo> unless you work *harder* during the weekend than during the week, I do not see how 2017-09-25 18:27:46 <^7heo> and I don't think I know anyone doing that. 2017-09-25 18:27:54 <^7heo> either people work just as much, or less. 2017-09-25 18:32:13 <_ikke_> thanks, that helped 2017-09-25 18:32:17 <^7heo> :) 2017-09-25 18:32:33 <@_ikke_> Is $(a#b} not POSIX complient? 2017-09-25 18:32:35 so many issues 2017-09-25 18:32:36 LOL 2017-09-25 18:33:02 ACTION 's eye twitches 2017-09-25 18:33:12 <_ikke_> danieli: do tell 2017-09-25 18:33:31 for one, your brackets are asymmetrical 2017-09-25 18:33:31 (} 2017-09-25 18:33:34 ^ 2017-09-25 18:34:08 <^7heo> danieli: I already said that twice... 2017-09-25 18:34:14 <_ikke_> I think my left and right brain were in disagreement 2017-09-25 18:34:19 LOL 2017-09-25 18:34:27 ^7heo: i realized after he posted it the second time, the provides one 2017-09-25 18:34:34 <^7heo> :) 2017-09-25 18:34:47 <_ikke_> When I was testing it in a shell, I did use {} 2017-09-25 18:35:15 strange 2017-09-25 18:35:19 might be that it's an ash thing? 2017-09-25 18:35:21 would be strange 2017-09-25 18:36:03 <_ikke_> foo=bar-baz 2017-09-25 18:36:05 <_ikke_> echo ${foo#bar-} 2017-09-25 18:36:41 ~ $ foo="bar-baz" 2017-09-25 18:36:41 ~ $ echo "${foo#bar-}" 2017-09-25 18:36:41 baz 2017-09-25 18:36:55 works in bash too 2017-09-25 18:36:57 <_ikke_> yes 2017-09-25 18:37:01 <_ikke_> ofcourse it works in bash :P 2017-09-25 18:37:01 straaaange 2017-09-25 18:37:15 what's the apkbuild you are trying to build? 2017-09-25 18:37:27 <_ikke_> danieli: alpine branded zabbix 2017-09-25 18:37:36 ah okay 2017-09-25 18:37:45 paste me the APK file in query? 2017-09-25 18:38:04 wait.. nvm, i'll just test with vanilla zabbix 2017-09-25 18:38:07 <_ikke_> danieli: I have it working now 2017-09-25 18:38:12 ah okay 2017-09-25 18:39:23 <_ikke_> Does it make sense btw to combine provide with replaces? 2017-09-25 18:39:46 it does, so it both provides said packages, and replaces them if installed 2017-09-25 18:39:55 <_ikke_> right, that was my thinking 2017-09-25 18:40:14 <^7heo> danieli: you're tired too. 2017-09-25 18:40:21 <^7heo> <@_ikke_> When I was testing it in a shell, I did use {} 2017-09-25 18:40:25 <^7heo> danieli> strange 2017-09-25 18:40:28 <^7heo> Nope, expected. 2017-09-25 18:40:33 <_ikke_> http://alpine.ikke.info/zabbix/x86_64/ 2017-09-25 18:40:36 <^7heo> ${} is the right syntax. 2017-09-25 18:40:49 <^7heo> $() is what _ikke_ incorrectly used in his APKBUILD file. 2017-09-25 18:40:53 i was talking about brackets, i thought $ was in the context 2017-09-25 18:41:03 <^7heo> ah ok 2017-09-25 18:41:08 <^7heo> nevermind then 2017-09-25 18:43:26 airplane tickets in norway are ridiculous, holy shit 2017-09-25 18:43:27 the prices 2017-09-25 18:47:15 <_ikke_> hmm, odd, when installing alpine-zabbix, it does not replace / conflict with plain zabbix 2017-09-25 18:47:21 strange 2017-09-25 18:47:23 oh 2017-09-25 18:47:32 wait, if it has a "replaces", it should replace 2017-09-25 18:47:59 <_ikke_> ah, d'oh 2017-09-25 18:48:37 <_ikke_> reference calls it replcaes 2017-09-25 18:48:39 <_ikke_> replaces 2017-09-25 18:49:00 <_ikke_> https://wiki.alpinelinux.org/wiki/APKBUILD_Reference#replaces 2017-09-25 18:49:04 what did you call it? 2017-09-25 18:49:16 <_ikke_> http://tpaste.us/mMDv 2017-09-25 18:49:22 <_ikke_> this is from the .PKGINFO 2017-09-25 18:52:06 hmm 2017-09-25 18:52:15 i have no freaking idea LOL 2017-09-25 18:52:23 might be that replaces requires a version and pkgrel, not sure 2017-09-25 18:58:11 <_ikke_> It does install the new packages, it just does not seem to remove that fact that the old package is installed 2017-09-25 19:35:15 <_ikke_> retry 5 (or something like that) 2017-09-25 19:45:28 <_ikke_> retry++ 2017-09-25 19:46:05 why isn't there an openjdk9 package? 2017-09-25 19:47:09 <_ikke_> danieli: because you haven't built it :P 2017-09-25 19:51:52 it's been out for a while 2017-09-25 19:51:54 i started an apkbuild for it 2017-09-25 19:52:00 ripping off arch's to some degree 2017-09-25 19:52:04 icedtea doesn't provide it 2017-09-25 19:56:32 <_ikke_> Ok, I think we should be able to install the alpine-zabbix* packages now 2017-09-25 19:56:40 did you figure it out? 2017-09-25 19:57:30 <_ikke_> It does install the packages, but the old packages keep the status installed to, and I cannot delete them 2017-09-25 19:57:35 <_ikke_> but not sure if that's intended or not 2017-09-25 19:57:58 <_ikke_> apk info -W shows the files belonging to the new package 2017-09-25 19:58:48 strange.. 2017-09-25 19:58:58 alpine-zabbix subpackages depend on alpine-zabbix and not zabbix, right? 2017-09-25 19:59:19 <_ikke_> danieli: let me check 2017-09-25 20:01:15 <_ikke_> It does seem so 2017-09-25 20:02:15 where are you testing this? 2017-09-25 20:03:31 <_ikke_> on a vim 2017-09-25 20:03:34 <_ikke_> vm* 2017-09-25 20:03:41 on a vim :') 2017-09-25 20:03:46 <_ikke_> hehe 2017-09-25 20:07:33 <_ikke_> http://tpaste.us/Rrb1 2017-09-25 20:12:53 yeah i see 2017-09-25 20:13:00 http://viz-js.com 2017-09-25 20:14:01 dependencies look correct 2017-09-25 20:20:53 <_ikke_> https://github.com/Ikke/aports/tree/alpine/zabbix 2017-09-25 20:21:07 <_ikke_> https://github.com/Ikke/aports/tree/alpine/zabbix/zabbix/zabbix 2017-09-25 20:21:22 alpine zabbix zabbix zabbix 2017-09-25 20:21:22 lol 2017-09-25 20:24:04 <_ikke_> hehe 2017-09-25 20:24:51 <_ikke_> I'll proceed in installing this then? 2017-09-25 20:26:13 sure, why not 2017-09-25 20:26:57 <_ikke_> I guess I'll first remove that zabbix-* packages, and then install these ones 2017-09-25 20:27:10 <_ikke_> zabbix should keep running in the mean time 2017-09-25 20:27:22 well, it'll keep running, but the webif will fail while the files are gone 2017-09-25 20:27:28 the data is in postgres anyway 2017-09-25 20:27:31 <_ikke_> yes 2017-09-25 20:27:45 <_ikke_> webif is not that much of a problem right now 2017-09-25 20:30:08 <_ikke_> Done 2017-09-25 20:39:31 checking which variants of the JVM to build... server 2017-09-25 20:39:31 configure: Unknown variant(s) specified: server 2017-09-25 20:39:31 configure: error: The available JVM variants are: server client minimal core zero zeroshark custom 2017-09-25 20:39:35 what even 2017-09-25 20:39:36 LOL 2017-09-25 20:44:57 <_ikke_> lol 2017-09-25 20:45:05 <_ikke_> server obviously != server 2017-09-25 20:46:19 obviously 2017-09-25 20:46:25 silly me 2017-09-25 21:15:37 why did java do this 2017-09-25 21:15:38 kill me 2017-09-25 21:21:58 danieli: [14:21] -- Permission Denied - You're not an IRC operator 2017-09-25 21:22:01 can't kill you, sorry 2017-09-25 21:22:06 Xe: dammit 2017-09-25 21:22:16 but freenode klined me for no reason earlier! 2017-09-25 21:25:30 maybe for hanging out on OFTC 2017-09-25 21:28:39 klining /0 smh 2017-09-25 22:30:39 http://jdk.java.net/9/ea 2017-09-25 22:30:40 i'm impressed 2017-09-25 22:30:41 LOL 2017-09-25 22:31:28 shame it's not openjdk 2017-09-25 22:45:36 danieli: wow 2017-09-25 22:45:55 openjdk9 is gonna be a bit more of a pain 2017-09-25 22:46:08 i sense it's gonna take lots of patches 2017-09-25 23:01:29 it is? 2017-09-25 23:29:06 clandmeter: they have a jdk build for musl/alpine, but not openjd 2017-09-25 23:29:06 k 2017-09-25 23:42:38 danieli: there's a branch called 'portola' with all the patches that are in that release 2017-09-25 23:42:51 foxkit: yeah, i've heard of it. i'll look for the patches 2017-09-25 23:42:56 danieli: there's an Oracle guy that comes around here every so often, michaelsomething is his nick, you can ask him more infos 2017-09-25 23:43:09 they also have a portola mailing list on jdk's infra, you can ask there maybe 2017-09-25 23:43:56 i found some messages from that list 2017-09-25 23:44:08 and yeah, he's a portola member 2017-09-25 23:45:34 s390x, neat! 2017-09-25 23:46:09 nvm, i read in the wrong place 2017-09-25 23:46:16 ACTION is very smart 2017-09-25 23:46:24 i can't sleep because of anxiety so i might as well work on this 2017-09-26 00:06:10 my understanding is that oracle is actually interested in openjdk9 on alpine (as well as supplying .apks) 2017-09-26 00:06:20 (for proprietary jdk9) 2017-09-26 00:08:20 proprietary, sure 2017-09-26 00:08:26 the openjdk portola project has some neat stuff in their hg 2017-09-26 00:08:53 but they are interested in openjdk9 on alpine, too. 2017-09-26 00:08:56 oracle already supplies jdk9 binaries for linux 2017-09-26 00:08:59 yes, portola is proof of that 2017-09-26 00:14:11 i can't get over this message 2017-09-26 00:14:12 configure: Unknown variant(s) specified: server 2017-09-26 00:14:12 configure: error: The available JVM variants are: server client minimal core zero zeroshark custom 2017-09-26 00:14:13 silly java 2017-09-26 01:07:07 does alpine have an "LTS" branch? 2017-09-26 01:10:52 not as far as I know 2017-09-26 01:20:03 the s390x builder still seems dead 2017-09-26 01:20:13 it's gonna have a lot of catching up to do if we even get it back 2017-09-26 02:01:33 are we holding back shorewall packages or does anitya just not flag them for some reason? 2017-09-26 03:39:16 So I'm getting this cryptic message from abuild: >>> ERROR: tinygles*: Has /home/... in rpath 2017-09-26 03:39:32 It's tripping up prepare_package, but I don't even define that in the APKBUILD 2017-09-26 03:41:51 Ah, it probably has something to do with me trying to create a symlink 2017-09-26 03:55:01 Actually, it's not the symlink. 2017-09-26 04:31:17 <_ikke_> craftyguy[m]: did you setup the prefix correctly? 2017-09-26 04:41:45 _ikke_: like for a 'make install'? I'm not doing that with this package, just a make, then manually installing files 2017-09-26 04:44:42 <_ikke_> craftyguy[m]: and what does your makefile do? 2017-09-26 04:44:58 <_ikke_> craftyguy[m]: It's normally part of ./configure 2017-09-26 04:57:41 Ah, this thing uses cmake, there's no configure. I literally have to do "cmake ." and "make". the interesting bits have to be 'installed' manually (e.g. with 'install') 2017-09-26 04:59:24 ikke: makefile produces one shared library, I'm then doing something like "install -Dm755 ${builddir}/lib.so ${pkgdir}/usr/lib/lib.so" in the package() method 2017-09-26 05:01:58 <_ikke_> craftyguy[m]: The rpath is set during linking 2017-09-26 05:02:29 Ah, I didn't know that. I'll take a closer look with that in mind! 2017-09-26 05:03:22 <_ikke_> https://stackoverflow.com/questions/6323603/ld-using-rpath-origin-inside-a-shared-library-recursive#6323946 2017-09-26 05:03:43 <_ikke_> algitbot: :* 2017-09-26 05:07:38 _ikke_: thanks! This one is entirely new to me, so I'll read up and try again 2017-09-26 05:23:30 The trick was adding this to the cmake call: -DCMAKE_SKIP_BUILD_RPATH=TRUE 2017-09-26 07:40:12 good morning 2017-09-26 07:41:00 anyone knows how to list packages installed with apk's virtual switch? 2017-09-26 07:41:11 fabled? 2017-09-26 08:24:01 clandmeter, apk info -R 2017-09-26 08:24:45 ah right. thanks. 2017-09-26 08:39:59 ncopa, should we fix abuild aport to override VERSION for the makefile so we get $pkgrel included? 2017-09-26 09:17:02 fabled: sounds like a good idea 2017-09-26 09:39:12 clandmeter, we have py-dialog and py3-dialog in different APKBUILDs 2017-09-26 09:39:31 ok 2017-09-26 09:39:34 i did that? 2017-09-26 09:39:38 what do you think if we merge it ? 2017-09-26 09:39:54 the source is the same? 2017-09-26 09:39:56 Dunno if you did 2017-09-26 09:40:10 no they are different packages 2017-09-26 09:40:22 but actually python version can be a variable 2017-09-26 09:40:45 http://downloads.sourceforge.net/project/pythondialog/pythondialog/${pkgver}/python2-pythondialog-${pkgver}.tar.bz2" 2017-09-26 09:40:46 source="http://downloads.sourceforge.net/project/pythondialog/pythondialog/${pkgver}/python3-pythondialog-${pkgver}.tar.bz2" 2017-09-26 09:43:25 hmm 2017-09-26 09:43:29 they dont have the same source 2017-09-26 09:43:43 what if one version is diff then the other? 2017-09-26 09:44:55 probably is different 2017-09-26 09:45:13 just wodering to have one APKBUILD rather than two(s) 2017-09-26 09:46:12 what will you do if one gets 3.4.1 and other is still behind? 2017-09-26 09:46:20 im not sure how they manage versioning. 2017-09-26 09:48:41 https://sourceforge.net/p/pythondialog/code/ci/master/tree/ 2017-09-26 11:17:01 looks like imagemagick modules are currently broken 2017-09-26 11:17:31 $ identify -list format 2>&1 | tpaste 2017-09-26 11:17:31 http://tpaste.us/ZVOz 2017-09-26 11:26:54 ok 2017-09-26 11:27:05 its it was the libtool update that broke it 2017-09-26 11:35:19 <^7heo> ncopa: do you know if musl has an equivalent to stdbuf? 2017-09-26 11:35:28 <^7heo> (trying to unbuffer tr(1)) 2017-09-26 11:36:10 musl has no cmd(1) at all 2017-09-26 11:36:14 only function(3) 2017-09-26 11:36:16 <^7heo> ok 2017-09-26 11:36:24 busybox may have it 2017-09-26 11:36:28 or install coreutils 2017-09-26 11:36:47 <^7heo> I asked on #musl if there was a way to control that buffering behavior via environment variables 2017-09-26 11:36:59 <^7heo> but nobody answered yet, so... I check what I can :) 2017-09-26 11:37:13 <^7heo> installing coreutils is a bit overkill for a simple one-time buffering problem. 2017-09-26 11:38:02 <^7heo> http://lists.gnu.org/archive/html/bug-coreutils/2008-11/msg00164.html 2017-09-26 11:38:03 <^7heo> :DDD 2017-09-26 11:38:12 <^7heo> that is such a hack. 2017-09-26 11:47:23 kaniini: i think we need to revert "main/libtool: disable -dlpreopen nonsense" commit 0ab24492aeafcf28bbe1be0ebac025d055e24c50 2017-09-26 11:47:30 it broke imagemagick 2017-09-26 11:47:40 see #7915 2017-09-26 12:26:05 ncopa: aha! it *was* libtool 2017-09-26 12:33:58 go ahead and revert 2017-09-26 12:34:24 what was it supposed to fix? 2017-09-26 12:38:06 <_ikke_> newapkbuild says it supports -C (CMake example), but the optparse does not accept it. 2017-09-26 12:39:01 i was so tired last night i accidentally did rm -rf ~/ when tab complete failed 2017-09-26 12:39:11 <_ikke_> auh 2017-09-26 12:39:12 it was in a container but still AAAARGH 2017-09-26 12:39:13 <_ikke_> auch 2017-09-26 12:39:24 that was the point when i just closed my laptop and realized i was way too tired to function 2017-09-26 12:39:49 i had just done like 12-13 commits on aports lol 2017-09-26 12:40:09 <_ikke_> https://www.sadtrombone.com/ 2017-09-26 12:40:30 I'll ask here as well, as the traffic on the main channel can hide things... any suggestions for just fetching packages to upgrade before actually upgrading? 2017-09-26 12:40:38 it lags in chromium, bleh 2017-09-26 12:41:28 TBB: rephrase? 2017-09-26 12:45:38 I have some systems behind ultra slow links, and the usage of the systems can be short enough that upgrading on the background might never finish 2017-09-26 12:46:32 i've had systems like that on oil rigs and similar, we usually brought the package files in on a physical medium 2017-09-26 12:47:07 we were considering that solution too, but my client has basically said no way 2017-09-26 12:47:16 so, what exactly do you want to do? 2017-09-26 12:47:21 "fetch packages before actually upgrading"? 2017-09-26 12:48:32 make sure the packages required to upgrade are downloaded and cached locally first, in several sessions if necessary, and only then the actual upgrade can be run 2017-09-26 12:50:01 the project I work in is a collection of really weird requirements :/ 2017-09-26 12:52:18 I guess having a local directory as a package cache and stretching an rsync operation across as many uses of the system as it takes and then flagging an upgrade as OK would do the trick 2017-09-26 12:52:32 you can always rsync the whole ./apks tree 2017-09-26 12:52:37 create your own mirror 2017-09-26 12:53:21 I could do that too but that would mirror the entire tree, the cost of initial sync would be prohibitive 2017-09-26 12:54:12 so at least I have to use two repos, one of which is for these 'background updates' 2017-09-26 12:54:48 s/have to/could/ 2017-09-26 14:04:12 I'm working on packaging a replacement for libGL, is there a way to tell apk to "remove mesa-gl even though other things depend on it, trust me, I know what I am doing" ? 2017-09-26 14:04:35 the -f option to apk isn't very forceful.. 2017-09-26 14:05:44 <^7heo> fabled: ^ 2017-09-26 14:07:05 craftyguy[m], not really. best option is likely to create dummy package, or perhaps "apk add -t mesa-gl" 2017-09-26 14:07:17 but there might be also so: dependencies 2017-09-26 14:07:23 so it's not going to be simple 2017-09-26 14:09:15 fabled: thanks, I'll play around with that a bit! 2017-09-26 14:12:04 i'm bumping me some package versions 2017-09-26 14:44:39 looks like sourceforge is down 2017-09-26 14:44:39 partially 2017-09-26 14:44:48 zbar.sourceforge.net 2017-09-26 14:45:02 The sourceforge.net website is temporarily in static offline mode. 2017-09-26 14:45:03 Only a very limited set of project pages are available until the main website returns to service. 2017-09-26 14:45:13 twitter: "SourceForge is experiencing issues. We are investigating." 2017-09-26 14:45:17 clandmeter: emby may need an update, but i am not able to build new version 2017-09-26 14:45:27 danieli: im gonna push the imagemagick update 2017-09-26 14:45:33 + a bunch of rebuilds 2017-09-26 14:45:42 emby and zbar are left 2017-09-26 14:45:48 and need rebuild 2017-09-26 14:45:53 sounds great 2017-09-26 14:46:15 but im pushing anyways 2017-09-26 14:46:18 did it work when you reverted the libtool -dlpreopen stuff? 2017-09-26 14:46:22 yes 2017-09-26 14:46:26 nice 2017-09-26 14:46:39 im updating mono too while at it 2017-09-26 14:47:25 clandmeter: in case your like to look at emby: http://tpaste.us/0Ndd 2017-09-26 14:49:01 + local libMagicWand="$(ls /usr/lib/libMagickWand-7.Q16HDRI.so.? | cut -d / -f4)" 2017-09-26 14:49:14 thats the most important thing for making it build with new imagemagick 2017-09-26 14:49:50 i already spent more time that i can afford on that imagemagick issue 2017-09-26 14:52:19 I'm building custom packages in a private repo. At the end of the build process, the index is (re)generated. It's not clear to me if I can move the newly built package into the larger repo and append the existing index with the newly generated. 2017-09-26 14:52:29 Is that possible? Has anyone else run into that? 2017-09-26 14:53:19 use the same keys 2017-09-26 14:54:40 should be possible 2017-09-26 14:54:48 the apkindex will be signed with your key though 2017-09-26 14:55:10 which is ok i suppose 2017-09-26 14:58:31 Yeah, the same keys will be used; That's not an issue. The larger question remains: can I concatinate two indexes together and have it work? 2017-09-26 14:59:07 should be possible if you merge them properly 2017-09-26 14:59:17 you need to deal with overlapping packages 2017-09-26 14:59:18 if any 2017-09-26 15:02:02 you can give the previous index to the index generator; wonder if it can handle two indices 2017-09-26 15:03:12 okay, my two repository approach didn't quite work. enabling cache and then doing cache sync using just the updates repo doesn't seem to download anything 2017-09-26 15:03:29 (not related to AG's problem) 2017-09-26 15:05:03 apk index --index APKINDEX.tar.gz *.apk 2017-09-26 15:05:23 apk index --index APKINDEX.tar.gz --output APKINDEX.tar.gz.new *.apk 2017-09-26 15:05:28 or similar 2017-09-26 15:05:33 then you need to sign the new 2017-09-26 15:12:01 tmh1999: can you please try build pixman and see if tests fails for you on s390x 2017-09-26 15:12:08 they block the s390x build server 2017-09-26 15:12:32 1520 buildoze 58173: /home/buildozer/aports/main/pixman/src/pixman-0.34.0/test/.libs/lt-region-contains-test 2017-09-26 15:12:38 ncopa : it tried last weekend and it also has test problem on my side 2017-09-26 15:12:39 1670 buildoze 58517: /home/buildozer/aports/main/pixman/src/pixman-0.34.0/test/.libs/lt-affine-test 2017-09-26 15:12:49 still figuring out 2017-09-26 15:13:22 the test doesn't show and bug, just hanging 2017-09-26 15:13:27 *any 2017-09-26 15:13:38 doesnt crash yes 2017-09-26 15:13:53 but hangs = bug in my book :) 2017-09-26 15:14:19 agreed. im on the run. be back later in 12 hours 2017-09-26 15:14:24 ok 2017-09-26 15:15:19 it's never a bug, it's a feature! 2017-09-26 15:15:25 you have wrong books. :) 2017-09-26 15:35:59 ncopa: FWIW, that didn't work. However, I just extracted the index, concatinated them together, then re-archived 2017-09-26 15:36:06 easy enough. thx for the help 2017-09-26 16:57:21 https://github.com/alpinelinux/aports/pull/2390 can we merge this one? it's a security update for weechat. /cc fabled clandmeter 2017-09-26 17:01:32 would be nice to see it also in 3.6 2017-09-26 17:04:05 <_ikke_> I think this would warant an update in 3.6 too 2017-09-26 17:08:18 well, many people runs IRC client on a server that's not edge for sure ;) 2017-09-26 17:17:07 clandmeter: thanks 2017-09-26 17:17:20 thank you! 2017-09-26 18:42:37 guile build on x86 segfaults 2017-09-26 18:43:14 http://tpaste.us/NRoP 2017-09-26 18:43:28 in pthread_mutex_unlock 2017-09-26 18:44:51 dunno how to debug it 2017-09-26 18:47:48 ncopa: it segfaults because the mutex has already been destroyed i believe 2017-09-26 18:48:11 that sounds likely 2017-09-26 18:48:14 sounds plausible 2017-09-26 18:48:31 so it double destroy the mutex? 2017-09-26 18:48:39 why does it not happen on x86_64? 2017-09-26 18:48:42 is it a race? 2017-09-26 18:49:14 it smells race 2017-09-26 18:49:33 i was able to build it from cmdline 2017-09-26 18:49:45 probs race 2017-09-26 19:03:55 if (sizeof (unsigned long) == 8) 2017-09-26 19:03:55 ret = (((unsigned long) c) << 32) | b; 2017-09-26 19:03:56 else 2017-09-26 19:04:07 brilliant, isnt it? 2017-09-26 19:04:30 hash.c:214:32: error: left shift count >= width of type [-Werror=shift-count-ove 2017-09-26 19:04:30 rflow] 2017-09-26 19:04:30 ret = (((unsigned long) c) << 32) | b; 2017-09-26 19:04:30 ^~ 2017-09-26 19:15:50 hum 2017-09-26 19:16:01 i wonder if its the gc that makes somethine recursively 2017-09-26 19:19:02 http://tpaste.us/xWwo 2017-09-26 20:04:59 ncopa, is grsec still an issue with hyperv? 2017-09-26 23:25:32 git config --global alias.git '!exec git' 2017-09-26 23:58:53 is anyone working on llvm5 right know? 2017-09-26 23:59:15 because I am 2017-09-27 00:00:35 ncopa: tmh1999: the s390x builder still looks stuck 2017-09-27 00:01:05 danieli : i ack that. I am working tonight 2017-09-27 00:01:26 working on it tonight 2017-09-27 00:03:00 all right tmh1999 :) 2017-09-27 00:03:15 it's a checksum that's failing, it doesn't really make sense 2017-09-27 00:03:18 what are you testing? 2017-09-27 00:04:40 also, laz-perf: The following tests FAILED: 2017-09-27 00:04:40 1 - io_tests (SEGFAULT) 2017-09-27 00:10:35 for libpng, my download gets a different checksum than the one in the last upgrade commit (https://git.alpinelinux.org/cgit/aports/commit/?id=fff6218a54f9b2c313059b24fae17df3acc8d015) 2017-09-27 00:29:00 ^7heo: mind if I take over maintaining gogs? 2017-09-27 00:29:47 seems fairly dead 2017-09-27 00:29:53 s/dead/stale/ 2017-09-27 00:33:49 <^7heo> danieli: if you want 2017-09-27 00:34:21 <^7heo> danieli: I would update it eventually 2017-09-27 00:34:47 it doesn't really matter if you plan to keep it up to date 2017-09-27 00:34:50 i'm updating it now though 2017-09-27 00:35:16 <^7heo> the current package has issues with upgradjng 2017-09-27 00:35:35 <^7heo> upgrading* 2017-09-27 00:35:38 it built fine for me now 2017-09-27 00:35:48 <^7heo> sure 2017-09-27 00:36:16 i removed the patch and glide install 2017-09-27 00:36:36 <^7heo> "it works on my machine" is not a good attitude as a maintainer 2017-09-27 00:38:03 i'm gonna test it thoroughly, don't worry - i think gogs quit using glide 2017-09-27 00:38:03 <^7heo> please do not take over if it's to do trivial stuff that would take me 2 minutes 2017-09-27 00:38:22 <^7heo> I abumped it fine 2017-09-27 00:38:35 <^7heo> (locally) 2017-09-27 00:38:58 calm down, it doesn't really matter who does it 2017-09-27 00:39:09 <^7heo> there's a reason it's I waited 2017-09-27 00:39:16 <^7heo> < danieli> calm down, it doesn't really matter who does it 2017-09-27 00:39:25 <^7heo> precisely 2017-09-27 00:39:33 are you gonna submit a patch, or should i? 2017-09-27 00:39:51 <^7heo> It's HOW it's done that matrers 2017-09-27 00:40:01 <^7heo> matters* 2017-09-27 00:40:14 how come you waited with updating the package, btw? 2017-09-27 00:40:18 <^7heo> so as I said 2017-09-27 00:40:42 <^7heo> < ^7heo> the current package has issues with upgrading 2017-09-27 00:41:00 <^7heo> that's why 2017-09-27 00:41:06 define 'current package' and 'upgrading' 2017-09-27 00:41:24 <^7heo> cuurent version 2017-09-27 00:41:38 the one in aports at the moment? 2017-09-27 00:41:46 <^7heo> yes 2017-09-27 00:41:54 okay, define 'upgrade'? 2017-09-27 00:42:03 <^7heo> apk add -u 2017-09-27 00:42:09 aha 2017-09-27 00:42:19 <^7heo> ? 2017-09-27 00:44:15 <^7heo> /query danieli 2017-09-27 00:47:42 <^7heo> spacefail 2017-09-27 00:48:23 spaced out 2017-09-27 01:22:35 could anyone help me with getting dropbear to use rsa keys while not allowing password login? 2017-09-27 01:23:58 I have tried using "dropbear -s -g" but it seems to not work. I have successfully imported my public rsa key to the server. 2017-09-27 01:25:55 I have even uncommented some lines in sshd_config 2017-09-27 01:26:06 m4chm4n: any particular reason you want dropbear over openssh? 2017-09-27 01:27:02 lightweight and it is what is working for me at the moment. 2017-09-27 01:27:36 I was having trouble installing and running openddh 2017-09-27 01:27:41 openssh* 2017-09-27 01:27:44 google says https://wiki.openwrt.org/doc/howto/dropbear.public-key.auth#disable_password_login 2017-09-27 01:28:04 oh that's openwrt 2017-09-27 01:28:27 yea everything worked up till the uci part 2017-09-27 01:29:03 where it was changing the authorizations 2017-09-27 01:50:27 well I disabled password logins 2017-09-27 02:13:07 well, Xe, I figured it out.... 2017-09-27 04:14:23 If I submit a patch to the aports mailing list to update a package I am not the 'maintainer' for, should I trust that the actual maintainer will see it, or should I explicitly cc them? 2017-09-27 04:15:05 It's too late now for the SDL2 update patch I just mailed off, but I just thought "hmm maybe I should have cc'd the maintainer" 2017-09-27 05:33:06 hmmh... upstream modified libpng source tarballs? 2017-09-27 11:38:28 god damn it freenode 2017-09-27 11:38:31 was there a split or something? 2017-09-27 12:08:07 <^7heo> danieli: rethorical question I assume? 2017-09-27 12:20:48 A split? on freenode? YOU DON'T SAY 2017-09-27 12:43:16 I asked because I wasn't sure what it was lol 2017-09-27 12:43:24 it's fairly obvious I guess 2017-09-27 14:03:32 i added a tmate aport, tmate is actually pretty damn handy 2017-09-27 14:03:40 i came across a bug with it in alpine on docker though 2017-09-27 14:03:46 it tries to read /dev/pts/1 and fails 2017-09-27 14:04:01 ncopa: i'm not too familiar with that - how does it work? 2017-09-27 14:18:33 danieli: does that container have devpts mounted? 2017-09-27 15:29:30 Xe: i knew i forgot something 2017-09-27 15:44:19 s390x builder so dead 2017-09-27 15:57:54 /bin/bash: line 6: 30371 Segmentation fault GUILE_AUTO_COMPILE=0 ../meta/build-env guild compile --target="s390x-ibm-linux-musl" -O1 -L "/home/buildozer/aports/main/guile/src/guile-2.2.2/module" -L "/home/buildozer/aports/main/guile/src/guile-2.2.2/guile-readline" -o "srfi/srfi-1.go" "../module/srfi/srfi-1.scm" 2017-09-27 15:57:55 that's cute 2017-09-27 16:42:21 danieli: i think it is the same pthread_mutex_destroy race as discussed yesterday 2017-09-27 17:46:51 kaniini: i would assume so 2017-09-27 17:46:56 it keeps s390x builder dead 2017-09-27 19:08:17 does weechat die for anyone else with SIGILL in libnettle.so.6 on arm for anyone else? 2017-09-27 19:08:32 <_ikke_> No, but I'm not running it on alpine though 2017-09-27 19:10:26 mitchty: strace it? 2017-09-27 19:12:09 i have to look at arm abi and assembly again, been a while but looks like r5 is wrong here assuming i remember the register layout correct https://gist.github.com/mitchty/1556ea6e3f61cd5d42055b4b532f4dc2 2017-09-27 19:12:19 in either case i'll debug later 2017-09-27 19:22:16 mitchty: were you using any third party extensions? 2017-09-27 19:23:20 Xe: i just setup sasl to connect to freenode basically is all, this is the first time i've tried using weechat 2017-09-27 19:23:44 mitchty: so you were connected via ssl? 2017-09-27 19:24:08 i mention this because `salsa20` is related to SSL crypto 2017-09-27 19:24:29 trying to yep, should it try changing the cipher lists weechat uses? 2017-09-27 19:24:29 mitchty: i'd suggest running the test suite for GNUTLS on that machine if you can 2017-09-27 19:25:02 i'm personally very hesitant to mess with cipher lists 2017-09-27 19:25:08 especially for things like IRC clients 2017-09-27 19:26:10 sure, i'm just curious if i might be able to select a different cipher to work around it temporarily while I figure the real problem out 2017-09-27 21:37:28 does armhf presume neon extensions are available? 2017-09-27 21:43:59 cause it looks like the sigill here is neon asm in libnettle.so.6, this armhf box doesn't have any neon support 2017-09-27 21:44:45 It probably shouldn't assume that 2017-09-27 21:46:57 https://gist.github.com/mitchty/16bd167bb228ccdf79b43944eb4e94f8 as a note, those q registers are most definitely not going to work on this system :) 2017-09-27 21:53:03 looks like if autoconf detects neon that gets built into libnettle by default, not run time selectable in that way, have to pass --disable-arm-neon otherwise whatever the build system has will influence the result 2017-09-27 21:54:12 https://git.lysator.liu.se/nettle/nettle/commit/ade7779c98a5426c7d86c8a01bbd7ad65980c9b9 2017-09-27 22:11:58 yep, compiling libnettle with --disable-arm-neon fixed it 2017-09-27 22:16:21 I wonder if there should be a subpackage (e.g. -noneon) or something, since neon seems to help quite a bit with performance in some apps if the cpu supports it. disabling use of those extensions for all armhf processors in alpine seems a bit heavy-hammer-ish 2017-09-28 00:45:43 well i'm not sure how it should be fixed, maybe an install script that looks for neon support and installs the non neon version? 2017-09-28 00:47:02 mitchty: I would say it should be a split package, but I don't know really 2017-09-28 00:47:43 foxkit: i'm good for now, but i'm going to guess this is a more general problem than just this single package 2017-09-28 01:06:55 libnih segfaults during tests unless it is linked with gnu -lintl and --enable-nls is passed while testing something completely unrelated to translation :/ 2017-09-28 01:07:10 and fails many other tests without it too 2017-09-28 02:55:06 https://github.com/alpinelinux/aports/pull/2393 patching and (especially) compiling llvm is pain 2017-09-28 05:56:53 any postmarket people around: wondering if you build -doc packages for KF5 and if so what are you using to provide the doc builder stuff 2017-09-28 06:49:53 moring 2017-09-28 06:50:03 i'm thinking on adding the 'build flavors' to abuild or lua-aports 2017-09-28 06:50:15 something similar to https://docs.travis-ci.com/user/customizing-the-build#Build-Matrix 2017-09-28 06:50:36 it would likely be easier to add it to lua-aports, so that it just executes abuild with different environment variables set 2017-09-28 06:51:00 and that would allow generating cross-compilers pretty much straight out of the box, by just setting CTARGET to different things 2017-09-28 06:52:24 same thing *might* be useful for different kernel configurations, e.g. packages built against different lua/python/php/whatever major version 2017-09-28 06:52:35 any thoughts? 2017-09-28 06:53:58 nice idea fabled 2017-09-28 06:54:15 it's been on todo list for a long enough time; and seems no one else has picked it up yet :) 2017-09-28 06:58:31 I'm looking forward for the minigw64 compiler 2017-09-28 06:59:22 https://bugs.alpinelinux.org/issues/5011 2017-09-28 07:00:07 we receieved also a possibile sponsorship for that 2017-09-28 07:02:29 https://github.com/jprjr/mingw-cross-make 2017-09-28 07:02:52 one thing is if different patches are needed depending on build flavor, dunno how to make patches conditional easily 2017-09-28 07:07:13 fabled, what about setting a "naming-standard" for patches 2017-09-28 07:07:14 ? 2017-09-28 07:07:25 like $ARCH-whatehver-name.patch 2017-09-28 07:08:12 abuild might be improved so that when there's a patch in source, if the naming is not respected, it fails 2017-09-28 07:10:03 but it would be flavor specific, not arch specific 2017-09-28 07:12:36 $ARCH-$FLAVOR ? 2017-09-28 07:21:28 morning 2017-09-28 07:21:40 im not sure i like the idea with different source patches for different flavors 2017-09-28 07:21:52 different source patches means different sources for same build 2017-09-28 07:22:28 conditional patches... it's a horror.. 2017-09-28 07:22:53 https://code.foxkit.us/adelie/aports/blob/master/main/coreutils/APKBUILD#L25-27 2017-09-28 07:23:08 I recommend not even trying it in alpine proper 2017-09-28 07:23:15 it is shameful that I have to do it at all :P 2017-09-28 09:06:26 anyone else has issues with gimp and jpeg's? 2017-09-28 09:21:56 clandmeter: seems to open here 2017-09-28 09:31:32 ncopa, displays ok? 2017-09-28 09:31:43 yes 2017-09-28 09:32:19 im getting this: http://i.imgur.com/trhkApK.png 2017-09-28 09:34:40 nice 2017-09-28 09:34:58 :3 2017-09-28 09:36:26 reminds me of stereograms, but the bend obviously makes it impossible 2017-09-28 10:35:30 danieli: re tmate: https://github.com/tmate-io/tmate/pull/117 2017-09-28 11:13:11 ncopa: great! 2017-09-28 11:22:58 nmeum, does your gimp work correct with jpg images? 2017-09-28 11:23:19 py3-pyo is build for arm, but it won't appear on the index. Anybody knows why? https://pkgs.alpinelinux.org/packages?name=py3-pyo 2017-09-28 11:24:41 clandmeter: nope, it doesn't. I don't use gimp very often therefore I never cared enough to fix it 2017-09-28 12:19:09 nmeum, thx will try to fix it. 2017-09-28 15:40:55 question, I got ghc 8.2.1 working for x86_64 but i'm still debugging a segfault on armhf and got pinged on the package being out of date, the pr for 8.2.1 has been open for a while but how is that kind of thing generally handled where one arch is fine but one is holding up the rest? 2017-09-28 15:41:52 my arm debugging skills aren't yet up to everything and while I'd love to use the newer ghc, debugging a lot of unknowns means it takes a while until I figure out whats broken 2017-09-28 15:52:17 anyway thought i'd ask, I'm still debugging it but i'm dealing with understanding more about the ghc rts right now and how it does code gen on arm 2017-09-28 15:57:54 Hi Alpine maintainers! I know y'all are swamped with a long backlog of package patches, but this is a friendly ping on two I submitted a week ago: tslib (https://lists.alpinelinux.org/alpine-aports/4907.html) and directfb (https://lists.alpinelinux.org/alpine-aports/4908.html). Note that the directfb package depends on the tslib package 2017-09-28 16:35:10 mitchty: i have just about the same technique on debugging software on unfamiliar-ish platforms 2017-09-28 16:35:13 can be a pain 2017-09-28 17:30:18 craftyguy: use github 2017-09-28 17:30:24 craftyguy: i have scripts for that 2017-09-28 17:30:56 ok, I will resubmit those patches in github 2017-09-28 17:31:30 should I just use github from now on, even though the Alpine wiki instructs us to use the mailing list? 2017-09-28 17:36:20 craftyguy: github will likely get more attention, most people hate using patchwork 2017-09-28 17:37:03 Ah I see. Ok, github it is! 2017-09-28 17:39:43 anyone have a pre-built EFI iso for alpine? 2017-09-28 18:13:17 gimp seems to have issues with color profiles 2017-09-28 18:16:30 how does alpine feel about a $pkgname-git existing for certain things? 2017-09-28 18:29:26 how do i get something build checked on, say, ppc? 2017-09-28 19:03:19 i believe the 'go' package was incorrectly flagged 2017-09-28 19:03:57 The reason I wanted our own git source server shadowing upstream, https://carlchenet.com/the-github-threat/ should be a bit more clear 2017-09-28 19:04:40 debian, fedora do them to some extent 2017-09-28 19:05:51 i have to agree 2017-09-28 19:05:55 putting all your eggs in one basket is bad 2017-09-28 19:06:00 for obvious reasons 2017-09-28 19:07:36 good thing is we are not huge as debian, and can afford at-least /main 2017-09-28 19:08:58 with git its even possible to create layered (unionfs) over read-only backups and active 2017-09-28 19:09:05 not that it may be used 2017-09-28 19:09:36 just trying to guard our interests 2017-09-28 19:13:34 one way to handle checksum is , http://irclogger.com/.alpine-devel/2014-11-06#1415268267 2017-09-28 19:14:22 problem is regeneration of archives cannot reliably give same checksum 2017-09-28 19:14:59 so need to checksum whats inside the pristine archive 2017-09-28 19:15:43 this is a practice that needs to be thought and adopted 2017-09-28 19:34:53 gtg 2017-09-28 20:53:01 https://jira.mariadb.org/browse/MDEV-9055 2017-09-28 20:53:06 mariadb got rid of libmysqlclient 2017-09-28 21:02:27 tmh1999: s390x seems stuck on lvm and guile 2017-09-28 21:03:05 fabled : after running make, does abuild do anything to built binaries besides getfattr, strip ? 2017-09-28 21:03:55 danieli : yeah I ack that. we have some problem with s390x atm 2017-09-28 21:04:06 yeah, i guessed as much 2017-09-28 21:04:07 any idea why? 2017-09-28 21:04:33 is the software the host or build system? 2017-09-28 21:04:39 would seem like it, that segfault seems strange 2017-09-28 21:05:03 danieli : I am trying to finalize the work on bootloader and disk (sys) installation on s390x so that we do not have to rely on running in containers. most of bugs in s390x now are due to running in container 2017-09-28 21:05:05 ncopa: ping, for reference - https://jira.mariadb.org/browse/MDEV-9055 mariadb got rid of libmysqlclient 2017-09-28 21:05:25 tmh1999: ah, i see. having issues getting the bare metal system set up? 2017-09-28 21:05:40 kinda 2017-09-28 21:08:07 how come? 2017-09-28 21:17:59 ncopa: i fixed up the ppc fix, removed the ucontext fix, and the install db patch still applies 2017-09-29 01:20:19 Is there any reason Travis in aports would fail here: alpine_run $USER .travis/build-pkgs 2017-09-29 01:20:43 There's literally no information in the travis output to help determine it was my package that caused it to fail, or something else 2017-09-29 01:21:07 craftyguy[m]: link to PR? 2017-09-29 01:22:13 Xe: https://github.com/alpinelinux/aports/pull/2403 2017-09-29 01:22:18 ty 2017-09-29 01:23:12 thank you for helping :) 2017-09-29 01:23:14 craftyguy[m]: https://travis-ci.org/alpinelinux/aports/builds/281119245#L656 2017-09-29 01:24:03 Ah, there's a thing that expands. Wow, I totally missed that very subtle arrow on the left 2017-09-29 01:24:13 yeah 2017-09-29 01:24:18 i did too at first 2017-09-29 01:24:35 hmm, well this package will always fail to build until PR #2404 is merged 2017-09-29 01:25:18 algitbot: go home, you're drunk 2017-09-29 04:28:18 are there any known issues with using `install_if` ? 2017-09-29 06:20:32 if I wrote something like `repoman` or `rpmlint` or whatever your favourite non-alpine distro has for checking large quantities of package recipes 2017-09-29 06:20:35 would that be accepted? 2017-09-29 06:20:49 the QA violations I keep finding are kinda shocking :/ 2017-09-29 06:20:55 main/libidn has no checksums. at all. 2017-09-29 06:38:45 craftyguy[m], if you bundle the commits travis will be happy 2017-09-29 06:41:25 foxkit, you mean in edge? i see checksums. 2017-09-29 06:42:23 sorry it was typo 2017-09-29 06:42:25 https://git.alpinelinux.org/cgit/aports/tree/main/libidl/APKBUILD 2017-09-29 06:42:27 libidl 2017-09-29 06:44:28 interesting, somebody already tried to fix it but broken it instead... 2017-09-29 07:29:35 foxkit: do you know how repoman or rpmlint work? are they trying to statically detect problems, or actually running package builds or binaries? 2017-09-29 07:30:00 i am not an alpine dev but I was thinking of using my shell parser to do something similar 2017-09-29 07:31:16 i noticed this thread: http://lists.alpinelinux.org/alpine-devel/5741.html with a somewhat scary sed script to "parse" APKBUILD 2017-09-29 07:31:16 https://github.com/tmpfile/table 2017-09-29 07:58:53 kaniini, what is the name of that musl compat library you recently made? 2017-09-29 08:09:41 gcompat 2017-09-29 08:14:08 thx 2017-09-29 08:21:22 kaniini: did you get libwidevine working with gcompat? 2017-09-29 08:58:20 ncopa: do you have a copy of libwidevine somewhere? I'm the principal author of gcompat 2017-09-29 08:59:07 some X things work, steam-updater and basic gl demos... but never tried something like widevine which I assume links to libva and such? 2017-09-29 09:00:05 foxkit, i just added it to aports 2017-09-29 09:00:08 in nonfree 2017-09-29 09:00:28 clandmeter: it's already in testing/ 2017-09-29 09:00:37 oh 2017-09-29 09:00:37 it is? 2017-09-29 09:00:47 you mean libwidevine? I thought you meant gcompat 2017-09-29 09:00:48 sorry 2017-09-29 09:00:58 np 2017-09-29 09:01:10 our chromium has some issue though 2017-09-29 09:01:29 I usually merge alpine aports into adelie on sunday or monday, so I will play with it then. 2017-09-29 09:01:31 i think it our apkbuild needs patching to include the wrapper 2017-09-29 09:01:50 well.. I will try anyway 2017-09-29 09:01:58 trying to finish plasma desktop integration :) 2017-09-29 09:02:13 nice 2017-09-29 09:02:22 are you going to push that too alpine as well? 2017-09-29 09:02:40 oilshell: repoman uses python to static read the ebuild files in gentoo tree and find common mistakes. rpmlint uses perl to do same with redhat rpms 2017-09-29 09:03:13 clandmeter: I'm not sure. I would like to collaborate with postmarketos people because I know they are doing KDE too. some of their apkbuilds are using beta/bleeding edge releases though, and mine are using the LTS versions 2017-09-29 09:03:20 so I don't want to disrupt their work 2017-09-29 09:03:40 if we can all three (alpine/postmarket/adelie) collaborate I think KDE desktop would be easy to achieve between the three :D 2017-09-29 09:04:03 sounds good to me 2017-09-29 09:04:21 at least a good alternative for xfce4 2017-09-29 09:04:23 all my musl patches have been accepted upstream and have landed in latest bugfix releases, so no patching has been necessary so far. but it is kind of heavy. I would like to try He Xi's gettext-tiny from musl ML. 2017-09-29 09:04:32 a proper maintained one that is. 2017-09-29 09:04:58 clandmeter: I do know I want to package lxqt for Alpine. it is basically xfce-like but qt instead of gtk. and maintained. 2017-09-29 09:05:32 the qt5 builds in community/ have been good but it is missing qt5-qtx11extras which both need for window manager stuff. so I am going to submit that this weekend, too. 2017-09-29 09:05:33 we had shots at others but they were never really maintained. 2017-09-29 09:06:12 be back in a moment, cat is mrowing for some reason (odd at 4am) 2017-09-29 09:06:46 you should feed it ones in a while, it helps :p 2017-09-29 09:21:39 actually he needed more water 2017-09-29 09:25:47 mrow 2017-09-29 09:26:08 now he figured out IRC... 2017-09-29 09:26:28 fun fact, my cat actually managed to type 'awk' at shell prompt when he walked on my keyboard. thought it was hilarious XD 2017-09-29 09:27:21 next time it will be `cat` :) 2017-09-29 09:27:25 LOL 2017-09-29 09:27:44 would be fun if he typed 'cat' 2017-09-29 09:27:44 narcissism 9000 2017-09-29 09:29:49 "pussy" would be more refined and showing higher intelligence 2017-09-29 09:30:29 "shutdown -p now pet me" would show even more intelligence 2017-09-29 09:30:31 i'm at oslo airport atm 2017-09-29 09:30:41 foxkit: rm -rf --no-preserve-root / would be even more 2017-09-29 09:30:53 oh, wait wait 2017-09-29 09:30:55 hey! he's needy, but not malicious ;) 2017-09-29 09:30:57 see if your cat can post both his private and public key 2017-09-29 09:32:01 cats don't use PGP, they use a more well-designed protocol, OSniff 2.0 2017-09-29 09:32:11 sniff snif 2017-09-29 09:32:15 snif snif* 2017-09-29 09:32:30 do builders have proper /etc/hosts entries for themselves? orbit2 test suite requires `hostname` to resolve. 2017-09-29 09:42:00 i cannot for the life of me make this golang package compile "bin" into $builddir, god dammit 2017-09-29 10:39:50 <_ikke_> cve involving ELF binaries built with PIE: https://access.redhat.com/security/vulnerabilities/3189592 2017-09-29 12:50:33 How do I submit a patch to abuild? Is alpine-devel the correct mailing list for that? 2017-09-29 13:07:12 zhuowei, yes or send a pr via gh. 2017-09-29 13:13:05 _ikke_: i read about that the other day 2017-09-29 13:13:11 old bug that apparently got "upgraded" to CVE 2017-09-29 13:15:57 ncopa: what are the primary TODOs for 3.7? 2017-09-29 13:48:09 htop 2017-09-29 13:56:44 consus: error: curses cannot draw to an IRC channel 2017-09-29 13:56:52 Ah dammit 2017-09-29 13:58:20 ll 2017-09-29 13:58:22 lol* 2017-09-29 13:58:51 <_ikke_> zhuowei: there is a specfic aports list 2017-09-29 14:01:50 _ikke_: does abuild do git? 2017-09-29 14:02:03 say there's a tag and not a release 2017-09-29 14:23:15 yes it does 2017-09-29 15:49:43 why is elixir built with one core only? 2017-09-29 16:10:06 <_ikke_> danieli: one thread you mean? 2017-09-29 16:10:15 <_ikke_> (ie, one job)? 2017-09-29 16:11:00 <_ikke_> Might be that multithreaded is giving issues 2017-09-29 16:11:02 _ikke_: -j1 2017-09-29 16:11:04 <_ikke_> right 2017-09-29 16:11:06 yeah, i guessed so 2017-09-29 16:11:08 but that shouldn't be 2017-09-29 16:11:18 it's elixir, it should compile fine with multi core processors 2017-09-29 16:11:49 <_ikke_> danieli: The makefile / build process also? 2017-09-29 16:12:54 <_ikke_> iirc, the Makefile needs explicit support for a jobserver 2017-09-29 16:25:24 <_ikke_> Quite interesting read: http://make.mad-scientist.net/papers/jobserver-implementation/ 2017-09-29 16:31:24 _ikke_ : what is the problem? many makefiles have bugs if you run them with -j 2017-09-29 16:31:56 that is, they are correct with -j1, but the bugs only show up with -j N where N > 1 2017-09-29 16:32:17 foxkit: thanks I just looked at repoman and rpmlint 2017-09-29 16:32:42 it looks like repoman parses metadata XML files in Python? I didn't see parts where it parsed the ebuild files themselves, which are bash scripts 2017-09-29 16:32:51 it looks like rpmlint is now written in Python: https://github.com/rpm-software-management/rpmlint 2017-09-29 16:33:19 anyway if you want to write a linter for Alpine, I believe you will run into the same problem of parsing shell scripts that I pointed out in that thread. (i.e. sed is a bad way to do it) 2017-09-29 16:33:37 thanks! 2017-09-29 16:34:01 I am thinking of adapting my shell parser for things like this: http://www.oilshell.org/blog/tags.html?tag=parsing-shell#parsing-shell 2017-09-29 17:09:12 <_ikke_> oilshell: I don't know, I was just responding to danieli 2017-09-29 17:09:34 <_ikke_> oilshell: I just assumed someone found a bug with N>1 and therefore limited it to 1 2017-09-29 17:21:59 _ikke_: OK yeah I should have directed it at danieli 2017-09-30 09:11:08 what happens when `make test` fetches 120 MB of crap? 2017-09-30 09:11:20 do we mark it !check b/c sandbox vio? 2017-09-30 09:24:07 <^7heo> I'd vote for dropping the package. 2017-09-30 09:24:15 <^7heo> and documenting why. 2017-09-30 09:26:05 wow, wtf, "Running 4280 test cases. Begin at 04:24:17" 2017-09-30 09:26:13 it actually fetched 790 MB 2017-09-30 09:26:15 of data 2017-09-30 09:26:22 ^7heo: I agree with droppin the package; it's a Google package 2017-09-30 09:26:24 libvpx 2017-09-30 09:26:31 but unfortunately firefox needs it to play webm 2017-09-30 09:26:44 <^7heo> the problem is that there's nowhere where the reasons why dropping packages are mentionned 2017-09-30 09:26:52 the git log 2017-09-30 09:26:57 <^7heo> yeah no 2017-09-30 09:27:15 <^7heo> does not work if you never commit it in the first place; because you found during tests that it was terribad 2017-09-30 09:27:44 <^7heo> for that to work alpine would need a 'refused' repo 2017-09-30 09:27:51 <^7heo> to put all the 'refused' packages 2017-09-30 09:28:35 <^7heo> even if only containing the package name as a folder and an `.empty` file to make git aware of the folder 2017-09-30 09:28:46 <^7heo> but that sounds likc a hack to me 2017-09-30 09:34:48 ^7heo: it's in main/ 2017-09-30 09:34:53 ^7heo: main/libvpx 2017-09-30 09:35:01 ^7heo: hence 'dropping' means removing from main/ 2017-09-30 09:35:05 <^7heo> ok 2017-09-30 09:35:06 with a git log 2017-09-30 09:35:11 <^7heo> yeah in that case 2017-09-30 09:35:18 <^7heo> but I mean, "generally" a procedure for that would be nice. 2017-09-30 09:35:31 refused/glibc 2017-09-30 09:35:44 <^7heo> huhuhuh 2017-09-30 09:35:44 <^7heo> nice. 2017-09-30 09:36:31 <^7heo> til someone does `git mv main main.bak; git mv refused main; git mv main.bak refused; commit -am 'screw you'; push` 2017-09-30 20:15:22 hello guys and gals!! I am having trouble with my /var/log/messages.... The time and date are far off the real date and time. and every entry has the same date and time stamp. 2017-09-30 20:32:36 m4chm4n: do you have a ntp client installed? 2017-09-30 20:44:16 Yes, it asked to install one when I ran setup-alpine 2017-09-30 20:44:30 I chose chrony 2017-09-30 20:51:56 what syslogd? 2017-09-30 20:54:30 <^7heo> danieli: I only touch syslog-ng 2017-09-30 20:54:35 <^7heo> danieli: I dunno for others 2017-09-30 20:54:43 <^7heo> danieli: but I had terrible experiences with rsyslog 2017-09-30 21:02:52 so is there a way to make the timestamps on my logs correct? they dont even change from the first entry to the last. 2017-09-30 21:32:00 ok thanks for the help..... Guess I will go hit some forums up to find a solution.... 2017-09-30 21:47:26 because he's definitely entitled to getting an instant solution on a Saturday night 2017-09-30 21:48:19 <^7heo> skarnet: you're not paid to do this? :) 2017-09-30 21:48:29 I wish 2017-09-30 21:48:38 ACTION pays skarnet 0.01 EUR 2017-09-30 21:48:44 you are now alpine support desk 2017-09-30 21:48:47 have fun 2017-09-30 21:48:49 :D 2017-09-30 21:48:53 ... I quit 2017-09-30 21:49:52 you do not get unemployment protection since you only worked for 4 seconds :( 2017-09-30 21:50:22 awwww 2017-09-30 21:51:34 Can an Alpine dev review PR 2403 & 2404 ? These were also submitted to the Alpine aports mailing list a week ago. Thanks :) 2017-09-30 22:26:07 https://github.com/alpinelinux/aports/pull/2404 looks good to me fwiw 2017-09-30 22:30:25 Xe: thanks! 2017-09-30 22:30:38 i don't have any official power though lol 2017-09-30 22:33:13 shrug