2022-11-01 07:03:21 so, this openssl issue only affects 3.0. alpine 3.16 uses 1.1. so we don't need to do much 2022-11-01 07:03:45 we dont really need to create an emergency release 2022-11-01 07:29:24 Correct, though I think an edge snapshot would be useful 2022-11-01 09:26:59 How do you deal with pyproject.toml projects again? 2022-11-01 09:29:30 https://wiki.alpinelinux.org/wiki/APKBUILD_examples:Python#Package_for_pyproject.toml_build 2022-11-01 09:30:06 I will pray this is up-to-date 2022-11-01 09:36:26 bratkartoffel: i think we need help with openjdk15/16 on ppc64le. openjdk16 built with openjdk15 before it was rebuilt with gcc12. not sure what is up this time 2022-11-01 10:22:40 if I want to fix a bug before Alpine 3.17 is released, how much time do I have to get it in? 2022-11-01 11:18:55 Newbyte: Somewhere between now and 2 weeks 2022-11-01 13:42:41 ncopa: so it's just ppc64le openjdk which makes trouble? 2022-11-01 13:43:35 i'd suggest to just leave it (>= 12) out for now so i can take some time to look into it 2022-11-01 13:44:10 openjdk11 is fine and the most widely used afaik, so if we have that in 3.17 then it fine for now 2022-11-01 13:47:57 bratkartoffel: I can provide a container again for you 2022-11-01 13:58:28 yes only ppc64le from what i know 2022-11-01 14:19:21 directfb cmake patch doesnt apply on the new sdl2 release 2022-11-01 14:19:26 yeet it/keep it? 2022-11-01 15:00:25 Hi, first time visitor to alpine IRC 2022-11-01 15:00:33 I'm trying to package a go module, as a static-linked exec. The command works fine in the shell, but under `abuild` creates a dynamic-linked exec. 2022-11-01 15:00:59 (and apols if this isn't the best place to ask the question – if there's a better forum, any guidance would be welcome!) 2022-11-01 15:01:30 Any thoughts on what's different under `abuild` that might make golang link dynamically instead of statically? 2022-11-01 15:12:56 Check /etc/abuild.conf, there are GOFLAGS defined there. 2022-11-01 15:13:52 manarth: maybe it helps to export CGO_ENABLED=0 2022-11-01 15:19:05 Thanks, that's one of the options I've tried 2022-11-01 15:19:16 At it's simplest, running this from a shell works: 2022-11-01 15:19:18 go build -ldflags '-linkmode external -extldflags="-static"' 2022-11-01 15:19:58 minica: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), statically linked, Go BuildID=9ox24QkCfaUxkJ0sTryL/EuyprQGgsZFbdryBLLRA/gnnSEcR47LZJL55sgjej/h5Qj81KE7LMQXZd6q1EM, not stripped 2022-11-01 15:20:45 And running the same command within APKBUILD (Alpine's package-management build tool): 2022-11-01 15:20:45 minica: ELF 64-bit LSB pie executable, x86-64, version 1 (SYSV), dynamically linked, interpreter /lib/ld-musl-x86_64.so.1, Go BuildID=XH7BDsXdgIdtYgu1b5Rk/vuPKj3NjM1GXHlXwoctA/ItfRWLb4yYjh7ckF69CW/CMoZLuybXba9mlT_QwXD, stripped 2022-11-01 15:25:08 Did you check the GOFLAGS that abuild sets? 2022-11-01 15:26:43 ikke: i think a container would help a lot, thanks 2022-11-01 15:27:21 Will provide you one in about half an hour. 2022-11-01 15:28:59 I compared env within abuild and within shell; nothing significant changed (just the work-dir) 2022-11-01 15:29:17 would it be set anywhere other than env? There's nothing I've seen in ~ 2022-11-01 15:29:33 This is a containerised vanilla alpine docker image 2022-11-01 15:35:00 haha 2022-11-01 15:35:03 I've figured it out 2022-11-01 15:35:07 (facepalm) 2022-11-01 15:35:25 what was it? 2022-11-01 15:35:33 in build(), I provided the correct params to properly build the package statically-linked. 2022-11-01 15:35:55 in package(), I invoked `GOBIN="$pkgdir/usr/bin" go install` 2022-11-01 15:36:25 expecting it to work in the same way that running `make INSTALL_ROOT="$pkgdir" install` would behave 2022-11-01 15:38:07 Heh 2022-11-01 15:38:09 I'm somewhat surprised it doesn't 2022-11-01 15:38:52 go install is meant to automatically fetch and build modules 2022-11-01 15:39:22 ah! 2022-11-01 15:39:23 yeah. A painfully long amount of debugging and trial and error :-D 2022-11-01 15:39:36 (I'm not much of a golang developer) 2022-11-01 15:39:38 This worked: 2022-11-01 15:39:41 build() { go build -ldflags '-linkmode external -extldflags="-static"' } package() { install -D -m 755 "$builddir"/minica \ "$pkgdir"/usr/bin/minica } 2022-11-01 15:39:45 well, now I too know that so you have probably saved future-me time too :P 2022-11-01 15:39:56 \o/ 2022-11-01 15:39:59 hooray 2022-11-01 15:44:34 openssl 3.0.7 is building here now 2022-11-01 15:44:40 does anybody know the CVE? 2022-11-01 15:48:12 Apparently not announced yet 2022-11-01 15:50:50 ncopa: where did you get 3.0.7 from? there is no such tag in their git repo nor release I see on their site 2022-11-01 15:51:24 Yeah, wondering the same 2022-11-01 15:51:39 from https://www.openssl.org/source/openssl-$pkgver.tar.gz where pkgver=3.0.7 2022-11-01 15:52:17 Just announced 2022-11-01 15:52:30 Though overloaded 2022-11-01 15:52:35 aha 2022-11-01 15:52:36 https://paste.xinu.at/abK6x3UhQ3S0tYHi8JT/ 2022-11-01 15:52:38 yeah it's quite slow 2022-11-01 15:52:42 apparently it's fucking nothing 2022-11-01 15:53:30 CVE-2022-3602 2022-11-01 15:54:15 A buffer overflow with a potential DoS or RCE 2022-11-01 15:55:29 CVE-2022-3786 2022-11-01 15:55:42 I was wondering why you posted buildah cve 2022-11-01 15:56:24 It's in the announcement 2022-11-01 15:57:20 Oh, that's an announcement from NixOS 2022-11-01 15:58:43 the tarball has CHANGES.md with CVE numbers 2022-11-01 15:58:51 Both CVEs are not listed yet on NVD 2022-11-01 15:59:27 CVE-2022-3602 is buildah 2022-11-01 15:59:39 oh wait 2022-11-01 15:59:42 it's 2021 2022-11-01 15:59:44 lol 2022-11-01 15:59:58 Yeah 2022-11-01 16:01:25 i should learn to read 2022-11-01 16:33:26 ikke: did the ghc test failure resolve itself? 2022-11-01 16:33:47 nmeum: I would not think so 2022-11-01 16:33:58 it should not be able to build without ghc-bootstrap installed 2022-11-01 16:34:37 did you find the time to extract the full build log for the test failure that you mentioned yesterday? 2022-11-01 16:35:03 nmeum: I linked you to it 2022-11-01 16:35:12 yea, but it just shows a dependency installation issue now :D 2022-11-01 16:35:28 oof :P 2022-11-01 16:35:36 https://build.alpinelinux.org/buildlogs/build-3-17-x86_64/community/ghc/ghc-9.0.2-r1.log 2022-11-01 16:35:41 Yeah, I se 2022-11-01 16:36:01 sorry was afk when you posted it yesterday 2022-11-01 16:36:14 hi 2022-11-01 16:36:41 can anyone see my messages 2022-11-01 16:37:00 yes, we can see your messages. hello 2022-11-01 16:37:08 oh hi 2022-11-01 16:37:12 im just testing this 2022-11-01 16:37:20 i never use before :) 2022-11-01 16:41:36 nmeum: sadly I moved the log file instead of copying, so it's overwritten 🤔 2022-11-01 16:42:01 :( 2022-11-01 16:42:10 unfourtunate 2022-11-01 16:42:47 is there a chance that you could run the build again? i would really like to see ghc in 3.17 2022-11-01 16:43:08 yes, certainly 2022-11-01 16:43:13 thanks 2022-11-01 16:43:47 unless we disable it, the build would not be completed until it passes ;) 2022-11-01 16:45:33 waiting for the 3-17-x86_64 builder to be idle 2022-11-01 16:45:49 not sure if that happens soon, though 2022-11-01 16:52:48 why *does* abuild depend on openssl? 2022-11-01 16:53:39 for package signatures via abuild-sign 2022-11-01 16:53:53 aha 2022-11-01 16:54:00 I'm surprised it doesn't have some standalone tool for that 2022-11-01 16:56:10 In general, the advise, especially cryptography related, is to use existing tools, not rolling your own 2022-11-01 16:57:29 yeah, absolutely, but there are existing tools that are basically just for signing/verification (like age) or I guess gpg 2022-11-01 16:58:02 age is/was too new 2022-11-01 16:58:39 that's fair 2022-11-01 16:59:55 abuild-sign was written 13 years ago 2022-11-01 17:00:33 openssl seems like a big dep to have but I guess especially 10+ years ago there was no other option 2022-11-01 17:02:03 I think openssl would be present anyway since you need tls support for apk anyway 2022-11-01 17:02:16 it fetches packages over https? 2022-11-01 17:02:20 It can, yes 2022-11-01 17:02:28 gotcha 2022-11-01 17:02:37 legit, thanks for entertaining my questions :) 2022-11-01 17:11:00 no problem 2022-11-01 17:12:23 nmeum: it just failed on ghc :-) 2022-11-01 17:13:15 the buildlog still shows a dependency error though 2022-11-01 17:14:52 yes, I just now started bootstrapping it again 2022-11-01 17:24:30 oopss, go 1.19.3 is a windows only fix 2022-11-01 17:24:47 heh 2022-11-01 17:24:53 what's the syntax again to NACK a cve? 2022-11-01 17:24:58 0: 2022-11-01 17:25:16 ie, fix it in version 0 2022-11-01 17:27:20 done, sorry for the noise 2022-11-01 18:01:18 nmeum: https://build.alpinelinux.org/buildlogs/build-3-17-x86_64/community/ghc/ghc-9.0.2-r1-bootstrap.log 2022-11-01 18:01:33 ikke: would there be interest in porting abuild-sign, apk, or whatever part of the Alpine infrastructure software, away from openssl (and preferably to bearssl)? 2022-11-01 18:01:50 skarnet: you' 2022-11-01 18:01:57 you'd have to ask fabled / ariadne 2022-11-01 18:02:40 well they just got pinged :) 2022-11-01 18:02:53 melange already does not use openssl :) 2022-11-01 18:03:11 WIth https://gitlab.alpinelinux.org/ayakael/aports/-/jobs/891938, I'm getting new 'found textrels` errors after packaging. I do not know what that textrels are 2022-11-01 18:04:04 Ariadne: yeah, I was asking about apk/abuild because I know you have chainguard stuff down :P 2022-11-01 18:05:14 skarnet: an abstraction layer for cryptographic ops would be welcome in apk, as would an abstraction layer to allow alternative backends other than libfetch. both of these could be justified as cutting the openssl dependency. 2022-11-01 18:05:34 iirc, there was already an MR open for that 2022-11-01 18:05:42 an MR open for what? 2022-11-01 18:05:51 not using ossl 2022-11-01 18:06:09 skarnet: there would be interest in you spending 10 years writing it and it not getting merged ;) 2022-11-01 18:06:21 more specifically, it was yours MR :P Ariadne 2022-11-01 18:06:28 psykose: that's exactly why I'm asking *early* 2022-11-01 18:06:33 https://gitlab.alpinelinux.org/alpine/apk-tools/-/merge_requests/59 2022-11-01 18:07:16 panekj: no, that was to port libfetch to use libtls. it is on hold because there is no point in moving forward on it when apk depends on libcrypto anyway. 2022-11-01 18:09:16 ayakael: text relocations, https://wiki.alpinelinux.org/wiki/APKBUILD_Reference 2022-11-01 18:09:16 skarnet: fwiw i would not be in favor of using bearssl at the moment. despite its memory safe architecture, it does not support TLS 1.3 which has been identified as a requirement for some sites using apk-tools. 2022-11-01 18:09:42 skarnet: also, we need a FIPS validation story for apk-tools 2022-11-01 18:10:01 o.O some sites enforce TLSv1.3 and don't accept TLSv1.2?? 2022-11-01 18:10:19 by "sites" i mean deployments of apk-tools distributions in corporate settings 2022-11-01 18:10:24 oh 2022-11-01 18:10:34 also, while we are ontopic of chainguard stuff, Ariadne what was the name of github org that has up-to-date OCI images 2022-11-01 18:10:47 doesn't surprise me, corporate always manage to come up with the dumbest possible decisions 2022-11-01 18:11:15 I remember it was "something-images" 2022-11-01 18:11:18 or something like that 2022-11-01 18:11:22 literally "chainguard-images" lol 2022-11-01 18:11:35 btw the fact that it's memory-safe is exactly *why* it doesn't support tlsv1.3... the 1.3 spec is flawed in that it's *impossible* to implement it in a memory-safe way 2022-11-01 18:11:48 skarnet: rustls supports it in a memory-safe way 2022-11-01 18:11:58 rustls doesn't malloc? 2022-11-01 18:12:02 Ariadne: you also had different gh org which didn't include chainguard name at all 2022-11-01 18:12:08 or maybe you renamed it 2022-11-01 18:12:15 panekj: politics happened 2022-11-01 18:12:50 because that's the point, you have to malloc, because you have to store the whole chain of certificates in memory since 1.3 allows it to be sent in the wrong order 2022-11-01 18:12:51 'sfine, I just couldn't remember the name and couldn't find it now :P 2022-11-01 18:13:19 up to 1.2 there's a requirement that the cert chain is sent in the proper order, so you can stream-process it 2022-11-01 18:13:32 1.3 does away with that and it's a terrible decision 2022-11-01 18:13:40 i mean, i agree 2022-11-01 18:13:51 but, alas, TLS 1.3 is a requirement 2022-11-01 18:14:05 my question is about rustls implementing it and being memory-safe 2022-11-01 18:14:11 because that doesn't sound compatible 2022-11-01 18:15:16 I for one am waiting for TLSv1.4 where the committee realizes it made a mistake and adds the requirement back 2022-11-01 18:18:56 ok well either way, it seems unlikely that the alpine security team would ACK switching the default from openssl to something which does not support the current TLS standard, and isn't FIPS validatable 2022-11-01 18:19:12 but from an apk-tools perspective, an abstraction would be welcome 2022-11-01 18:30:37 panekj: thanks! 2022-11-01 18:33:41 you cannot add textrels to this 2022-11-01 18:34:48 https://wiki.gentoo.org/wiki/Hardened/Textrels_Guide https://wiki.gentoo.org/wiki/Hardened/Introduction_to_Position_Independent_Code 2022-11-01 18:35:10 there's an exception for riscv64 because the textrels are bogus and checking what they are is actually benign due to toolchain issues 2022-11-01 18:35:12 but not for others 2022-11-01 18:35:18 you have to actually fix it, apologies 2022-11-01 18:35:35 Time to open up another issue... 2022-11-01 18:35:51 sometimes the issue is just not having an -fPIC somewhere in a c/++ compilation step 2022-11-01 18:35:56 sometimes it's bad custom assembly 2022-11-01 18:36:33 sometimes it's 'good' custom assembly where something like libx264 (iirc) intentionally has them where it's faster, and it's a choice to make if you want them or not 2022-11-01 18:36:39 usually it's just the first one though or some build issue 2022-11-01 18:38:06 Was this check newly introduced? 2022-11-01 18:38:22 it's very old 2022-11-01 18:38:29 no it's ancient 2022-11-01 18:38:31 okay good to know 2022-11-01 18:38:41 i'd guess something in your last dotnet7 changes (new version, new something) breaks it 2022-11-01 18:38:56 if something is mentioned in wiki, it's usually ancient 2022-11-01 18:40:38 Indeed, you guessed it, my favorite problem child haha 2022-11-01 18:42:25 if anything i'd really like if the riscv64 issue was fixed 2022-11-01 18:42:29 i never figured out why it happend 2022-11-01 18:42:48 iirc when i tried lld instead they went away so it would be a usual binutils issue, but maybe my memory sucks 2022-11-01 18:55:51 Ariadne, skarnet: apk side of things already has build time abstraction layer for crypto in master 2022-11-01 18:56:09 Libfetch is problematic 2022-11-01 18:56:12 yeah, it is just the libfetch part 2022-11-01 18:56:57 I've been wanting to get rid of it. But there's still so many other things to do first 2022-11-01 18:57:15 what does libfetch do? if it's make a http connection, I already have all the pieces to tlsify it 2022-11-01 18:57:33 other protocols like ftp may be harder to handle 2022-11-01 18:57:34 I'm hoping to get next APK spurt within this month. I got side tracked with other $ work 2022-11-01 18:58:05 Libfetch is http/https client. It has ftp also, but I'm not sure if it's a requirement anymore. 2022-11-01 19:37:37 we still have 11 packages in aports that use ftp as primary source 2022-11-01 19:37:52 :/ 2022-11-01 19:38:14 ptrc: that's not what libfetch is used for 2022-11-01 19:38:26 ohh, right 2022-11-01 19:38:34 that would be abuild fetching packages 2022-11-01 19:38:42 in aports 2022-11-01 19:38:47 i confused it with the weird aports curl wrapper 2022-11-01 19:38:56 s/aports/abuild/ 2022-11-01 19:38:57 ptrc meant to say: i confused it with the weird abuild curl wrapper 2022-11-01 19:39:46 yeah, that wrapper uses either wget or curl 2022-11-01 19:39:49 https://gitlab.alpinelinux.org/alpine/abuild/-/blob/master/abuild-fetch.c 2022-11-01 19:39:51 Was just verifying 2022-11-01 19:40:44 surprisingly, some mirrors do in fact still serve packages over ftp 2022-11-01 19:40:49 some are down, like ftp://mirror.ps.kz/alpine/ 2022-11-01 19:41:04 We are not monitoring ftp 2022-11-02 01:14:18 bl4ckb0ne: i did explicitly drop it indeed, i guess whatever uses that is going to have a fun time :) 2022-11-02 01:14:27 realistically i'd like for someone to look at removing directfb instead 2022-11-02 01:14:49 i think i was the one that originally complained 2022-11-02 01:15:11 that and some random emulator (yuzu i think) that failed otherwise 2022-11-02 01:15:16 iirc some stiff in pmos depends in the directfb backend 2022-11-02 01:15:26 i had to patch sauerbraten 2022-11-02 01:15:31 id be down to yeet it too 2022-11-02 01:15:33 ye 2022-11-02 01:15:37 so my thought is to.. fix that 2022-11-02 01:15:44 i've never even had a good experience on fb devices 2022-11-02 01:15:53 anything using drmkms or whatever instead is much, much better 2022-11-02 01:16:05 this is really an ancient decades old interface+library at this point 2022-11-02 01:16:27 in terms of aports nothing really makes use of this, so i guess it's just those few apps someone might use 2022-11-02 01:16:48 the sdl directfb backend is deprecated 2022-11-02 01:16:53 but they still dont yeet it 2022-11-02 01:17:10 well that's ok, doesn't matter what they yeet or don't i guess 2022-11-02 01:17:15 https://gitlab.com/postmarketOS/osk-sdl thats the project using it 2022-11-02 01:17:17 do you know what the comment in the apkbuild refers to for booting things in pmos 2022-11-02 01:17:18 ah 2022-11-02 01:18:17 this seems to already have !isDirectFB() checks? what stops it from using the other backends? 2022-11-02 01:19:55 ACTION shrugs 2022-11-02 01:19:55 ¯\_(ツ)_/¯ 2022-11-02 01:19:59 we could ping them 2022-11-02 01:20:08 or let them repackage sdl with directfb 2022-11-02 01:21:32 pmos repackaging never goes well 2022-11-02 01:21:36 asking would be preferred 2022-11-02 01:22:04 https://gitlab.freedesktop.org/wayland/weston/-/issues/581#note_1225248 hah 2022-11-02 01:23:00 > Seriously, dropping fbdev? 2022-11-02 01:23:54 d-d-d-drop directfb 2022-11-02 01:26:12 bl4ckb0ne: go open an issue $somewhere about it for me :) 2022-11-02 01:49:47 remind me about it tomorrow 2022-11-02 01:49:48 ill do 2022-11-02 01:50:10 i can probably drop it at the sdl level if osk-sdk doesnt use it anymore 2022-11-02 01:51:33 i'll do it since it also involves a few rebuilds and removing directfb itself 2022-11-02 01:51:36 just want to make sure it's clear 2022-11-02 01:52:33 i really dont mind doing it if it can clear you some time 2022-11-02 01:52:37 i deeply hate it 2022-11-02 01:53:38 it takes me 5 minutes :p what takes me forever is talking to people 2022-11-02 02:25:02 people are dumb, at leat you can delete software without going to prison 2022-11-02 02:30:52 hardly dumb 2022-11-02 02:30:53 just scary 2022-11-02 02:42:32 elly: how's chromium adventures going 2022-11-02 03:27:47 psykose: in general good, but I have not gotten time to try landing any patches upstream yet 2022-11-02 03:28:05 unfortunately I am a manager and not an engineer any more so my time to hack is very limited 2022-11-02 03:28:20 that's alright :) no rush or anything, just wondering if you needed me to unblock something or some help with something specific to here 2022-11-02 03:28:33 no, I think things are alright! 2022-11-02 03:28:38 good to know ^_^ 2022-11-02 03:30:54 it would be very cool if chromium didn't take so long to build 2022-11-02 03:38:49 thankfully just a few changes isn't that long if no gn stuff changed and thinlto is disabled 2022-11-02 03:39:26 and with stuff changed it's mostly the non-ccache'd parts that take a while (java, node, tsc, python scripts..) 2022-11-02 03:39:29 but yeah aside from that.. 2022-11-02 03:40:11 how does one do incremental development on something like this using abuild? 2022-11-02 03:40:43 youdon't or you can repeatedly run abuild build if you want 2022-11-02 03:41:15 if you want to actually get a full built package each time, abuild rootpkg makes one (but doesn't resign/regenerate the local index to make it installable.. funny), then you can rm -r pkg/ to redo it after rebuilding i guess 2022-11-02 03:41:25 but there's no issue just running things from the source tree for these kinds of changes 2022-11-02 03:41:43 yeah, I just need to make sure that I have all the prepare steps properly done beforehand &c 2022-11-02 03:42:02 abuild deps unpack prepare 2022-11-02 03:42:27 aha, thank you 2022-11-02 03:42:28 though the gn stuff is in build, so for this i guess.. run build and cancel it lol 2022-11-02 03:42:34 it's a tiny bit arcane but eh 2022-11-02 03:42:52 that's alright, I basically just need the ability to test changes without having to do a full rebuild every time 2022-11-02 03:43:10 abuild prepare applies all the patches, right? 2022-11-02 03:43:26 abuild prepare runs prepare(), default_prepare has some patch application 2022-11-02 03:43:30 cool 2022-11-02 03:44:00 rest of prepare is some copypasted devendoring 2022-11-02 03:44:17 and a fix or two 2022-11-02 03:45:48 time to abuild build and wait a bit :) 2022-11-02 03:47:25 what hardware are you building it on 2022-11-02 03:48:27 my framework laptop 2022-11-02 03:48:38 it is: not super fast :) 2022-11-02 03:48:55 I wish I was using my work workstation but alas 2022-11-02 03:50:33 aha 2022-11-02 03:50:38 yeah i'd give it quite a few hours 2022-11-02 03:50:49 and i do recommend adding use_thin_lto=false 2022-11-02 03:50:53 otherwise that final link is another hour 2022-11-02 03:52:35 yeah 2022-11-02 03:52:48 the temptation to simply land a bunch of changes upstream and then hope they work is strong :) 2022-11-02 03:53:45 bad idea :p 2022-11-02 04:29:36 blech, the test_fonts tarball is missing a required font called NotoSansCJK-VF.otf.ttc 2022-11-02 04:30:09 i'd assume that matters only for actual tests 2022-11-02 04:30:12 but also that thing is quite old 2022-11-02 04:30:18 i don't think anyone touched the commit 2022-11-02 04:30:21 unless it's always missing 2022-11-02 04:31:48 on 2022-04-26 upstream chromium changed the hash, so the hash has been wrong for a while I think 2022-11-02 04:31:58 it does only matter for building tests but I tripped over it trying to build base_unittests 2022-11-02 04:32:03 ddevault: himitsu-ssh now doesn't build either https://build.alpinelinux.org/buildlogs/build-3-17-x86_64/community/himitsu-ssh/himitsu-ssh-0.1-r0.log 2022-11-02 04:32:07 himitsu doesn't either actually 2022-11-02 04:32:23 next time could you rebuild everything along with hare when upgrading it if it's this volatile 2022-11-02 04:32:26 elly: ah, makes sense 2022-11-02 04:32:34 if you know what it's supposed to be you can change it 2022-11-02 04:34:47 yeah, I will update it to the proper hash 2022-11-02 04:36:33 it would be very cool if I could at least get base_unittests to build and run and re-enable the check step with them 2022-11-02 04:36:44 it would be 2022-11-02 04:36:48 sure, i can try it in parallel to you as well 2022-11-02 04:36:50 gimme the hash 2022-11-02 04:36:57 336e775eec536b2d785cc80eff6ac39051931286 is the new one 2022-11-02 04:37:03 thanks! 2022-11-02 04:37:04 source: https://source.chromium.org/chromium/chromium/src/+/main:third_party/test_fonts/test_fonts.tar.gz.sha1;l=1?q=third_party%2Ftest_fonts%2Ftest_fonts.tar.gz.sha&ss=chromium 2022-11-02 04:37:11 (that file is authoritative) 2022-11-02 04:39:13 i was using chromium.googlesource.com this whole time, nice to see there is an actually usable website instead for reading things 2022-11-02 04:39:57 heh :) 2022-11-02 04:40:43 :) 2022-11-02 04:40:48 ninja: file is missing and not created by any action: '../../content/test/data/web_ui_test.test-mojom' 2022-11-02 04:40:51 hmmmm 2022-11-02 04:41:26 i would guess the gn hacks in prepare either break that in part, or the release tarball is incomplete 2022-11-02 04:41:32 more leaning towards the latter 2022-11-02 04:42:06 the existing prepare hacks already have to touch some files that are missing like that 2022-11-02 04:42:36 so i'd guess the google tarball doesn't ship that 2022-11-02 04:42:55 what on earth, where did all the files in //content/test/data go 2022-11-02 04:45:18 something in here mangles it https://source.chromium.org/chromium/chromium/tools/build/+/main:recipes/recipe_modules/chromium/resources/export_tarball.py;l=14?q=export_tarball.py&ss=chromium i would guess 2022-11-02 04:45:32 yup, the tarball is incomplete 2022-11-02 04:45:55 aha: "it can also remove files which are not strictly required for the build" 2022-11-02 04:46:05 so I'm guessing building test targets like this is simply not supported if you're using the tarball 2022-11-02 04:46:41 i do have a snapshot overall that works for arbitrary gitrevs (and i do the same mangling) which i made since not many tarballs are published into the release branches, could make that work with tests, but it would be a little too much work to constantly need to do that for every single release 2022-11-02 04:46:43 weh 2022-11-02 04:46:49 i do that for electron already, same thing 2022-11-02 04:47:02 yup, it deliberately strips ~all the test data when packaging the source tree 2022-11-02 04:47:10 minus the hardrequired ones yep 2022-11-02 04:47:11 so I think a tarball based build will not ever have a working check target 2022-11-02 04:47:14 and still misses one given the touch 2022-11-02 04:47:14 mhm 2022-11-02 04:47:22 maybe it was recent since they used to work a year+ ago 2022-11-02 04:48:28 well, this is unfortunate 2022-11-02 04:49:05 I can check how big the tarballs get if we don't do this stripping but I have a feeling the answer is "very big" 2022-11-02 04:50:02 semi-informed guess: +20% for removing test stripping only 2022-11-02 04:50:49 it would take a few iterations to figure out what is needed for either just base_unittests or 'mostofeverything_tests' and playing with that exclude list 2022-11-02 04:50:57 unless the whole list is just removed of course 2022-11-02 04:51:15 +20% at worst* 2022-11-02 04:51:25 which is probably too high for whoever decides anyway, but have to measure 2022-11-02 05:02:57 oh gosh, I tried it on my workstation and the intermediate archive (before xz gets to it) is multiple gigabytes 2022-11-02 05:03:34 the .tar is like 14 yeah 2022-11-02 05:04:09 ..do chromium developers not realise how big it is, generally? :D 2022-11-02 05:04:14 yeah we don't 2022-11-02 05:04:56 didn't expect that, funnily. i guess when you work on it all the time in a specific area it's not as apparent 2022-11-02 05:05:25 oh! good news: https://commondatastorage.googleapis.com/chromium-browser-official/chromium-107.0.53 2022-11-02 05:05:28 04.87-testdata.tar.xz 2022-11-02 05:05:31 aha 2022-11-02 05:05:31 oops, that got split, but you get the idea 2022-11-02 05:05:41 that contains all the missing data files for tests 2022-11-02 05:05:48 fun fun 2022-11-02 05:05:50 it is, unfortunately, another 250MB 2022-11-02 05:06:26 so, 15% extra 2022-11-02 05:06:33 oh that's 250MB *compressed* 2022-11-02 05:06:38 aye 2022-11-02 05:07:03 so yeah it's not all that much really 2022-11-02 05:07:19 anyway, looks like you unpack that over the source tarball and you're maybe good to go 2022-11-02 05:07:42 and yeah we a) already have source checkouts, b) always do incremental builds, and c) use a compile farm so builds are very fast regardless 2022-11-02 05:07:51 also we all use component builds always to avoid the link step :P 2022-11-02 05:08:21 yep, fancy component build saves a ton of steps 2022-11-02 05:14:01 ninja: job failed: python3 ../../build/gn_run_binary.py generate_fontconfig_caches 2022-11-02 05:14:02 fun fun 2022-11-02 05:14:59 https://source.chromium.org/chromium/chromium/src/+/main:third_party/test_fonts/fontconfig/generate_fontconfig_caches.cc 2022-11-02 05:14:59 hah 2022-11-02 05:15:59 auto cache = fontconfig_caches + "/" + kCacheKey + "-le64.cache-7"; 2022-11-02 05:16:06 ..mine ends in -8 2022-11-02 05:16:15 let me guess, the new version of fontconfig changed that number :) 2022-11-02 05:17:10 rude 2022-11-02 05:17:26 https://gitlab.freedesktop.org/fontconfig/fontconfig/-/commit/d863f6778915f7dd224c98c814247ec292904e30 2022-11-02 05:17:28 yep 2022-11-02 05:17:50 that should be fixable upstream I guess in the test_fonts repo 2022-11-02 05:18:09 oh wow are you trying to do tests 2022-11-02 05:18:31 fun fact I just checked on my workstation and my chromium src checkout is 268GB 2022-11-02 05:18:41 sam_: it would be nice if at least *some* of them worked 2022-11-02 05:18:45 auto cache = fontconfig_caches + "/" + kCacheKey + "-le64.cache-" + FC_CACHE_VERSION_NUMBER; 2022-11-02 05:18:47 there, fixed 2022-11-02 05:18:50 feel free to send that 2022-11-02 05:18:57 defined in the included fontconfig.h 2022-11-02 05:19:06 btw, https://crbug.com/1380654 & its blocking bugs 2022-11-02 05:19:20 https://gitlab.freedesktop.org/fontconfig/fontconfig/-/issues/330 lol 2022-11-02 05:19:26 added to bookmarks 2022-11-02 05:19:54 nice 2022-11-02 05:20:32 ah, it's FC_CACHE_VERSION instead, a line below 2022-11-02 05:20:35 same thing 2022-11-02 05:23:55 first actual error, allocator_shim_default_dispatch_to_partition_alloc_unittest.cc has mallinfo() 2022-11-02 05:24:09 and allocator_shim_unittest.cc has pvalloc() 2022-11-02 05:24:10 more hacks for me 2022-11-02 05:25:03 there's an existing patch about mallinfo in the aports tree I think 2022-11-02 05:25:58 yeah, lots of places with that 2022-11-02 05:28:17 https://source.chromium.org/chromium/chromium/src/+/main:base/third_party/symbolize/README.chromium argh... "this patch should be upstreamed" to a package with a 2006 copyright date *also from google* 2022-11-02 05:28:49 ^_^ 2022-11-02 05:34:59 tests almost entirely built 2022-11-02 05:35:14 safe_sprintf.h:232:39: error: conversion from 'std::nullptr_t' to 'const internal::Arg' is ambiguous 2022-11-02 05:35:16 yes yes i get it 2022-11-02 05:36:02 a journey of a thousand kloc begins with a patch to a single line: https://chromium-review.googlesource.com/c/chromium/src/+/3997878 2022-11-02 05:38:10 browsing these sites with font-overrides disabled which give the symbols is hilarious sometimes 2022-11-02 05:38:16 https://img.ayaya.dev/UeghFurWZkV2 2022-11-02 05:39:00 lol yes 2022-11-02 05:39:08 try cs.chromium.org with no symbol glyphs 2022-11-02 05:39:10 it looks silly 2022-11-02 05:40:48 https://img.ayaya.dev/NaFLraLrDtC1 2022-11-02 05:41:09 yup 2022-11-02 05:41:17 folder folder folder folder folder folder 2022-11-02 05:41:51 well, alright, the first two chromium changes are up - they're both trivial but will allow for dropping two of the patches in aports once they roll to stable 2022-11-02 05:44:05 base_unittests almost finished, just one more thing to fix for linking 2022-11-02 05:44:12 ld.lld: error: undefined symbol: base::debug::StackTrace::OutputToStreamWithPrefix(std::ostream*, char const*) const 2022-11-02 05:44:13 yawn 2022-11-02 05:44:30 usual execinfo stuff i guess 2022-11-02 05:44:35 good stuff 2022-11-02 05:44:42 psykose: fyi, you can run abuild index after rootbld to get the packages signed and added to the index 2022-11-02 05:44:51 I mean rootpkg 2022-11-02 05:46:00 yep 2022-11-02 05:46:17 elly: https://img.ayaya.dev/Ka2TE0Tz0LgB changes so far 2022-11-02 05:46:28 fc-cache-version and musl-nullptr should be fine as is if you want to yoink more stuff 2022-11-02 05:48:40 ack 2022-11-02 05:48:51 the fc-cache-version one looks easy, that musl-nullptr one looks a little sketchy to me 2022-11-02 05:51:09 i can't tell you what glibc is doing there 2022-11-02 05:51:13 me either 2022-11-02 05:51:20 by standard it's https://en.cppreference.com/w/cpp/types/NULL 2022-11-02 05:51:24 so musl does, sec 2022-11-02 05:51:59 https://github.com/bminor/musl/blob/8f9259450aa43a6fd539e428e61e2961b725fbae/include/stdlib.h#L11 2022-11-02 05:52:10 since 1.2.3, 1.2.2 didn't have that part 2022-11-02 05:52:20 in glibc i can only find references to one of the other two 2022-11-02 05:52:50 as for why the code fails, shrug, some language semantics between nullptr and an actual value or alternatively the number, depends on what glibc is doing 2022-11-02 05:53:52 yeah, but the intent of the test is specifically to make sure that if you try to SafeSPrintf NULL in a couple of different formats you get the right results, so passing an explicit 0 in is testing something quite different 2022-11-02 05:54:12 I am not sure if I want to try to figure it out though, but I can just disable the test under not-glibc maybe :) 2022-11-02 05:54:18 (upstream) 2022-11-02 05:55:38 someone that simultaneously understands nullptr semantics and the SafeSPrintf soup simultaneously would be able to figure out the semantic difference 2022-11-02 05:55:45 Arg has constructors for const char* and char* 2022-11-02 05:56:17 ya, I don't understand what safesprintf is supposed to do internally but I think I know who would 2022-11-02 05:56:18 is constcharptrconstructor(std::nullptr_t) valid? is ((void*)0) valid instead? easy to test for this case i guess, gives the answer 2022-11-02 05:56:27 i bet the former is not 2022-11-02 05:56:32 so the error makes sense 2022-11-02 05:56:50 for pvalloc, I think the upstream code should not use pvalloc at all (it is obsolete) and should just use posix_memalign(), which would neatly make it work under musl also 2022-11-02 05:57:01 it already does use posix_memalign 2022-11-02 05:57:04 it uses everything at once 2022-11-02 05:57:13 the tests use every api imaginable in that memory file 2022-11-02 05:57:45 icky 2022-11-02 05:57:47 which makes sense as it's for the partition alloc to see if it implements the interfaces correctly i guess or something 2022-11-02 05:57:52 yeah 2022-11-02 05:57:57 hum 2022-11-02 05:58:15 but if chromium drops all pvalloc/valloc from the entire codebase, and the shim code, yes, it becomes possible to just remove 2022-11-02 05:59:13 even if we did there would be no way to guarantee system libs don't use it 2022-11-02 05:59:19 so I guess it is stuck there 2022-11-02 05:59:41 alright 2022-11-02 05:59:49 well, it's 11pm here, enough chromiuming for one night 2022-11-02 05:59:58 thank you for IRC-pairing about it 2022-11-02 06:00:10 o/ 2022-11-02 06:01:29 I think that with moderate effort it will be possible to obsolete around half the patches in the chromium aport, and possibly simplify the apkbuild a lot and get at least some tests running 2022-11-02 06:04:34 mhm :) 2022-11-02 06:04:37 thanks for your time 2022-11-02 06:04:45 it's a fun 7am for me 2022-11-02 06:05:43 I have odd feelings about spending a day doing paid work on chromium, then spending my evening doing unpaid work on it 2022-11-02 06:07:06 that one is much easier for me as everything is unpaid so i don't even think about it 2022-11-02 06:09:16 the joys and sorrows of having a lot of free time I guess 2022-11-02 06:12:39 https://img.ayaya.dev/HX0p9epD5LYA 2022-11-02 06:12:41 link success 2022-11-02 06:13:12 https://img.ayaya.dev/vxGtS9YdeZUL 2022-11-02 06:13:14 score! do they run? 2022-11-02 06:13:16 15 fails, 3 ceashes 2022-11-02 06:13:21 very fixable 2022-11-02 06:13:23 almost done then 2022-11-02 06:13:25 98% of the way there 2022-11-02 06:13:37 lol 2022-11-02 06:13:57 icu tests can be musl locale related (it has some very niche specifics) 2022-11-02 06:14:03 crashes are probably hmm 2022-11-02 06:14:29 eh, will see 2022-11-02 06:14:32 That ScopedFDOwnership test is probably related to scoped-file-no-close.patch 2022-11-02 06:14:34 stacktrace shouldn't even run 2022-11-02 06:14:37 that one is that, yeah 2022-11-02 06:14:45 and elf reader is probably some other niche thing 2022-11-02 06:14:48 failures are more interesting 2022-11-02 06:14:54 but it's nice to know almost everything works 2022-11-02 06:14:59 yeah :) 2022-11-02 06:15:12 i think there's a few more things to fix to get another 2 testsuites in (you mentioned another 2 i think that would be useful) 2022-11-02 06:16:40 unit_tests and components_unittests have a lot of the remaining value imo 2022-11-02 06:16:54 however unit_tests is likely to be a much bigger undertaking 2022-11-02 06:18:31 right 2022-11-02 06:18:34 will keep in mind 2022-11-02 06:19:08 gfx_unittests and net_unittests may also be interesting 2022-11-02 07:10:55 morning 2022-11-02 07:11:27 gnome-keyring-daemon ends up in a loop-forever when i log in. and conumes 100% of one cpu core 2022-11-02 07:11:43 i dont know why 2022-11-02 07:13:21 how do you run it 2022-11-02 07:15:00 https://gitlab.gnome.org/GNOME/gnome-keyring/-/issues/123 2022-11-02 07:15:00 ah 2022-11-02 07:15:02 it's this 2022-11-02 07:15:09 we do have that glib 2022-11-02 07:15:25 https://gitlab.gnome.org/GNOME/glib/-/commit/ca905744dffb844663b5bd6b42f33e2d44f9b4cd fix commit for glib 2022-11-02 07:15:29 go for it 2022-11-02 07:17:09 well, it comes from https://gitlab.gnome.org/GNOME/glib/-/merge_requests/3029 2022-11-02 07:17:10 shrug 2022-11-02 07:30:51 ikke: saved the ghc build log this time. will take a look later 2022-11-02 07:31:27 Thanks. I named it differently so that the normal process would not overwrite it. 2022-11-02 07:35:44 psykose: thank you! will fix it after breakfast 2022-11-02 07:36:17 tasty breakfast i hope :) 2022-11-02 07:52:52 psykose: ack 2022-11-02 09:48:30 ncopa: thanks for the fixes 2022-11-02 09:55:26 elly: gfx fails 397/5663, net fails https://img.ayaya.dev/ZlBrVKSa9jjE 2022-11-02 09:55:49 net fails are split between the vpython thing which has some override iirc and nss failing, which iirc fails on firefox too, i think it's silently broken for this cert usecase 2022-11-02 09:55:54 something i should probably write down somewhere 2022-11-02 09:56:08 but yeah looks quite good so far, just a few fixes to run net+base 2022-11-02 09:57:04 components/unit build like 40k so i'll see that another time just for reference 2022-11-02 09:57:04 psykose: do you mind take over chromium maintainership officially? You do the work there anyway 2022-11-02 09:57:33 sure, i'll grab it :) 2022-11-02 09:57:35 just a formality 2022-11-02 09:57:52 i will say i don't know what i'm doing though :p 2022-11-02 09:58:41 i guess i now maintain every web browser sans qtwebengine 2022-11-02 09:59:05 but that just has me rebase the chromium patches anyway for qt6 2022-11-02 09:59:49 done, didn't bump pkgrel to not waste a few hours on builders 2022-11-02 10:00:01 will be updated pkgrel on next release of anything there 2022-11-02 10:21:01 thank you. please feel free to poke me if you need help 2022-11-02 10:22:39 of course :) 2022-11-02 10:23:05 ah, right, we have to fix gdc 2022-11-02 10:23:11 it's the last real blocker iirc for 3.17 2022-11-02 10:24:13 i think the issue is that https://github.com/gcc-mirror/gcc/blob/16e2427f50c208dfe07d07f18009969502c25dc8/libphobos/libdruntime/core/sys/posix/sys/socket.d#L222 creates a reference to that symbol in some way 2022-11-02 10:24:35 for which glibc has a symbol https://codebrowser.dev/glibc/glibc/sysdeps/unix/sysv/linux/cmsg_nxthdr.c.html 2022-11-02 10:24:39 in musl this is just macros 2022-11-02 10:24:41 it doesn't exist 2022-11-02 10:24:57 so, ldd /usr/lib/libgphobos.so.3 2022-11-02 10:24:59 Error relocating /usr/lib/libgphobos.so.3: __cmsg_nxthdr: symbol not found 2022-11-02 10:25:08 this makes the ldc build fail 2022-11-02 10:25:16 not sure why this wasn't an issue on edge 2022-11-02 10:25:26 ok. I'll have a look at that 2022-11-02 10:25:30 or maybe it's only things that link gphobos so the existing ldc was fine 2022-11-02 10:25:54 ah, and !40967 is the only thing blocking moving openssl1.1 to community so we don't have to worry about it for 2 years 2022-11-02 10:26:13 that's all, everything else should be clear for 3.17 after ghc is resolved too 2022-11-02 10:26:25 aside from testing and fixing bugs that pop up of course 2022-11-02 10:36:08 technically, the problem is under musl, not glic: https://github.com/gcc-mirror/gcc/blob/16e2427f50c208dfe07d07f18009969502c25dc8/libphobos/libdruntime/core/sys/posix/sys/socket.d#L2040 2022-11-02 10:39:28 not sure what you mean? 2022-11-02 10:40:31 ah 2022-11-02 10:40:33 the part of the file 2022-11-02 10:40:33 yeah 2022-11-02 10:41:23 it's in the uclibc part and the glibc part but not the musl part, so i think it's not built as musl then 2022-11-02 10:41:27 good catch 2022-11-02 10:41:29 didn't realise that 2022-11-02 10:57:15 can dnsmasq be update to v2.87 in v3.15 and v3.16 pls, before ossl3 gobles it ? 2022-11-02 10:57:54 before what? 2022-11-02 10:58:17 openssl3 2022-11-02 10:59:00 i don't understand the relevance 2022-11-02 10:59:09 I am guessing openssl3 would effect almost all pkgs 2022-11-02 10:59:34 openssl3 isn't default in 3.15 and 3.16 2022-11-02 10:59:45 there are no changes to any existing release 2022-11-02 10:59:47 mayabe would require more effort 2022-11-02 10:59:51 wrt openssl3 2022-11-02 11:00:02 ok 2022-11-02 11:00:02 and we did the openssl3 migration months ago 2022-11-02 11:01:13 and even if all of that, how would ossl3 "goble it" 2022-11-02 11:01:30 v2.86 has regression + sec fixes, it does qualify for backporting, right ? 2022-11-02 11:02:07 its just a pun - kinda things getting lost is current work 2022-11-02 11:02:16 which sec fixes are you referring to 2022-11-02 11:03:34 https://git.alpinelinux.org/aports/commit/?id=b4f63809c1e034e6f710896d2b15dbb39464f726 2022-11-02 11:03:34 prolly Fix write-after-free error in DHCPv6 server code. CVE-2022-0934 refers. 2022-11-02 11:03:42 that is already fixed in 3.15 and 3.16 2022-11-02 11:04:09 I still don't understand the relation of dnsmasq and ossl3 2022-11-02 11:04:29 be it a pun or not 2022-11-02 11:04:31 this linked commit has 0 secfixes 2022-11-02 11:04:54 regression fix? 2022-11-02 11:05:04 which one 2022-11-02 11:07:38 v2.87 work in al3.15 but v2.86 does, this option, "--address=/#/127.0.0.1", if someone confirms it does I can give retry 2022-11-02 11:07:50 v2.86 does not^ 2022-11-02 11:08:55 > Replace --address=/#/..... functionality which got missed in the 2.86 domain search rewrite. 2022-11-02 11:09:32 but was fixed in v2.87 2022-11-02 11:09:58 there is already a backport related to that in 3.16 2022-11-02 11:10:31 regardless, did you verify upgrading to edge with 0 other changes actually fixes the issue 2022-11-02 11:10:34 or 3.16 apparently 2022-11-02 11:11:04 v3.15 also pls 2022-11-02 11:13:28 https://thekelleys.org.uk/dnsmasq/CHANGELOG - Fix problem with the configuration 2022-11-02 11:15:15 i asked a question 2022-11-02 11:18:14 which one, this linked commit has 0 secfixes ? 2022-11-02 11:19:51 the last thing i said 2022-11-02 11:20:09 elly: gfx failures for some fun https://img.ayaya.dev/Qce1GhBz7L21 2022-11-02 11:20:15 some of those look funny, some very weird 2022-11-02 11:22:43 Ariadne: I have a patch for gcc, do you still want that to be posted to https://gitlab.alpinelinux.org/alpine/alpine-gcc-patches? seems like that has not been used for a while? 2022-11-02 11:22:47 upgrading to edge ? I cannot do on all stable boxes, but it does work 2022-11-02 11:23:01 to 3.16 2022-11-02 11:23:51 Ariadne: is this up-to-date? https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/main/gcc/APKBUILD#L192 2022-11-02 11:24:07 will move to v3.16 in couple of weeks, its still does not show here, http://dl-cdn.alpinelinux.org/alpine/v3.16/main/armv7/ 2022-11-02 11:24:16 dnsmasq-2.86-r3.apk 2022-11-02 11:24:33 it's there though 2022-11-02 11:24:38 https://dl-cdn.alpinelinux.org/alpine/v3.16/main/armv7/dnsmasq-2.86-r3.apk 2022-11-02 11:24:45 ncopa: you should add it to the alpine/ repo, nmeum also deals with it now 2022-11-02 11:24:52 backports don't bump package version, they bump revision 2022-11-02 11:25:09 yes, please add it to the alpine repo 2022-11-02 11:25:16 there is a branch for 12.2 2022-11-02 11:25:31 https://gitlab.alpinelinux.org/alpine/alpine-gcc-patches/-/tree/alpine/12.2 2022-11-02 11:25:57 ok, then v3.17 release would update it to v2.87 ? 2022-11-02 11:26:12 nmeum: thank you! 2022-11-02 11:27:36 i misunderstood. i thought alpine/12.2 was alpine v3.12, which it is not 2022-11-02 11:28:48 ncopa: ah, also people keep running into upgrade issues with the default bootsize now due to a threshhold for the kernel's size being passed, so https://gitlab.alpinelinux.org/alpine/alpine-conf/-/issues/10536 should be fixed sooner rather than later so the next .iso has a more correct setup result 2022-11-02 11:29:13 Ran into that as well on ppc64le 2022-11-02 11:37:59 ok. will have a look at that asap 2022-11-02 12:05:57 dnsmasq pkgs from http://dl-cdn.alpinelinux.org/alpine/v3.17/main/armv7/ only works 2022-11-02 12:06:08 thanks 2022-11-02 12:43:59 ok, libphobos should be fixed now 2022-11-02 12:44:58 nice! 2022-11-02 12:50:41 ncopa: !40980 2022-11-02 12:51:59 ikke: thank you! 2022-11-02 14:01:41 applying CVE patches is great(for sec oriented distro), but that should not be the end point if upstream has released a new version in given (alpine-upgradable) time-frame 2022-11-02 14:02:34 this will distort the concept of stable version vs. edge 2022-11-02 14:13:03 psykose: ncopa I ran into the same issue as https://gitlab.alpinelinux.org/alpine/alpine-conf/-/issues/10536, didn't got time to look into fixing it. Is there a way to fix it without reinstalling Alpine ? 2022-11-02 14:14:04 i'd say boot a live usb with gparted and repartition 2022-11-02 14:14:13 but make backup first 2022-11-02 14:14:47 like this: https://gparted.org/livecd.php 2022-11-02 14:15:51 Yeah it is a plain install at the moment, there's not much on it. Maybe reinstalling is not such a bad option then as I still have to do livecd booting 🙂 2022-11-02 14:16:24 so, you can always do: lbu package 2022-11-02 14:16:34 it will create a tarball with the config 2022-11-02 14:17:39 Ah yeah, that's even easier to reinstall 🙂 Thanks! 2022-11-02 14:17:43 then you use that with setup-disk -o myhost.apkovl.tar.gz 2022-11-02 14:46:14 psykose: so did you took care of the yeetening 2022-11-02 14:46:23 it is morning and i have some spare time 2022-11-02 14:52:18 ping craftyguy about osk-sdl 2022-11-02 17:09:41 ncopa: with that fix merged, downgrade to 3.15 is now working agian 2022-11-02 17:10:22 psykose: seems like downgrade to 3.14 works fine without it 2022-11-02 17:11:05 to 3.13 as well 2022-11-02 18:34:21 is normal that python tries to write on __pycache__/ ? https://tpaste.us/5kWM maybe the current cache is outdated for that package? 2022-11-02 18:35:02 it only happens to jinja2 when starting qutebrowser 2022-11-02 21:00:45 psykose: ncopa: I'm not sure where the appropriate place to write this down is, but these 3 patches will now not be needed as of chromium 109: canonicalize-file-name, credentials-sys-types-header, systypes 2022-11-02 21:02:26 elly: not sure you were aware, but psykose officially adopted the chromium package now 2022-11-02 21:02:30 oh okay cool 2022-11-02 21:02:35 I was not indeed 2022-11-02 21:40:49 very nice 2022-11-02 22:22:58 nmeum: seems like ghc is what's currently blocking aarch64 and x86_64 2022-11-02 22:23:05 going to try bootstrapping aarch64 now 2022-11-02 22:29:37 nmeum: for some reason, on aarch64 it cannot find llvm14 2022-11-02 22:31:04 https://build.alpinelinux.org/buildlogs/build-3-17-aarch64/community/ghc/ghc-9.0.2-r1-bootstrap.log 2022-11-02 22:52:34 you need 2022-11-02 22:52:43 export PATH="$PATH:/usr/lib/llvm14/bin" 2022-11-02 22:53:04 bl4ckb0ne: i need confirmation it works first, the rest takes me no time :) 2022-11-02 22:53:13 works -> is ok 2022-11-02 22:53:16 psykose: in the APKBUILD? 2022-11-02 22:53:19 ikke: yeah 2022-11-02 22:54:02 you could make it llvm$_llvmver/bin since it would be a repeated issue every time it's not the default 2022-11-02 22:54:24 also weird, i did see those downgrade issues on 3.14 ci too 2022-11-02 22:55:21 elly: thanks! :) 2022-11-02 22:55:52 psykose: if so, we can certainly do the same for 3.13/14 2022-11-02 22:56:04 maybe i was dreaming 2022-11-02 22:56:21 I have only seen it for 3.15 so far 2022-11-02 22:57:03 is ddv going to fix hare? 2022-11-02 23:08:42 dunno 2022-11-02 23:09:02 ddevault: on 3.17-aarch64 hare also fails to build for some reason while it didn't on edge https://build.alpinelinux.org/buildlogs/build-3-17-aarch64/community/hare/hare-0_git20221031-r0.log 2022-11-02 23:09:15 ack 2022-11-02 23:09:18 can you file a hare ticket about this 2022-11-02 23:13:08 done 2022-11-02 23:13:33 thanks 2022-11-03 00:09:12 just a thought, what if all -dbg- pgks were build in main.dbg or community.dbg 2022-11-03 01:32:45 Hello! I'm hoping to help get the "helm" package moved from "testing" to "community" but I'm having trouble finding out how to do so. I created a GitLab issue: https://gitlab.alpinelinux.org/alpine/aports/-/issues/14317 What else can I do? I'm a Gentoo Developer, so I'm more than happy and (I believe) capable to assist. 2022-11-03 01:55:06 it requires the maintainer to want to move it 2022-11-03 01:58:35 psykose: do you happen to know why default-pthread-stacksize.patch changes the minimum stack size from 0 to 2MB? 2022-11-03 02:00:07 as far as I know the default stack size is 2MB anyway, so returning 0 (and leaving the size at the default) ought to be just fine 2022-11-03 02:00:20 platform_thread_mac.mm would give the reasoning 2022-11-03 02:00:24 it's not, the default is 128k 2022-11-03 02:00:46 it's one of the most common musl porting issues indeed 2022-11-03 02:01:09 you'll see lots of them 2022-11-03 02:01:13 aha, okay 2022-11-03 02:01:20 i still owe psykose some research on why we do it for binutils or something but i keep forgetting :( 2022-11-03 02:01:35 technically, the mac.mm limit is set to 8mb 2022-11-03 02:01:38 i'm not sure why 2022-11-03 02:01:44 it goes like that for most macos ports 2022-11-03 02:01:48 llvm does the same 2022-11-03 02:01:52 if macos { 8mb stack } 2022-11-03 02:02:33 ok, platform_thread_mac has logic to return the max of: default stack size, PTHREAD_STACK_MIN, or the stack rlimit 2022-11-03 02:02:34 either macos is doing some special (doubt), or people that run into the issue (macos having a low default like musl) are too lazy to profile the requirement so they pick an absurd number 2022-11-03 02:02:38 huh 2022-11-03 02:02:40 i wonder if it's like we for musl and we tend to just copy paste it from the last place we saw it 2022-11-03 02:02:42 or if there's an actual reason 2022-11-03 02:02:46 Apple BigStacks (TM) 2022-11-03 02:03:01 objc does use quite a lot of stack I think, but not sure 2022-11-03 02:03:03 yeah, but the comment states the default rlimit is 8 2022-11-03 02:03:28 in any case, 2mb works everywhere as a general default 2022-11-03 02:03:55 it's the smallest glibc rlimit afaik 2022-11-03 02:04:03 https://ariadne.space/2021/06/25/understanding-thread-stack-sizes-and-how-alpine-is-different/ ah, here we go 2022-11-03 02:04:19 this entire page is probably useful for you as well https://wiki.musl-libc.org/functional-differences-from-glibc.html#Thread-stack-size 2022-11-03 02:04:56 cheers, thanks 2022-11-03 02:05:21 sam_: i've mostly forgotten now but iirc it was raised a while ago without justification? i didn't find a specific bugzilla reference to 'x crashes stack raised now doesnt', it was done ahead of time 2022-11-03 02:05:47 but my memory is quite bad so i can't say for sure 2022-11-03 02:06:31 in any case i don't think it's the end of the world to have it there 2022-11-03 02:13:56 nobody documented why they decided that this one random thread needed a 32KB stack instead of the default size, sigh :P 2022-11-03 02:16:34 there's always git blame 2022-11-03 02:16:59 but for tiny sizes the implied reason is "we know the hard limit and this is a memory saving" or then also on top of that "we spawn 1k of this" 2022-11-03 02:17:17 yeah 2022-11-03 02:17:31 I think this one is just premature optimization though since there's one such thread 2022-11-03 03:05:37 psykose: well, I eventually figured it out after a fashion... the stack size when that thread was initially introduce was 4kb, and it's just been bumped up since then as stuff crashed :) 2022-11-03 03:05:50 hah 2022-11-03 03:05:51 tiny :) 2022-11-03 03:06:00 indeed 2022-11-03 03:06:16 but I now have a pretty solid rationale for just setting it to 32k, which is that empirically 16k is too small 2022-11-03 03:16:52 I guess this counts as a "proper" fix for the stack sizes: https://chromium-review.googlesource.com/c/chromium/src/+/4001142 2022-11-03 03:20:37 massive 2022-11-03 03:21:02 haven't seen this constexpr bool style before in the ifdefs 2022-11-03 03:21:20 cute 2022-11-03 03:21:46 looks fine to me. i assume the change from *2 /*4(asan) to *1 /*2(asan) is intentional given the rationale 2022-11-03 03:40:05 yeah it is 2022-11-03 03:40:34 and yeah the constexpr bool thing is sort of a hack - I wish I could write: constexpr bool tsan = defined(THREAD_SANITIZER); but alas 2022-11-03 05:50:59 nmeum: ghc on aarch64 passed (after adding llvm14-bin to PATH) 2022-11-03 06:28:32 ptrc: Would you like to take over maintaining py3-sphinx-argparse or maybe mark it unmaintained? I packaged it for qtile (I think) but I'm happy with Openbox so I'm not going to package qtile. 2022-11-03 11:04:12 im working on ghc on build-3-17-x86_64 2022-11-03 13:05:00 ikke: I think the tests are disabled on aarch64 though 2022-11-03 13:05:10 nmeum: ok, that would explain 2022-11-03 13:08:25 is dl-cdn down? 2022-11-03 13:08:37 nmeum: should be back 2022-11-03 13:08:44 trying to switch to a different back-end 2022-11-03 14:51:15 psykose: on one setup, did not want to mix v3.15+edge, though newer dnsmasq works 2022-11-03 14:51:19 now using stable dnsmasq(for dhcp)+tinydns(for dns), thanks for help 2022-11-03 15:26:28 ktprogra1: sure, i can take over maintainership 2022-11-03 15:41:43 nmeum: seems like the current ghc maintainer was last active 5 years ago 2022-11-03 15:43:26 yea, i am aware that he isn't active anymore 2022-11-03 15:44:09 with the patch the ghc testsuite passes locally for me, not sure if you are also running into additional issues on the builders 2022-11-03 15:44:21 I will merge it if it also passes on the ci 2022-11-03 15:44:34 ptrc: thanks! 2022-11-03 15:44:42 on aarch64, we needed to add /usr/lib/llvm14/bin to PATH 2022-11-03 15:44:51 I did not commit that though 2022-11-03 15:46:36 makes sense since llvm15 is now our default llvm implementation 2022-11-03 15:46:42 so feel free to commit that 2022-11-03 15:47:03 not sure why it is not needed on x86_64 though 2022-11-03 15:47:17 because it doesn't use llvm there according to psykose 2022-11-03 15:47:25 ah! 2022-11-03 15:47:36 right, ghc has multiple compiler backends 2022-11-03 15:47:51 She also hinted that aarch64 might support the native backend now? 2022-11-03 15:48:00 it's also why it's slow 2022-11-03 15:48:46 native backend is in uhh 2022-11-03 15:48:49 9.2.1 2022-11-03 15:49:02 ah ok 2022-11-03 15:49:02 probably not a default though i have no idea 2022-11-03 15:49:13 looking at haskell stuff hurts it's like the worst toolchain ever envisioned 2022-11-03 15:51:13 https://gitlab.haskell.org/ghc/ghc/-/merge_requests/3641 https://gitlab.haskell.org/ghc/ghc/-/merge_requests/5575 2022-11-03 16:00:02 nmeum: right, aarch64 failed now because of that 2022-11-03 16:07:51 *nod*, just push your PATH change :) 2022-11-03 16:08:18 I don't have it anymore :P 2022-11-03 16:08:27 I hotfixed it on the builder 2022-11-03 16:08:44 I can also just push it to the MR 2022-11-03 16:08:48 sounds good 2022-11-03 17:13:11 psykose: yeah it's like rust except 3 times as bad 2022-11-03 17:18:52 was about to disagree 2022-11-03 17:18:54 then noticed 2022-11-03 17:18:57 in the last version of rust 2022-11-03 17:19:01 they went from 3 versions of libc 2022-11-03 17:19:02 to 1 2022-11-03 17:19:05 and in this version 2022-11-03 17:19:08 they went from 1 2022-11-03 17:19:09 to fucking 2022-11-03 17:19:10 FOUR 2022-11-03 17:19:12 ????????????????????????????????????????????? 2022-11-03 17:19:35 https://img.ayaya.dev/gXLON43x6NS8 2022-11-03 17:19:39 who smoked the crack again 2022-11-03 17:19:55 I feel less bad about the chromium package 2022-11-03 17:22:01 likewise 2022-11-03 18:24:55 psykose: looked into the hare issues you pointed out for aarch64 and my conclusion is I have no idea 2022-11-03 18:25:01 just werks for me 2022-11-03 18:25:24 yeah, just like it passed on edge when technically 3.17 should be the same 2022-11-03 18:25:41 lemme try reproducing on a 3.17 system 2022-11-03 18:25:59 err 2022-11-03 18:26:06 3.17 doesn't exist yet, right 2022-11-03 18:26:11 the repos do i think 2022-11-03 18:26:12 hm 2022-11-03 18:26:22 my only guess is gcc12 qbe? it was last built with gcc11 on edge 2022-11-03 18:26:27 not for aarch64 2022-11-03 18:26:41 It will only upload the repo after all packages are built 2022-11-03 18:26:54 to test that you could rebuild only qbe yourself and see if they fail suddenly 2022-11-03 18:27:11 I have a qbe build on gcc12 2022-11-03 18:27:14 lemme see... 2022-11-03 18:27:15 not that then 2022-11-03 18:27:32 one that I haven't tested it with, I deliberately didn't test with that because I wanted to try to reproduce whatever the builders might have 2022-11-03 18:27:39 i.e. the packaged version of qbe 2022-11-03 18:27:53 builders have a gcc12 version as 3.17 is built from scratch there 2022-11-03 18:28:15 not reproducible on a fresh qbe build on gcc12 2022-11-03 18:28:24 nor harec on gcc12 2022-11-03 18:28:35 ddevault: anything I can check / test on the builder to help you? 2022-11-03 18:30:07 can you try to build the hare package in a shell and see if it fails in an interactive environment? 2022-11-03 18:30:12 sure 2022-11-03 18:30:14 if so, can you give me access to said interactive environment? 2022-11-03 18:30:30 No, that I cannot 2022-11-03 18:30:33 figured 2022-11-03 18:30:47 also, psykose, we may not be able to get a himitsu-ssh fix done in time for the release, we have to do some refactoring to work around the breaking API change 2022-11-03 18:31:16 bbiab 2022-11-03 19:27:39 long builds take long.. 2022-11-03 19:27:45 Waiting for ff to finish 2022-11-03 20:09:27 finaly 2022-11-03 20:11:34 ddevault: failed interactively as well 2022-11-03 20:11:45 4 test failures 2022-11-03 21:25:48 psykose: hm, how should we keep track of which chromium patches will be no longer needed? 2022-11-03 21:25:57 me telling you their names on IRC seems like a bad approach :) 2022-11-03 21:27:30 i write them down somewhere 2022-11-03 21:27:32 good enough for me 2022-11-03 21:27:35 alright 2022-11-03 21:27:57 at best if forgotten they don't apply and it's obvious anyway 2022-11-03 21:28:06 at worst they stay and it's a useless patch that doesn't do anything 2022-11-03 21:28:15 yeah that's fair 2022-11-03 21:28:18 not exactly rocket surgery 2022-11-03 21:31:24 psykose: https://www.youtube.com/watch?v=THNPmhBl-8I 2022-11-03 21:34:01 :) 2022-11-03 21:44:41 oh cmon, I guessed the punchline as soon as the skit started :P 2022-11-03 21:45:51 (comedy isn't exactly SOFTWARE ENGINEERING) 2022-11-04 02:26:24 doesn't-really-matter question; I see "meson compile ${JOBS:+-j ${JOBS}}" in quite a few APKBUILDs, but abuild sets SAMUFLAGS which should take care of parallelism properly afaik? Is that needed? (doesn't really hurt, just curious why it's here) 2022-11-04 03:32:31 Asmadeus: maybe it predates SAMUFLAGS? (guess) 2022-11-04 07:32:27 ikke: does it also fail if you build outside of the abuild environment? 2022-11-04 08:28:24 ikke: if so, apply this diff: https://paste.sr.ht/~sircmpwn/422020ee515b7e3ce90cb5afbd9fc99d163d3aad 2022-11-04 08:28:30 ikke: then run hare build -o hare-tests -lc 2022-11-04 08:28:37 ikke: then you can do valgrind ./hare-tests, might narrow it down 2022-11-04 08:41:53 also pushed a fix to hare master which is /probably/ unrelated to the errors, but who knows 2022-11-04 10:02:58 ddevault: will do that in a bit 2022-11-04 10:04:39 ikke: cheers 2022-11-04 11:34:42 ddevault: it also fails when building outside of abuild 2022-11-04 11:34:52 will try the patch now 2022-11-04 11:40:47 ddevault: Error: Module (root) has no Hare input files 2022-11-04 11:40:55 at which step? 2022-11-04 11:41:13 .bin/hare build -o hare-tests -lc 2022-11-04 11:41:27 which hare version? 2022-11-04 11:41:33 oh no 2022-11-04 11:41:35 don't do hare build, do hare test 2022-11-04 11:42:08 that works 2022-11-04 11:42:52 ddevault: https://tpaste.us/bn6X 2022-11-04 11:43:14 uh 2022-11-04 11:43:16 wat 2022-11-04 11:43:20 can you do hare test -v 2022-11-04 11:43:28 hare test -v [rest of the args you already used], that is 2022-11-04 11:44:17 Do you need the complete output? 2022-11-04 11:44:28 yes please 2022-11-04 11:44:50 https://tpaste.us/70eO 2022-11-04 11:45:47 can you set LDFLAGS=-static 2022-11-04 11:45:50 then try again 2022-11-04 11:46:31 ok, succeeded now 2022-11-04 11:49:33 ddevault: and now? 2022-11-04 11:49:38 valgrind ./hare-tests 2022-11-04 11:49:43 actually, first, just run ./hare-tests 2022-11-04 11:49:46 see if it reproduces the same bug 2022-11-04 11:50:24 segfault 2022-11-04 11:50:32 err 2022-11-04 11:50:42 where? got a backtrace? 2022-11-04 11:50:48 ulimit -c unlimited, reproduce, gdb ./hare-tests core 2022-11-04 11:52:06 actually I see it too 2022-11-04 11:52:14 bleh, I'll get back to you later 2022-11-04 11:52:20 what a bizzare segfault 2022-11-04 11:52:21 https://tpaste.us/1lq4 2022-11-04 11:52:24 ok 2022-11-04 11:52:33 probably doesn't like our linker script 2022-11-04 11:52:34 ddevault: trying to get you a container with 3.17 community available 2022-11-04 11:52:41 alright, thanks 2022-11-04 11:59:33 will have to continue later 2022-11-04 13:24:15 ikke: this whole endeavour has led me to spend most of my day figuring out how to implement ASan for Hare btw 2022-11-04 14:26:05 hare a.k.a. yalll 2022-11-04 14:59:34 andypost[m]: https://bugs.bareos.org/view.php?id=1487#c4823 2022-11-04 15:00:33 " 2022-11-04 15:00:33 Yes, currently there are no known issues that break functionality. " 2022-11-04 15:00:57 So, seems that we are fine with php 8.1 and bareos webui according with upstream's words 2022-11-04 15:07:32 fcolista: thank you! great news0 2022-11-04 16:59:20 so ASan is pretty straightforward https://paste.sr.ht/~sircmpwn/ea48493b6476df2eb67444ddfcbd5c0507538d86 2022-11-04 17:22:17 ye olde 'error detected' 2022-11-04 17:22:57 it beats the even more classic "error not detected" :P 2022-11-04 17:57:30 could change the error to "you done fucked up" 2022-11-04 17:57:43 or add line numbers or the address or something useful like that 2022-11-04 20:34:29 gitlab has been upgraded to 15.3 2022-11-04 20:35:07 Issues now include tasks 2022-11-04 20:36:21 rebasing without triggering a pipeline! 2022-11-04 21:16:39 there is a file in dev.alpinelinux.org/archive owned by root 🤔 tor-src.tar, not even compressed. 105MB 2022-11-04 21:21:02 It has several versions of tor tar archives 2022-11-04 21:21:49 https://tpaste.us/k649 2022-11-04 21:22:16 that's not the archive folder :p 2022-11-04 21:22:50 That's in the tar file 2022-11-04 21:23:04 ah 2022-11-04 21:23:09 bizarre 2022-11-04 21:23:12 just delete it i guess 2022-11-04 21:23:42 nothing in aports references tor-src 2022-11-04 21:23:49 ok 2022-11-04 21:24:03 p00f 2022-11-05 06:07:17 ikke: new gitlab broke algitbot commit messages in -commits 2022-11-05 07:13:54 phun 2022-11-05 07:20:09 giga fun 2022-11-05 07:26:44 Hmm, doesn't seem to have to do with the upgrade 2022-11-05 08:41:21 ikke: armhf is still stuck fyi 2022-11-05 10:07:06 at this point that's the normal stable state of an arm builder 2022-11-05 10:08:44 you missed the other 25 times probably 2022-11-05 10:08:54 of every other builder, individually, combined 2022-11-05 10:09:19 maybe it's just the stable state of a builder then, no matter the arch 2022-11-05 10:09:45 (the annoying part is always me noticing it, feeling too lazy to mention it, then seeing that nobody else saw it 40 hours later) 2022-11-05 10:09:49 sure wish i could just do it myself 2022-11-05 10:10:25 I vote to give you percussive maintenance rights on builders 2022-11-05 10:11:16 that's too much power you see 2022-11-05 10:12:50 the permissions model doesn't make the difference between the power to give a gentle kick to wake a builder up, and the power to bind and torture and kill it 2022-11-05 10:12:53 that's a shame 2022-11-05 11:22:39 tpaste.us is down ? 2022-11-05 12:11:59 Is pkgs.a.o down? 2022-11-05 12:13:29 yes, there is some network issue, but it works via ipv6 2022-11-05 12:18:12 and now gitlab is, hehe 2022-11-05 12:18:47 that's me :) 2022-11-05 12:18:51 trying to get ipv6 enabled 2022-11-05 12:18:55 it;s back 2022-11-05 21:39:53 Is it possible to exclude specific files from dbg subpackage? Package build is dying with "objcopy: Unable to recognise the format of the input file", so I would like to exclude the offending file. Is that possible? 2022-11-05 21:43:27 graywolf: define dbg(), call default_dbg, then remove that file from $subpkgdir 2022-11-05 21:46:07 ikke: I guess other way around, since the default_dbg is dying. So I guess the sequence could be "define dbg(); move the file(s) to /tmp; call default_dbg; return the files to their location" 2022-11-05 21:46:46 Does this type of a thing have a chance to pass review into aports? 2022-11-05 21:55:16 Great, it does build now, thanks a lot. 2022-11-05 22:25:09 this sounds like a bad idea 2022-11-06 01:26:48 I have no issue with comments like this if they are followed by an elaboration 2022-11-06 01:37:37 elaboration for why this sounds like a bad idea (imo): package scripts should not create anything outside of $startdir (aports/repo/pkgname), no matter if it's a "known" writable directory like /tmp or not, because there can be some cases where it's not guaranteed 2022-11-06 02:04:57 yes, you should at least move the files to a build-owned directory. best to put it in srcdir actually, since startdir might not be writable. my bigger problem is regarding why some file is detected to be stripped but actually can't 2022-11-06 02:05:52 the underlying problem should be fixed, whether that's a bug in abuild or it is some precompiled binary 2022-11-06 02:20:52 Hello71: or it might be a bug in the toolchain, if it only occurs on some architecture like rv64 2022-11-06 06:05:16 anyone have an armv7 or x86 builder handy they can test a patch for me on? it's for the hanging test(s) in strongswan as commented in the APKBUILD 2022-11-06 06:07:14 sure 2022-11-06 06:09:10 moment, will pm it. 2022-11-06 06:09:29 you can just pastebin and post it here 2022-11-06 06:09:45 though i'm seeing if i can even reproduce the hang 2022-11-06 06:10:10 seems to not 2022-11-06 06:10:22 weird 2022-11-06 06:11:04 http://ix.io/4f59 2022-11-06 06:11:48 this is just playing around, i do not mean to imply any sort of correctness in both the problem or the solution 2022-11-06 06:14:15 hah, and on a retry it hangs 2022-11-06 06:14:39 see if you can collect some data from gdb on that, 2022-11-06 06:14:44 i haven't been able to get much 2022-11-06 06:15:55 hm, so now it passed tests without patches for me, for the first time out of ten or fifteen 2022-11-06 06:16:20 yeah sometimes it does 2022-11-06 06:16:28 still gets stuck with the patch 2022-11-06 06:16:37 sadly no ptrace in this container 2022-11-06 06:16:49 ok, disregard then. thanks for taking a look. 2022-11-06 07:18:06 ikke: s390x is stuck yet again :) 2022-11-06 10:01:04 so it is written, and so it shall be 2022-11-06 11:12:21 ikke: would you look at that, same thing again :p 2022-11-06 11:13:24 Trying the same thing and expecting different results. What was that again? 😋 2022-11-06 11:35:14 should still kick it 2022-11-06 11:37:06 sure 2022-11-06 11:38:10 ddevault: I saw that prometheus has begun with LTS releases, the first one is 2.37.x but it is only supported through january, I'm guessing not too long after our 3.17 release and perhaps not worth downgrading to but rather keep an eye at and stay at future LTS releases going forward 2022-11-06 11:38:21 makes sense to me 2022-11-06 15:11:09 ddevault, ikke: all 3.17 builders are done so you can split the branch now 2022-11-06 15:11:18 er 2022-11-06 15:11:21 ncopa: 2022-11-06 15:11:27 wrong ping hehe 2022-11-06 15:12:29 psykose: first there will be rc's 2022-11-06 15:12:45 ah 2022-11-06 15:12:51 annoying 2022-11-06 15:43:03 whts the eta for 3.17? 2022-11-06 15:43:34 1 or 2 weeks 2022-11-06 15:43:51 neat 2022-11-06 16:23:10 psykose: mind if I move your containers on nld5 to a new server? 2022-11-06 16:23:19 go for it 2022-11-06 16:26:24 lots of ccache stuff :P 2022-11-06 16:29:32 ccache -C 2022-11-06 18:58:57 nmeum: can I move your x86* and riscv dev containers? 2022-11-07 03:53:57 ikke: the bot no longer assigns mrs 2022-11-07 05:27:13 cannot reach the webhook endpoint 2022-11-07 05:37:38 HTTP was not meant to be an RPC protocol 2022-11-07 05:45:37 gitlab implements rfc 88 when 2022-11-07 05:47:33 card readers / line prints, lmao 2022-11-07 05:47:51 THE SOCKET CZAR 2022-11-07 05:48:07 https://datatracker.ietf.org/doc/rfc740/ 2022-11-07 05:48:54 I won't deny that HTTP is the best we yet have. I still don't think I'm unentitled to say that it's not meant to be an RPC protocol. 2022-11-07 05:49:02 whoa, ARPANET support 2022-11-07 05:49:15 I read that out loud... in an american accent 2022-11-07 05:49:51 sorry, I'm too chaotic, I should get back to hacking 2022-11-07 05:52:03 how about soap 2022-11-07 05:52:31 (pro tip: don't let me make software) 2022-11-07 05:53:19 ssh is a relatively alright RPC protocol 2022-11-07 05:53:33 although more remote program call than remote procedure call 2022-11-07 06:13:24 morning! 2022-11-07 06:13:32 yay! \o/ all builders are idle! 2022-11-07 06:14:22 ahuh 2022-11-07 06:14:35 psykose disabled the packages that were failing :P 2022-11-07 06:15:04 at least some 2022-11-07 06:15:10 now lets go over https://gitlab.alpinelinux.org/alpine/aports/-/issues and fix whatever we can for 3.17 2022-11-07 06:15:24 you make it sound like i disabled all of aports and not like 3 packages 2022-11-07 06:15:25 :p 2022-11-07 06:15:46 thats fine. 2022-11-07 06:15:52 it was just hare things 2022-11-07 06:15:57 yup 2022-11-07 06:16:07 aside from that i fixed multiple hundreds of things 2022-11-07 06:16:09 it was a fun week 2022-11-07 06:16:11 i should take the opportunity to tag an edge release 2022-11-07 06:16:31 psykose: you are awesome and do an awesome work 2022-11-07 06:16:49 psykose: sorry, did not mean to denigrate your awesome work :) 2022-11-07 06:16:53 :) 2022-11-07 06:17:43 there are a bunch of issues in https://gitlab.alpinelinux.org/alpine/alpine-conf/ that needs to be fixed 2022-11-07 06:17:52 aye 2022-11-07 06:18:00 and in mkinitfs 2022-11-07 06:19:17 ncopa: this week I'm also planning to move alpine-www and alpine-msg to the new server 2022-11-07 06:26:44 psykose: gitlab can reach the qa-bot webhook endpoint again, but testing it through the webif returns a 404, so need to see if it works 2022-11-07 06:30:06 I should just C-\ 2022-11-07 06:30:16 ikke: make an mr :-) 2022-11-07 06:31:26 didn't work 2022-11-07 06:42:29 should work now 2022-11-07 06:47:53 im tagging an edge snapshot release now. please hodl your pushes for a few mins, thanks 2022-11-07 06:49:03 ikke: works 2022-11-07 06:49:17 ok. v20221107 tag was pushed. thanks! 2022-11-07 06:49:42 :) 2022-11-07 06:54:10 ikke: s390x ci is also down 2022-11-07 06:58:25 somethign bad happened 2022-11-07 06:58:39 the release snapshot got not uploaded for soem reason 2022-11-07 06:59:05 maybe because both 3.17 and edge generated it? 2022-11-07 06:59:33 they upload to different locations, so it shouldnt 2022-11-07 06:59:50 not that then 2022-11-07 07:01:44 tar: etc: Not found in archive 2022-11-07 07:01:44 tar: Exiting with failure status due to previous errors 2022-11-07 07:13:11 ok. this broke release script: 23e66e85c95beef9d3f72a2ccc510671fdb3462d 2022-11-07 07:13:37 ah 2022-11-07 07:13:41 guess it needs to depend on release 2022-11-07 07:13:50 the problem is that script expects os-release to be in alpine-base 2022-11-07 07:14:46 https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/scripts/genrootfs.sh#L41 2022-11-07 07:17:39 sigh 2022-11-07 07:18:15 ok, i need to clean this up. can you please hold your git push'es for a few more minutes? 2022-11-07 07:18:55 I wonder what the best way to clean it up woudl be 2022-11-07 07:19:09 i can manually patch the builders 2022-11-07 07:19:15 or delete the tag 2022-11-07 07:19:24 and fix script and create new tag 2022-11-07 07:19:47 just call it 08 and pretend you're in future time 2022-11-07 07:20:18 or leave 07 broken and create a new tag tomorrow 2022-11-07 07:44:13 fine to push then? 2022-11-07 07:46:56 please hang on a sec 2022-11-07 07:48:51 np 2022-11-07 08:21:32 ok. i give up 2022-11-07 08:21:39 i will do new tag tomorrow 2022-11-07 16:34:09 ikke: edge can't upload due to mirror space 2022-11-07 16:34:11 as you probably know already 2022-11-07 16:35:02 huh 2022-11-07 16:35:13 or some other issue 2022-11-07 16:35:15 dl-master.a.o should have plenty 2022-11-07 16:35:21 hm 2022-11-07 16:35:28 well, nothing on edge has updated in a day 2022-11-07 16:37:32 http://dl-master.alpinelinux.org/alpine/edge/community/x86_64/ seems to have new packages 2022-11-07 16:37:58 how is all this infrastructure funded? 2022-11-07 16:38:05 APKINDEX.tar.gz 07-Nov-2022 16:16 2022-11-07 16:38:08 elly: all sponsored 2022-11-07 16:38:12 hm 2022-11-07 16:38:18 i guess it's just pkgs.a.o that's failing then 2022-11-07 16:38:20 fun 2022-11-07 16:38:21 yes 2022-11-07 16:38:52 elly: various linode/equinix sponsorships mostly, then there's some cloud credits we don't really use that much 2022-11-07 16:38:56 there was an arm server somewhere 2022-11-07 16:39:01 some ibm thing for s390x 2022-11-07 16:39:08 something similar for ppc64le 2022-11-07 16:39:36 neat! 2022-11-07 16:39:48 all the arm stuff is on an equinix 2x altra 80 core server 2022-11-07 16:39:58 160 core :P 2022-11-07 16:40:02 ah 2022-11-07 16:40:04 2x80 2022-11-07 16:40:09 ask them for the 2x128 /s 2022-11-07 17:47:39 What about adding a template like archwiki's Pkg to wiki.a.o? 2022-11-07 20:22:45 ikke: the riscv builder also doesn't build anything for some reason 2022-11-07 20:23:15 probably disconnected from msg 2022-11-07 20:39:59 somehow rv64 does not get dhcp anymore 2022-11-07 20:40:06 I thought that clandmeter fixed that 2022-11-08 07:24:44 librados17 fails to properly replace librados16. the `community/ceph` APKBUILD has some logic to replace the previous version when a new one exists but it seems to fail and I don't really understand it 2022-11-08 08:20:59 PureTryOut: what is the error message? 2022-11-08 08:21:16 `ERROR: librados17-17.2.5-r2: trying to overwrite usr/lib/ceph/libceph-common.so.2 owned by librados16-16.2.10-r10.` 2022-11-08 08:21:22 and 3 more files 2022-11-08 08:29:30 the logic is from clang/llvm APKBUILD’s, also with that logic bareos builds before ceph17 even though all the packages depending on ceph all have ceph-dev as a makedepends. so bareos now has ceph17-dev as a makedepends so it builds after ceph17, might be related? 2022-11-08 08:36:17 PureTryOut: there are no replaces 2022-11-08 08:36:56 there is a function called in the subpackage that sets some replaces 2022-11-08 08:36:57 but the whole ceph16 + ceph17 thingy is messy 2022-11-08 08:37:12 agreed 2022-11-08 08:38:07 oh, im looking at old version 2022-11-08 08:38:33 looks like it was fixed in 7fcc10b4a9826b8aee3f96bb870dcfabad46bcdc 2022-11-08 08:42:23 that sets pkgrel to 2 but looks like still pkgrel 1 2022-11-08 08:42:45 ERROR: librados17-17.2.5-r2: trying to overwrite .... 2022-11-08 08:42:49 looks like -r2 to me 2022-11-08 08:43:07 but i think the replaces is broken 2022-11-08 08:43:13 that didn't fix it for me, as I did upgrade to -r2 2022-11-08 08:43:16 replaces="${subpkgname/"$_majorver"/}" 2022-11-08 08:44:12 file an issue, we'll assign it to maintainer 2022-11-08 08:44:17 that removes the major version, if you install py3-rados it does install py3rados17 with librados17 2022-11-08 08:44:22 Will do 2022-11-08 08:44:55 right, so with subpkgname=librados17 2022-11-08 08:45:16 and _majorver=17 2022-11-08 08:45:25 yes 2022-11-08 08:45:26 it should give replaces=librados 2022-11-08 08:45:59 https://gitlab.alpinelinux.org/alpine/aports/-/issues/14337 2022-11-08 08:46:06 ERROR: librados17... trying to overwrite ... owned by librados16....` 2022-11-08 08:46:19 replaces=librados won't replace librados16 though 2022-11-08 08:47:16 exactly 2022-11-08 08:47:17 but 2022-11-08 08:47:41 i dont see any replaces at all in librados17 2022-11-08 08:48:27 oh, apk info -a does not show it 2022-11-08 08:48:41 $ apk fetch librados17 --stdout | tar -O -zx .PKGINFO | grep replaces 2022-11-08 08:48:41 replaces = librados 2022-11-08 08:48:44 its there 2022-11-08 08:50:21 darn, i was too late with my comment on the issue :') 2022-11-08 08:50:35 does replaces need changing to _majorver - 1 then? 2022-11-08 08:50:36 anyways, are math expressions supported in ash / apkbuilds? 2022-11-08 08:51:00 a16bitsysop: yeah, it was left without a version because ceph15 didn't have a prefix iirc 2022-11-08 08:51:05 s/prefix/suffix/ 2022-11-08 08:52:46 would that need 2 replaces? As unless someone just upgraded it would still depend on just librados 2022-11-08 08:53:30 hmm, yeah, i guess so 2022-11-08 08:54:00 do we have anything that depends on ceph16? 2022-11-08 08:54:03 or can we remove it? 2022-11-08 08:54:30 does not look like we have anything depending on it 2022-11-08 08:54:39 it’s for people running clusters so they can pin a version 2022-11-08 12:13:35 Hello, quick question. It seems like zoneminder require the perl module Sys::CPU in order to not spam the logs. I was about to package it from GitHub (mzsanford/perl-sys-cpu), but it doesn't have any releases in GitHub. What is the best thing to do in this case? 2022-11-08 12:44:36 Are there tags? 2022-11-08 13:45:04 there's https://metacpan.org/release/MZSANFORD/Sys-CPU-0.61/view/CPU.pm 2022-11-08 13:45:09 it most certainly has versions 2022-11-08 13:45:34 (unfortunately, the latest seems to be 2013, but that's not unusual for perl modules) 2022-11-08 14:06:09 ikke: No tags either unfortunately 2022-11-08 14:08:35 I see three ways forward. 1. Create an Issue in the project and ask for a release. 2. Package Sys::CPU with the current date when you create a snapshot of the repository. 3. Patch the package of zoneminder to handle if Sys:CPU is missing. Any other suggestions? 2022-11-08 14:11:00 you just use the cpan tarball url 2022-11-08 14:15:42 added 2022-11-08 14:20:12 psykose: Got it, thank you, I missed that one: https://cpan.metacpan.org/authors/id/M/MZ/MZSANFORD/Sys-CPU-0.61.tar.gz 2022-11-08 14:38:54 !41131: I added a subpackage which pulls dependencies and a additional config file to start the gtk application using weston. I named it $pkgname-weston but it also contains a openrc init script. Is this ok or should I move the init script to another subpackage $pkgname-weston-openrc? 2022-11-08 16:52:11 ncopa: thanks for the merge! \o/ 2022-11-08 17:37:45 besides irssi, what other aports could/should be moved from main to community? 2022-11-08 17:41:11 weechat in a similar fashion 2022-11-08 21:42:44 how about dansguardian? it hasn't been maintained for a few years, right? 2022-11-08 21:42:57 and there's the maintained e2guardian in community 2022-11-09 01:07:44 is it possible to set depends/arch per (custom-named) subpackage in an APKBUILD? 2022-11-09 01:11:01 yeah, you can just set the `depends` variable in the subpackage function 2022-11-09 01:11:12 as for the arch, it goes after the subpackage function name 2022-11-09 01:11:46 see https://gitlab.alpinelinux.org/alpine/aports/-/blob/8fcbb9da/community/podman/APKBUILD#L41 2022-11-09 01:12:26 thanks 2022-11-09 01:12:31 in the example we have `$pkgname-docker:docker:noarch` 2022-11-09 01:12:42 what does the inner `:docker:` do? 2022-11-09 01:13:49 it's the function name 2022-11-09 01:13:56 see line 95 2022-11-09 01:15:04 nice thanks 2022-11-09 02:16:35 psykose: I saw that you updated the depends in zoneminder. It also needed !41186 :) 2022-11-09 08:32:24 psykose: libxml++-2.6 package exists both in testing/ and community/, could you take a look there please? 2022-11-09 08:35:30 what does it break 2022-11-09 08:42:24 psykose: there's no point in having it twice? 2022-11-09 08:42:38 sounded at least 10x more urgent than that 2022-11-09 08:43:04 not urgent :) 2022-11-09 08:43:39 just pinged you because you maintain one of them and also touched the other one in the last months 2022-11-09 08:44:17 yeah somehow didn't see it 2022-11-09 08:44:18 hah 2022-11-09 11:16:14 :D 2022-11-09 11:19:21 :) 2022-11-09 11:19:26 what's up 2022-11-09 11:30:43 python3.11 is going well 2022-11-09 11:35:57 im trying to fix as many of the alpine-conf issues as possible so I can tag 3.17 rc1 2022-11-09 11:51:31 sounds good :) 2022-11-09 12:09:05 ncopa: you added the feature-usb patch to mkinitfs pkgrel 4, however the build log do NOT show the usb.modules file being patched for some reason 2022-11-09 12:09:30 it is strange, the checksum of the patch is checked, but then the patch is not applied 2022-11-09 12:13:56 minimal: edge or 3.16? 2022-11-09 12:15:19 bah. i know why 2022-11-09 12:15:42 $pkgname-features-usb::https://gitlab.alpinelinux.org/..... is broken 2022-11-09 12:16:02 i pushed mkinitfs 3.7.0_rc1 now 2022-11-09 12:17:13 ok, will retest 2022-11-09 12:20:31 forgot the .patch on the end of it :) 2022-11-09 15:27:39 is somebody taking care of the mesa upgrade to 22.3? 2022-11-09 15:29:26 oh its still a rc 2022-11-09 15:31:30 it'll be out in like over a month 2022-11-09 15:32:35 i saw the mail earlier and i thought it was already out 2022-11-09 15:46:44 nah, lil more 2022-11-09 15:46:49 and then waiting for .1 2022-11-09 15:46:57 but i'll have an mr for .0 for people to test or whatever 2022-11-09 15:49:51 ping me when you make it, my arc gpu isnt working on 22.2.3 2022-11-09 15:49:57 phoronix says i need 22.3 2022-11-09 15:59:57 you can build an rc if you want 2022-11-09 16:00:18 locally? 2022-11-09 16:00:22 ya 2022-11-09 16:00:38 i can make the PR and turn it into the official once its release 2022-11-09 16:00:41 ill do that 2022-11-09 16:01:16 already done actually 2022-11-09 16:01:22 :O 2022-11-09 16:01:53 alright then 2022-11-09 16:01:53 i eat a lot of carrots you see 2022-11-09 16:02:32 it makes you fast? 2022-11-09 16:02:56 as if 😔 2022-11-09 16:06:19 time to test that gpu eh 2022-11-09 16:06:35 i bet you will get at least 10 fps in vkcube 2022-11-09 16:07:58 i bet 12 at least 2022-11-09 19:17:08 how widely used is the v3 package format vs v2? 2022-11-09 19:17:32 v3 is not released yet 2022-11-09 19:18:23 ah, okay 2022-11-09 19:18:24 in alpine it isn't, chimera uses it i guess 2022-11-09 19:21:42 chimera? 2022-11-09 19:22:16 https://chimera-linux.org/ 2022-11-09 19:23:23 nice, another non gnu linux 2022-11-09 19:23:43 oh cool 2022-11-09 19:25:23 just noticed that linux-rpi depends on both linux-firmware-brcm and linux-firmware-cypress, wondering if that should be the case (rather than the RPI disk image and/or setup-alpine adding those firmware packages) - some people like myself don't use RPI Wifi and so don't need the firmware 2022-11-09 19:27:18 I have linux-firmware-none installed and so didn't expect these packages to be installed 2022-11-09 21:57:41 psykose: mesa r2 out 2022-11-09 21:58:25 sure, updated 2022-11-09 22:00:09 ty 2022-11-09 22:18:34 ikke: the dotnets and s390x are stuck 2022-11-09 22:23:35 Business as usual 2022-11-09 22:24:22 aye 2022-11-09 22:24:44 I don't see dotnet building? 2022-11-09 22:25:15 I see it building on aarch64 2022-11-09 22:25:22 did you mean that? 2022-11-09 22:27:22 yea 2022-11-09 22:27:43 and not on :) 2022-11-10 03:06:03 can gitweb be used to do something like git format-patch commithash11^..commithash17 ? 2022-11-10 03:13:28 i don't think even cgit can, and it's a far better web interface than gitweb 2022-11-10 03:18:38 nevermind 2022-11-10 03:18:40 found it 2022-11-10 03:19:40 what's the answer? 2022-11-10 03:19:41 it's either //commitdiff_plain/hash1?hp=hash2 or /?a=commitdiff_plain;p=repo_name;h=hash1;hp=hash2 2022-11-10 03:19:45 thx 2022-11-10 09:34:40 thanks, but that is just the diff of two commits, it does not include all the commits in between, and it seems to be h=hash2;hp=hash1 2022-11-10 10:04:30 hp is hash_parent 2022-11-10 12:55:35 ncopa: Is the idea to postpone the release after the aarch64/arm* builder move? 2022-11-10 12:56:09 when is the move 2022-11-10 12:56:13 14th 2022-11-10 12:56:20 hmm 2022-11-10 12:56:21 and estimated to take 4 days 2022-11-10 12:56:44 sob sob 2022-11-10 12:56:50 the endless waiting.. for the next rust version.. 2022-11-10 12:56:52 :p 2022-11-10 12:57:00 heh 2022-11-10 12:57:26 ikke: when was the arm move? i gues we dont ahve any choice do we? 2022-11-10 12:57:39 ^ 2022-11-10 12:57:43 The 14th 2022-11-10 12:57:47 and having to rebase a million py3.11 conflicts, hehe 2022-11-10 13:15:40 psykose: thanks for fixing ipython :) 2022-11-10 13:15:48 Wanted to use it yesterday and ran into issues 2022-11-10 13:15:52 np :) 2022-11-10 13:16:00 been doing nothing but fixing python all over the place recently 2022-11-10 13:16:01 (I did fix it locally, didn't get to push it yet) 2022-11-10 13:16:08 3.11 will be ready as soon as the split happens 2022-11-10 13:16:08 yeah, I've noticed :) 2022-11-10 13:16:30 tomllib in base means nothing uses tomli anymore 2022-11-10 13:16:36 except like 1 thing for tests 2022-11-10 13:16:39 nice 2022-11-10 13:16:47 makes bootstrapping easier I suppose 2022-11-10 13:16:58 eh, same old, we fixed that already 2022-11-10 13:17:04 right 2022-11-10 13:17:07 with gpep517? 2022-11-10 13:17:49 the main ''bootstrap issue'' presently is things needing setuptools+wheel to build wheels but py3-wheel needs setuptools and so on 2022-11-10 13:17:55 so on that path we still have told setuppy's 2022-11-10 13:17:59 the old* 2022-11-10 13:18:21 gpep517 is mostly the same as build -2 deps with the same path, but it's also faster as it skips compression and skips the usually wrong dependency check by default 2022-11-10 13:19:06 flit-core works with no deps but it vendors a wheel copy to do that, which is ok at that level i guess 2022-11-10 13:19:14 or wait does it 2022-11-10 13:19:22 no, it doesn't 2022-11-10 13:19:23 neat 2022-11-10 13:47:36 ptrc: wrt the above, patch or patches was what I wanted instead of commitdiff_plain, granted there's not patch set limit (default is 15 commits) 2022-11-10 14:38:32 im tagging an edge snapshot release now 2022-11-10 14:38:51 ack 2022-11-10 15:08:58 looks like librespot-openrc is missing some perms on the log file 2022-11-10 15:09:14 > start-stop-daemon: unable to open the logfile for stdout `/var/log/librespot.log': Permission denied 2022-11-10 15:09:32 fixed by touching the file and `chown librespot:audio /var/log/librespot.log` 2022-11-10 15:09:55 clandmeter: do you want a patch? 2022-11-10 15:10:07 sure i'll fix it 2022-11-10 15:10:26 bl4ckb0ne: typically solved by adding `checkpath ..` to start_pre 2022-11-10 15:10:27 unstoppable 2022-11-10 15:10:58 bl4ckb0ne: fyi, clandmeter is on holiday atm, so won't be reading this 2022-11-10 15:11:06 ack 2022-11-10 15:14:14 psykose: while you're at it the conf.d file is probably lacking `supervisor=supervise-daemon` 2022-11-10 15:38:34 Not everyone likes that to be enabled by default 2022-11-10 15:39:05 So the consensus is to set that in conf.d 2022-11-10 16:28:56 something is wrong with mugshot: $ doas apk fix 2>&1 | tpaste 2022-11-10 16:28:56 https://tpaste.us/qa6a 2022-11-10 16:30:05 no, that's an apk bug 2022-11-10 16:30:22 if a file turns into a directory, you see that in the transaction 2022-11-10 16:30:37 the only way to fix that is a pre-install that.. deletes the file and mkdirs before the install 2022-11-10 16:30:47 same for the reverse, directory -> file 2022-11-10 16:31:01 https://gitlab.alpinelinux.org/alpine/apk-tools/-/issues/10856 2022-11-10 16:31:07 egg-info is valid as either 2022-11-10 16:31:17 so sometimes the packages just change the type 2022-11-10 16:31:26 the previous version had only a file 2022-11-10 16:31:39 new installs aren't affected of course, just upgrade 2022-11-10 16:32:10 :/ 2022-11-10 16:32:36 b 12 2022-11-10 16:32:39 c 14 2022-11-10 16:32:42 dismiss that 2022-11-10 16:34:08 Don't leave us hanging, tell us if it was a hit or a miss 2022-11-10 16:37:10 👍 2022-11-10 16:38:18 ncopa: actually, apk fix doesn't fix it right? it still needs a temporary pre-upgrade to remove the prior dir with a version check.. 2022-11-10 16:38:22 bleh 2022-11-10 16:42:20 psykose sinked my destroyer :< 2022-11-10 20:27:14 psykose: I saw that fix and was wondering if we bothered enough to backport it lol 2022-11-10 20:32:42 ya 2022-11-11 07:05:49 we should look over those and find out which ones we need to fix before 3.17.0: https://gitlab.alpinelinux.org/alpine/aports/-/milestones/194#tab-issues 2022-11-11 16:47:32 something is broken on ppc64le 2022-11-11 17:14:23 ok builders are not done yet. seems like we will have to wait another week for rc1 :-( 2022-11-11 17:27:37 somebody please take a look at https://gitlab.alpinelinux.org/alpine/abuild/-/merge_requests/165 2022-11-11 18:05:28 finally.... 3.16.3 tagged 2022-11-11 18:05:42 on a friday evening... what could possibly go wrong 2022-11-11 18:06:08 @ncopa, ty for the response, I commented below it. 2022-11-11 18:07:30 weird choice of comment placement 2022-11-11 18:08:04 also s/comamnds/commands 2022-11-11 18:18:58 ok, here comes 3.17.0_rc1 2022-11-11 18:24:26 humpf 2022-11-11 18:24:32 im no channel operator again 2022-11-11 18:24:53 can someone please change /topic Alpine Linux distribution development | release 3.17.0_rc1 is out! 2022-11-11 18:25:51 can't you op yourself? 2022-11-11 18:26:25 i probably can but im lazy to google how to do so 2022-11-11 18:30:24 :) 2022-11-11 18:30:26 there we go. thanks panekj :) 2022-11-11 18:30:30 was hopalong removed from aports? 2022-11-11 18:32:01 it's in testing 2022-11-11 18:32:37 oh its tracking wlroots0.12-dev 2022-11-11 18:32:39 thats why i missed it 2022-11-11 20:18:56 I have a question about tarsnap and our packaging policy. It seems that tarsnap is source-available but not free: https://github.com/Tarsnap/tarsnap/blob/master/COPYING 2022-11-11 20:19:34 > Redistribution and use in source and binary forms, without modification, is permitted for the sole purpose of using the "tarsnap" backup service provided by Tarsnap Backup Inc. 2022-11-11 20:20:53 it's nonfree indeed 2022-11-11 20:22:34 I see that we use license="custom" in APKBUILD. So, I'm not sure if we include nonfree packages in our repo, or is it the package user's responsibility to make sure they abide by the provided license? 2022-11-11 20:23:16 no, this is literally just nonfree 2022-11-11 20:23:28 and shouldn't be there 2022-11-11 20:23:29 : ) 2022-11-11 20:25:18 Should I submit an MR removing it? It's in main, and I don't know about how packages are removed from that repo though. 2022-11-11 20:27:28 62640c302501c3191d94002dcd8a52a8eb181b07 2022-11-11 20:35:01 I don't know if I should say thanks or sorry. haha :| 2022-11-11 20:35:34 thanks for pointing it out ^_^ 2022-11-11 20:38:24 that argument is a tad silly 2022-11-11 20:41:41 which one 2022-11-11 20:43:14 it's by definition not free software 2022-11-11 20:43:34 if you made your own 'tarsnap server' it would break the licence as you wouldn't be using the official one 2022-11-11 20:43:57 psykose: the one when they say they would not be paid 2022-11-11 20:44:01 ah 2022-11-11 20:47:19 wait, is the tarsnap package going away? or just moving to community 2022-11-11 20:47:21 because I use that 2022-11-11 20:47:29 it's gone already 2022-11-11 20:47:33 (from aports) 2022-11-11 20:47:38 gone completely* 2022-11-11 20:47:42 poof 2022-11-11 20:47:52 ow, okay 2022-11-11 21:09:03 sadly it's just how it goes sometimes 2022-11-11 21:09:45 dont still my thunder this week end psykose ill take care of -0.15 monday i swear 2022-11-11 21:09:52 :) 2022-11-11 21:10:18 i'll truly let you do it if you want to 2022-11-11 21:10:27 just add a replaces=wlroots in the 0.15 one and that's about it 2022-11-11 21:10:34 as soon as anything supports this i'll merge it 2022-11-11 21:11:01 well, and anything that doesn't needs makedepends= adjusted without rebuild 2022-11-11 21:14:44 i think at least river will do the upgrade during the week end, cage probably too 2022-11-11 21:20:16 ^_^ 2022-11-11 21:21:19 but yeah id like to do it 2022-11-11 23:58:23 My package got flagged as out of date when its up-to-date. Any way to reset it as its a false flag? 2022-11-11 23:59:39 ayakael: do you mean pkgs.a.o/flagged? 2022-11-12 00:00:32 if so, it was probably auto-flagged from anitya (https://release-monitoring.org/) 2022-11-12 00:01:26 aaand it's because someone put the `dotnet6-runtime` package under ".NET Runtime" project which pulls the latest version: https://release-monitoring.org/project/220385/ 2022-11-12 00:01:37 Yeah, one of the false flags was that. Another top-gun flagger flagged my pkg between my merging it and builders still building. Some trigger happy types. 2022-11-12 00:02:10 There you go, I'll investigate a way to change that cause .NET 7 will be included in a seperate dotnet7 package 2022-11-12 00:03:08 you can just change it to 'dotnet7(-runtime)' on the anitya website; you need a fedora account to do that though 2022-11-12 00:03:51 copy that, thanks :) 2022-11-12 00:04:45 fwiw, the flag doesn't mean anything really, it's just a way to notify the maintainer and other packagers that something might be out of date; there's quite a few of invalid ones too 2022-11-12 00:06:24 I appreciate the flag when it's been more than a week - things can fall in between the cracks. Sometimes it just comes too quickly for comfort (like within 24 hours). 2022-11-12 00:08:08 yeah, that sounds like fedora, they run checks constantly and alpine sends flag emails daily at 2AM UTC 2022-11-12 00:19:11 Adjusted, thx much :)) 2022-11-12 00:20:46 Another question I have is related to versionned package. dotnet7 is to be included in MR https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/39122 . Given dotnet6 already being in community, should we skip testing and go directly to community for dotnet7? The test suite is green and the difference between the two are updated patches and 2022-11-12 00:20:46 version, but nothing else. 2022-11-12 00:21:53 what's the release cycle for dotnet? as in, is dotnet7 gonna be supported for 2 years? 2022-11-12 00:22:32 It's supported for 18 months, so not quite. 2022-11-12 00:23:11 I had planned with psykose to include it in 3.17, so scrambled this week to get it ready, and it's looking good. 2022-11-12 00:24:34 oh, nevermind, i confused it with main; community only needs to be supported for current release (usually ~6 months) 2022-11-12 00:24:44 so 18 months is absolutely fine 2022-11-12 00:24:54 Awesome :)) 2022-11-12 14:13:05 ikke: 3.17-x86_64 is stuck : ) 2022-11-12 14:17:30 even x86_64 gets stuck nowadays 2022-11-12 14:17:34 it's the economy 2022-11-12 14:18:17 everything gets stuck all the time 2022-11-12 14:18:23 usually silly testsuites 2022-11-12 14:19:00 either they hang forever or have a 1s timeout and fail in a circle under any load 2022-11-12 14:22:22 what is `want_check` used for, I thought that was for the builders to skip tests 2022-11-12 14:23:09 true with check and false with !check 2022-11-12 14:23:58 in the options? 2022-11-12 14:25:06 yes 2022-11-12 14:26:51 could you default builders to !check while boot strapping and then change back after? It seems a waste of resources running the tests twice 2022-11-12 14:29:23 bootstrapping doesn't have !check 2022-11-12 14:30:03 it also doesn't build twice 2022-11-12 14:31:03 but normally the tests get run on gitlab, then a second time on the builder. 2022-11-12 14:31:21 what does that have to do with bootstrapping 2022-11-12 14:32:01 if it’s the same versions as edge, haven’t the tests already run on gitlab 2022-11-12 14:32:23 i have no idea what you are trying to say 2022-11-12 14:33:20 the ci pipelines and the builders are separate and both build the package and both run the tests 2022-11-12 14:33:39 i have no idea what this has to do with bootstrapping 2022-11-12 14:36:44 or what it has to do with want_check which is true unless !check is in options or ABUILD_BOOTSTRAP=1 which is not set anywhere 2022-11-12 14:38:20 no just the double test run, as from a contributor point of view as well. It is really hard to see whats happening when builder tests fail, and they often just get disabled when they fail on the builder 2022-11-12 19:02:22 I'm completely unaware what the state of CTF support in Alpine is. See https://lwn.net/Articles/795384/ for context. 2022-11-12 19:03:45 I ran into an segfault and the backtrace ends in LLVM code. Rebuilding llvm15 with llvm15-dbg added as subpackage results in a 4.5 GiB llvm15-dbg, and llvm-static also grew by more than 4 GiB in size. I think I would have a hard time selling a MR to add llvm15-dbg by default (: 2022-11-12 19:04:52 So, did anyone try to build debug info with CFT instead of DWARF before? And package that as a $pkgname-dbg subpackage? 2022-11-12 19:06:35 the only things i can find for ctf is 404 page links and patches and switches that are not recognised 2022-11-12 19:06:38 and no, nobody has done it 2022-11-12 19:06:45 you probably want -g1 but it's still 1gb 2022-11-12 19:50:40 whoopsie: ERROR: Failed to create usr/lib/debug/usr/lib/llvm15/lib/libLLVM-15.so.debug: No error information 2022-11-12 19:51:24 Thiat one is reproducible. `tar xf` and manually copying that 4.5 GiB does work. Could it be that `apk` is assuming that file sizes in bytes fit in 32 bit? 2022-11-12 19:52:29 not impossible, for individual files 2022-11-12 19:52:54 well, it's probably not apk 2022-11-12 19:53:00 but the command that tries to split the .debug 2022-11-12 19:53:18 Triggering bugs seems to be my super power these days ... A few weeks ago I was fixing a bug I triggered. The bugfix triggered another bug on a different platform. Debugging that bug triggered a bug in GDB... 2022-11-12 19:53:22 objcopy --only-keep-debug 2022-11-12 23:25:28 it's bugs all the way down 2022-11-13 01:34:21 hmm, the newest chromium crashes when casting 2022-11-13 01:39:20 very cool 2022-11-13 01:39:48 yeah, unfortunately with a very useless stack trace 2022-11-13 10:08:21 psykose: mold 1.7.0 released and author is considering changing licence from AGPL-3.0 to BSL-1.1 in next version 2022-11-13 10:08:34 s/version/versions 2022-11-13 13:58:03 Cogitri: I have an issue with the `clang-sys` crate. When I build software using `clang-sys` (in my case as dependency of the `bindgen` crate) via the rust / cargo version installed locally via `rustup` (from the `rustup` pkg), the build segfaults. It does work with the packaged rust / cargo, though. I sadly do need nightly features for my use case, so just sticking with the packages won't help. 2022-11-13 13:58:41 A backtrace with debug symbols in `libclang` can be seen here: https://github.com/immunant/c2rust/issues/724#issuecomment-1312711370 2022-11-13 14:02:24 Note that `c2rust` does build with our rust and cargo. I have the exact same issue with build rust apps in RIOT (with the exact same backtrace), which do need nightly features. Also note that the latest stable toolchain from `rustup` also fails with the exact same backtrace. I read the backtrace as that the segfault occurs during allocation needed when calling the constructor of a static const C++ class instance right at startup, namely 2022-11-13 14:02:24 https://github.com/llvm/llvm-project/blob/84822479006e177c6388ba77acf2c12a73243870/clang/lib/Format/FormatTokenLexer.cpp#L345-L350 2022-11-13 14:02:43 Does this ring any bell? 2022-11-13 15:48:40 you need -Ctarget-feature=-crt-static 2022-11-13 15:48:43 panekj: i'm aware 2022-11-13 15:48:59 (that -C goes to RUSTFLAGS) 2022-11-13 15:49:21 maribu: ^ 2022-11-13 16:07:10 lol trusting upstream rust builds on musl 2022-11-13 16:07:25 some people are so brave they don't even realize 2022-11-13 16:14:22 yeah they straight up don't work haha 2022-11-13 16:15:02 and the proposal to change from +crt-static to -crt-static for *-musl has just stood there for 2 years or so because it's a "breaking change" 2022-11-13 16:15:17 same with actually basing the libc crate on 1.2.x of musl instead of 1.1.x pre time_t changes 2022-11-13 16:15:20 miserable 2022-11-13 16:16:07 wait, is libc crate still 1.1.x? 2022-11-13 16:16:10 yes 2022-11-13 16:16:11 or is that just in their vendor tree 2022-11-13 16:16:23 hm 2022-11-13 16:16:27 the code is based on 1.1 2022-11-13 16:16:31 i.e. time_t is c_long 2022-11-13 16:16:41 so how does it work if you try to use the compiler on 32-bit platforms 2022-11-13 16:16:44 it doesn't 2022-11-13 16:16:56 so what do you do for 32-bit platforms 2022-11-13 16:16:59 anything that mixes libc::time_t with a crate that exposes native things it breaks 2022-11-13 16:17:03 nothing, the build fails 2022-11-13 16:17:07 most things don't do things like that 2022-11-13 16:17:11 but for ones that do it's broken 2022-11-13 16:17:12 the compiler works? 2022-11-13 16:17:21 compiler works afaict 2022-11-13 16:17:24 ok 2022-11-13 16:17:37 i'm just checking in case i actually decide to add 32-bit powerpc in chimera eventually 2022-11-13 16:17:57 though that might not happen 2022-11-13 16:17:58 idk yet 2022-11-13 16:19:28 i kinda both want to do that and at the same time really don't want to 2022-11-13 16:19:30 the only thing broken is libc::time_t so anything that actually decides to use that 2022-11-13 16:19:38 it is almost nothing 2022-11-13 17:45:11 Will be shutting down the arm/aarch builder now 2022-11-13 17:45:23 so it won't be available for the majority of next week 2022-11-13 17:51:40 okie 2022-11-13 21:03:33 psykose: thanks for the pointer! That indeed fixed the issue with compiling c2rust with the rustup toolchain. Let's see if I somehow also get the actual application I want to build buiding :) 2022-11-13 21:04:08 this is probably the 15th time i have seen someone fail to do something with the rustup toolchain 2022-11-13 21:04:11 and i had to tell them to do that 2022-11-13 21:06:34 i wish i knew how to at least mitigate it upstream 2022-11-13 21:06:51 ours defaults to -crt-static but even with +crt-static it also works because i fixed some link args 2022-11-13 21:07:04 i don't know how the defaults work for anyone trying to use it, i think they just don't 2022-11-13 21:07:37 maybe they'd accept an ifdef soup version of the changes but meh 2022-11-14 05:50:40 Hi, OSV developers would appreciate someone with Alpine development knowledge taking a look at https://github.com/ossf/osv-schema/pull/88 2022-11-14 06:20:16 Caesar: what specific information would you need? 2022-11-14 07:09:06 ikke: I think they want to make sure what they're specifying looks right to an actual Alpine developer 2022-11-14 11:53:39 Caesar: Alpine versions are commonly specified both with and without v. also note that alpine has edge branch similar to debian unstable: roughly the package set planned for next stable release 2022-11-14 13:54:35 Hi all. 2022-11-14 13:54:35 Is there any technical issue/incompatibility for not having firewalld packaged ? Or is it just that nobody packaged it ? 2022-11-14 14:51:33 doesnt firewalld depend on systemd? 2022-11-14 14:55:00 It doesn't seem do 2022-11-14 14:55:10 so* 2022-11-14 14:59:34 Gentoo provides an openrc init file: https://gitweb.gentoo.org/repo/gentoo.git/tree/net-firewall/firewalld/files/firewalld.init?id=af1a0e74155be7d67cf7e0d7109c795e276fa78f 2022-11-14 15:23:56 looks like it should be possible to build/use it with systemd. So I guess its just that nobody has packaged it yet 2022-11-14 15:24:40 s/with/without/? 2022-11-14 15:26:16 s/with/without/g 2022-11-14 15:26:21 indeed 2022-11-14 15:26:57 shouldn't the bot fix messages right now? 2022-11-14 15:27:18 apparently it's awol 2022-11-14 15:29:06 test 2022-11-14 15:29:10 s/test/test2/ 2022-11-14 15:29:14 at last 2022-11-14 15:29:20 or mia 2022-11-14 15:30:26 welcome back alpine-meetbot 2022-11-14 15:30:26 s/awol/back again/ 2022-11-14 15:30:31 s/awol/back again/ 2022-11-14 15:30:37 test 2022-11-14 15:30:39 s/test/working 2022-11-14 15:30:39 ikke meant to say: working 2022-11-14 15:30:54 psykose: your peace is over :p 2022-11-14 15:30:58 grr 2022-11-14 15:31:21 psykose: what should I use instead of removed py3-setuptools-scm-git-archive ? 2022-11-14 15:31:27 nothing 2022-11-14 15:31:36 setuptools scm does the same by default 2022-11-14 15:32:36 for fancy verysmart python scripts that check you can just patch it out 2022-11-14 15:33:18 py3-ansi2html complains at missing deps. It has setuptools_scm_git_archive stated as dep in its pyproject.toml 2022-11-14 15:34:00 phunny 2022-11-14 15:38:14 gpep517 doesn't even check those so you don't have to do anything 2022-11-14 15:39:59 Oic 2022-11-14 15:41:25 I guess gpep517 replaces py3-build and py3-installer? 2022-11-14 15:42:39 it uses py3-installer 2022-11-14 15:43:08 but in most apkbuilds py3-installer is used directly anyway 2022-11-14 15:51:28 Oic, thank you. 2022-11-14 19:15:41 "Cogitri: I have an issue with..." <- Do you have -crt-static defined in your RUSTFLAGS? 2022-11-14 19:16:15 It probably dies because the static `musl` in the rustup-rustc conflicts with the dynamic `musl` linked to `libclang` 2022-11-14 19:16:33 Cogitri: psykose already mentioned that and that solved it :-) 2022-11-14 19:19:40 Yes thx :), That was indeed the issue. psykose thankfully already pointed me to the solution :) 2022-11-14 19:20:05 Ah, grand :) 2022-11-15 00:13:02 Thanks Hello71 2022-11-15 06:16:38 Caesar: this might be helpfull: https://alpinelinux.org/releases.json 2022-11-15 11:04:46 Hello all! https://gitlab.alpinelinux.org/ -> Gateway Timeout 2022-11-15 11:05:03 SecT0uch: thanks, we are looking at it 2022-11-15 11:47:41 SecT0uch: It's back now 2022-11-15 12:09:26 Thanks :) 2022-11-15 15:42:25 https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/41404 -> could that be approved please? 2022-11-15 22:02:55 hi, are the arm builders offline and if so, will arm packages be automatically be rebuilt for current versions when they come back? 2022-11-15 22:04:00 trying to do a version bump for a non-urgent package, wondering whether to leave MR as draft and wait 2022-11-15 22:07:41 yes, they are being moved atm 2022-11-15 22:08:00 and yes, they will build the packages / changes that have been pushed since then 2022-11-15 22:08:10 ikke: awesome, thanks 2022-11-15 22:10:04 note that that counts for the builders 2022-11-15 22:10:33 for ci jobs in MRs, you'd have to rebase your branches after they have been enabled again 2022-11-15 22:17:05 ikke: does that mean a new MR with something like, "re-enable arm builds" even if it is the same version? 2022-11-15 22:24:06 ah, I'll just MR again on the next version bump and the builders will probably be back by then 2022-11-15 22:24:22 (nvm, thanks) 2022-11-16 03:28:56 gosh sometimes busybox is so bad... sed -i not checking it actually wrote its output file properly :| 2022-11-16 03:28:59 http://lists.busybox.net/pipermail/busybox/2022-November/089964.html 2022-11-16 03:29:32 I guess that somewhat qualifies for cherry-picking into aports? 2022-11-16 03:52:11 (didn't wait an opened https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/41438 ) 2022-11-16 04:00:11 nmeum: ^ 2022-11-16 07:29:27 ugh... thats bad 2022-11-16 08:54:00 patch looks fine, just merged it. thanks Asmadeus! :) 2022-11-16 09:12:29 Thanks! :) 2022-11-16 10:08:53 Can someone please take a look at !41311 ? I would really like to have this in 3.17 2022-11-16 10:24:08 nmeum: you just had to write this wouldn't get reviewed, and Denys just did! He made a fair point (fclose won't error if there's nothing buffered on musl), and I just sent a v2... for sed I wouldn't think there is any case where we don't have anything buffered (and failing to create the file -is- caught), so I don't think there's any rush to update to the v2, but I'll open a new MR when it gets 2022-11-16 10:24:14 merged at least so we don't diverge more than we need to 2022-11-16 10:25:07 great that Denys found the time to review it 2022-11-16 10:26:14 would be cool to update the patch in aports as well but no need to rush it 2022-11-16 10:26:39 happy to do it now if you prefer, I've tested it on alpine so both versions work (for the case I fell into at least) 2022-11-16 11:33:50 Asmadeus: thank you for following it up. I can review an updated patch it if nmeum does not have time for it now 2022-11-16 11:35:04 I can do it later today, I was just thinking that if Denys is reviewing it anyhow we might as well wait until it lands in busybox master before we have to update the patch again for a potential v3 :D 2022-11-16 11:35:10 but pushing the v2 is also fine with me, up to you 2022-11-16 11:39:49 I've opened https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/41443 for v2, but I agree -- let's give him a day or so 2022-11-16 15:36:20 i think I have fixed all apk dot --errors in edge now 2022-11-16 15:36:32 i think we should start write the release ntoes for 3.17 2022-11-16 15:37:23 could we have a place to write them for 3.18 right after 3.17 release too 2022-11-16 15:37:31 i kind of forgot everything i wanted to note down 2022-11-16 15:38:25 i think we normally have something in wiki 2022-11-16 16:00:04 release notes: https://gitlab.alpinelinux.org/alpine/infra/alpine-mksite/-/merge_requests/54 2022-11-16 16:01:59 ncopa: line 2 has 3.16.0 instead of 3.17.0 2022-11-16 16:02:28 for more detailed notes: https://wiki.alpinelinux.org/wiki/Release_Notes_for_Alpine_3.17.0 2022-11-16 16:03:21 minimal: thanks. fixed now. 2022-11-16 17:11:02 Hello all. I'm having an issue (musl bug ?): 2022-11-16 17:11:02 OSError: protocol not found 2022-11-16 17:11:02 $ python -c 'import socket; print(socket.getprotobyname("dccp"));' 2022-11-16 17:11:02 dccp is present in /etc/protocols 2022-11-16 17:33:46 Never mind, it seems they are hardcoded in musl 2022-11-16 18:19:52 looks like execinfo.h is a glibc thing, but `man backtrace` still reports I can use it 2022-11-16 18:20:04 is this just a standard linux manpage that doesn't apply to alpine? 2022-11-16 18:20:19 what's the recommended approach for building software that requires execinfo.h? 2022-11-16 18:20:51 patch it out 2022-11-16 18:21:16 is 3.17 already tagged? i might be a bit late with the wlroots bump 2022-11-16 18:22:20 only rc1 2022-11-16 18:26:26 we have wlroots 0.16.1 on the way 2022-11-16 18:26:39 idk if its best to stay with 0.15 as the main package until the release 2022-11-16 18:27:07 or do it anyway, i snuck a patch for 0.15.1 2022-11-16 18:27:28 or i can patch 0.15.1 in 3.17, whatever floats your goats 2022-11-16 18:56:37 psykose: !41290 is good to go 2022-11-16 20:49:35 arm builder is back 2022-11-16 22:29:18 hello, I've problem with some python packages - those are generating wheels with version 0.0.0 but not using setuptools_scm, anyone knows how can I handle that? Issue is about !41397 2022-11-17 07:12:11 ncopa: is it correct that alpine:3.17 already exists on docker hub? 2022-11-17 07:13:53 correct. i pushed it yesterday 2022-11-17 07:15:52 but it is not alpine:latest yet 2022-11-17 07:16:59 Ah ok. I was surprised as 3.17 itself was not released yet 2022-11-17 07:17:12 i shoud have mentioned it yesterday, sorry 2022-11-17 07:17:26 the idea was to give a preview so people can start test it 2022-11-17 07:18:05 maybe it should be tagged alpine:3.17-rc1 2022-11-17 07:18:18 We have automation to automatically create MRs for upgraded images, so it now suggests to upgrade everything to 3. 17 2022-11-17 07:18:46 There is an rc as well 2022-11-17 07:19:39 https://github.com/docker-library/official-images/pull/13548/commits/fd8acfb753dba590ada38b915b84e1e47003c935#diff-8c805a48d8543a1fe5cdcee8aa34f23366aa01f27b7f2e276df99901bbac86a9R17 2022-11-17 07:19:49 ikke: oh cool :) 2022-11-17 07:20:30 there shouldnt be any huge or breaking changes at this point, but there are exceptions 2022-11-17 07:21:19 we might be able to do rc2 today 2022-11-17 07:21:33 and the final 3.17.0 next week 2022-11-17 07:26:33 ncopa: RenovateBot 2022-11-17 07:26:46 Similar to dependabot 2022-11-17 07:27:20 any reason you're not using dependabot? afaik, it supports Docker images 2022-11-17 07:28:40 or is that on non-github platform 2022-11-17 07:32:10 Not sure, I have not set it up 2022-11-17 07:32:49 I think for gitlab, it's a separate project 2022-11-17 07:33:20 FYI, I'm talking about $dayjob 2022-11-17 07:39:49 👍 2022-11-17 14:49:29 Upstream for a package I maintain has released a new version. It's called 0.5, but the tag for it on github is 1.5. Wondering what the best way of handling this is? I've submitted an issue in the github repo, but there's been no response. 2022-11-17 14:49:32 https://github.com/dabeaz/sly/blob/master/setup.cfg 2022-11-17 14:49:39 https://github.com/dabeaz/sly/tags 2022-11-17 14:49:44 https://github.com/dabeaz/sly/issues/105 2022-11-17 14:50:48 I suppose use 0.5 as pkgver and override it for the rest 2022-11-17 14:51:05 if it turns out to be 1.5, you can just bump it 2022-11-17 14:51:24 v0.5 is on PyPi, with a note saying it will no longer be maintained there. only on Github. 2022-11-17 14:51:57 I currently pull from a specific commit on Github, so I guess I could continue to do that. 2022-11-17 14:52:13 There wasn't a tag for v0.4 for example. 2022-11-17 15:43:45 Once upon a time I had my name misspelled in the maintainer line of an APKBUILD, and since then pkgs.a.o has always remembered it even though no package has it anymore. Could "Bart Ribberse" be removed from it? 😅 2022-11-17 15:49:09 PureTryOut: done 2022-11-17 15:50:05 Thanks! 2022-11-17 16:28:21 ok, i think we could probably do rc2 tomorrow 2022-11-17 16:49:41 I'm planning to raise MR for cloud-init version upgrade in a matter of hours which I'd like to get into 3.17 if possible 2022-11-17 17:00:50 👍 2022-11-17 17:08:00 I would like to move 'nyx' and 'py3-stem' to community but I should fix the checks first (it does wrong "setup.py check" now), is still possible for 3.17? 2022-11-17 17:14:26 donoban: if you can do it within one week or so 2022-11-17 17:16:33 ok, I'm pretty bussy this days but I hope to do it in that timeframe 2022-11-17 21:45:35 /etc/conf.d/ would generally not have +x set on files ? 2022-11-17 21:47:33 could someone check if mosquitto install sets /etc/conf.d/mosquitto properly 2022-11-17 22:01:41 apk fetch —stdout mosquitto | tar -ztv | grep conf 2022-11-17 22:32:36 apk add mosquitto; ls -l /etc/conf.d/ | grep mosquitto 2022-11-17 22:34:42 https://tpaste.us/ 2022-11-17 22:34:48 https://tpaste.us/Rxn5 2022-11-17 22:36:10 https://git.alpinelinux.org/aports/tree/main/mosquitto/APKBUILD?h=3.16-stable#n90 2022-11-17 22:38:19 #n89 ^ 2022-11-17 22:39:42 755 in master 2022-11-17 22:52:42 just for info (reply to iw earlier), mobiles have hotspots, it could help in scripts testing 2022-11-17 23:31:42 psykose: do you want me to put the wlroots0.15 patch for 3.17? 2022-11-17 23:31:51 that one looks fine yeah 2022-11-17 23:31:56 ita fixes a couple of bugs on wlroots based compositors 2022-11-17 23:32:06 how much time left until 3.17 2022-11-17 23:32:14 a week 2022-11-17 23:32:50 ill make the pull request for the patch tomorrow, its bath night for the goblins 2022-11-17 23:42:22 having bathed my brother exactly once, i don't envy you 2022-11-18 01:16:29 "There has been a runner system failure, please try again"? 2022-11-18 06:47:25 clandmeter: powered off nld5 2022-11-18 07:15:12 vkrishn: thats a bug indeed. thanks 2022-11-18 07:23:13 i have a super weird issue today. busybox ifup lo no longer work 2022-11-18 07:24:27 ifup: too few parameters for line "iface" 2022-11-18 07:24:46 auto lo 2022-11-18 07:24:46 iface lo inet loopback 2022-11-18 07:24:56 with ifupdown-ng it works 2022-11-18 07:25:14 i can also not see any relevant changes in busybox recently 2022-11-18 07:26:36 i havent touched my /etc/network/interfaces 2022-11-18 07:35:18 i must have had ifupdown-ng installed and for some reason it got uninstalled i guess 2022-11-18 14:55:23 mozjs102 tests deadlocked on build-3-17-armv7 and build-edge-x86 2022-11-18 14:55:52 so something is broken in mozjs102, maybe only on 32bit 2022-11-18 20:44:26 ncopa: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/41532 is to some degree undoing the limiting of supported hosts you did some time ago, as it re-enables building for x86. It comes however with a twist: I classify ARM, AVR and RISC-V as "mainstream" microcontrollers, and the rest of the architectures (openrisc, MSP430, aarch64) as "exotic". Support for exotic architectures is limited to x86_64 and aarch64. So it will 2022-11-18 20:44:26 reduce the load on the PowerPC builders (that previously also supported exotic architectures). 2022-11-18 20:45:13 patches to https://lists.alpinelinux.org/~alpine/aports not being considered yet, been a while ? 2022-11-18 20:47:19 Note: aarch64 is correctly classified as exotic in this setting. It may be mainstream for systems that run a "proper OS" (such as Alpine). But bare-metal targets without an OS are typically MMU-less microcontrollers with << 1 MiB of RAM. This combination with aarch64 is indeed exotic. (But firmware development for aarch64 would also require a bare-metal toolchain. But then again, I expect that to happen on x86_64 and aarch64.) 2022-11-18 20:50:13 I would be extremely happy if I could tomorrow install some bare-metal toolchain on x86, as won't be able to carry my regular machine with me tomorrow and the only machine available will be an old notebook powered by an 1 GHz 32 bit Celeron CPU. 2022-11-18 21:25:25 vkrishn: maybe the import from mainlinglist to gitlab is broken? 2022-11-18 21:25:42 ncopa: it is 2022-11-18 21:25:44 for some time now 2022-11-18 21:26:00 and nobody has time to fix it 2022-11-18 21:28:22 maribu: i dont mind, as long as the CI passes and the build/merge is done over the weekend 2022-11-18 21:28:41 i mean, as long as it does not gets in the way of the 3.17 release 2022-11-19 16:31:17 For the build errors on armv7 and aarch64 of dotnet7-stage0, they should clear-up after a retry. I'm reporting the bug upstream. 2022-11-19 17:06:34 ayakael: issue is the use up quite some space 2022-11-19 17:06:36 20G per build 2022-11-19 17:06:43 and they all share the same host 2022-11-19 17:06:55 so 4 of them building at the same time eat up 80G 2022-11-19 17:07:57 So I make sure only one or two build at the same time 2022-11-19 17:08:12 Right, that would explain that. I already run with '/p:CleanWhileBuilding=true' which cleans-up after every build step, but indeed that means that instead of using 60G the build uses 20G. 2022-11-19 17:11:58 build-3-17-armv7 is building now 2022-11-19 17:12:28 after it finished, I'll start build-edge-aarch64 again 2022-11-19 19:14:13 psykose, ncopa: Thx for the embedded toolchain for x86 :) 2022-11-19 19:17:36 It is an interesting experience to sit again at the old machine I was using as pupil 15 years ago. It was already running Alpine, I forgot that I installed it. It was Alpine edge with GCC 8.3, so rather quite some time ago. 2022-11-19 19:20:06 I am a bit let down that according to /proc/cpuinfo even that old machine from 2003 (which I recall to be outdated at that point in time) is in fact vulnerable to meltdown. I always assumed it to be too old for that :-D 2022-11-19 19:21:45 I am now rebuilding linux-lts with CONFIG_FB_TRIDENT=m on it. Let's see if it can handle the build without OOM over night :D 2022-11-20 07:27:48 I underestimated how long building a kernel takes on old machines :D 2022-11-20 07:28:12 ayakael: dotnet7-build fails on aarch64 and armv7 with "EXEC : error : Multiple input files matching same simple name" 2022-11-20 13:46:53 Hello, I tried to upgrade from 3.16 to "3.17" today, but got some cyclic dependency and I solved it by first uninstalling easyeffects. Then during the upgrade there were issues with py3-uinput... tpaste: https://tpaste.us/NONB 2022-11-20 13:47:39 Nothing pinned in /etc/apk/world and all packages were from 3.16 (nothing pinned or installed from edge) 2022-11-20 17:51:23 is ppc64le still a supported arch for Alpine? 2022-11-20 17:54:22 yes 2022-11-20 17:54:34 how so? 2022-11-20 17:55:28 ikke Copy that, investigating 2022-11-20 17:56:57 ikke: just added a new package and the tests failed only on ppc64le so was wondering whether to exclude the testing for ppc or to just exclude ppc as an arch for package 2022-11-20 18:51:35 ikke It looks like a known-issue without a clear fix yet, https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/41593 looks to fix another issue that might fix this one. 2022-11-20 18:56:43 ikke It looks like a known-issue without a clear fix yet, https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/41593 looks to fix another issue that might fix this one. 2022-11-20 18:56:49 Woops sorry 2022-11-20 18:56:52 heh :) 2022-11-20 19:03:16 Anyway I can make it easier for you with the storage issue? 2022-11-20 19:04:13 Right now, there are 2 builders per arch tracking master, so it means it doubles the space requirements until 3.17 is released 2022-11-20 19:04:57 ok good, yes indeed after it'll be easier as update on edge wont update 3.17 at the same time 2022-11-20 19:06:09 oh misunderstood, thought ya'll had 2x the builders 2022-11-20 19:07:31 Each builder is an lxc container 2022-11-20 19:07:47 where armhf, armv7 and aarch64 all share the same host 2022-11-20 19:16:53 ayakael: hmm, failed on x86_64 2022-11-20 19:18:05 similar error as on the builders, it appears 2022-11-20 19:21:50 Indeed, I reckon it's due rebuilding with `dotnet7-build` rather than the initial `dotnet7-stage0-bootstrap`. I'm forcing use of the latter to confirm. 2022-11-20 19:21:53 due to* 2022-11-20 19:29:23 If it is the case, it's like because of the RID issue that this MR actually looks to fix, cause I had tested rebuilding dotnet7 with itself previously and didn't have that issue. 2022-11-20 19:29:27 likely* 2022-11-20 19:51:25 That did it. For this next one, I'm gonna force use of build with stage0, next dotnet7 upgrade I'll go back to using itself to build. 2022-11-20 20:09:54 ayakael: so we can merge the MR once it's green? 2022-11-20 23:12:15 ikke Yup, should work :) 2022-11-21 04:45:46 Tested dotnet7 with the new powershell version on edge and it's looking good. RID issues are fixed. Thanks for y'all's help :) 2022-11-21 07:53:29 Quick question: To figure out if it's a problem with the upcoming 3.17 or just my installation. After upgrading to 3.17 GL aceleration inside libvirt QEMU VMs stopped working. GL works just fin in the host running wayland. Any known issue or config change I've missed? 2022-11-21 07:55:06 EvTheFuture: what did you upgrade from? 2022-11-21 07:55:18 might be some qemu module got renamed or moved to a subpackage 2022-11-21 07:58:12 ncopa: 3.16 fully updated uesterday 2022-11-21 07:59:26 To clarify, the VM starts but no GL acceleration inside it. glxgears push around 4 fps. When reverting back to 3.16 it works just fine. 2022-11-21 08:00:03 As a side not the other problem i face is that for some reason the background loaded in sway doesn't show up after the upgrade. 2022-11-21 08:01:01 Yeah, and there seem to be a problem that I need to reinstall some packages and also reinstall font-awesome.in order to get the font loading in waybar properly like before the upgrade. 2022-11-21 08:01:27 https://tpaste.us/NONB 2022-11-21 09:00:51 @ncopa: I looked around but didn't find anything obvious when searching with apk search. 2022-11-21 09:21:02 !41599 2022-11-21 11:35:04 this PR has been assigned to me, what's the expectation here?: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/41526 2022-11-21 11:35:21 I can't seem to rebase or approve a merge, do I just comment LGTM or something? 2022-11-21 11:35:49 Yes, that'll do 2022-11-21 14:23:30 Hello, could someone review !41598 and tell if I did a good job please ? 2022-11-21 15:41:38 nothing to add for release notes for 3.17? https://wiki.alpinelinux.org/wiki/Release_Notes_for_Alpine_3.17.0 2022-11-21 15:58:43 raspbeguy: sorry but im prioritizing the 3.17 release for now :-/ 2022-11-21 16:03:50 I don't seem to see anything on https://wiki.alpinelinux.org/wiki/Release_Notes_for_Alpine_3.17.0 is there a draft somehwere? 2022-11-21 16:09:05 ayakael: it is an invitation for everybody to add whatever they think should be mentioned in the release notes 2022-11-21 16:09:12 its a wiki 2022-11-21 16:09:46 compare with this: https://wiki.alpinelinux.org/wiki/Release_Notes_for_Alpine_3.16.0 2022-11-21 16:10:25 Gotcha 2022-11-21 16:10:27 this is the draft: https://gitlab.alpinelinux.org/alpine/infra/alpine-mksite/-/merge_requests/54 2022-11-21 16:33:05 ncopa: that's ok, no biggies. I secretly hoped that it would make it to the next release, guess it will wait 6 more months. 2022-11-21 16:54:04 the dotnet7? it is in 2022-11-21 16:54:20 we branch 3.17-stable when 3.17.0 is released 2022-11-21 16:56:38 Yup, it is in master/community and is available in the 3.17 repo currently. 2022-11-21 16:56:53 It wasn't added in in time for rc3 though, so that's why it isn't there 2022-11-21 17:18:40 hello, can I ask some python packager to help me? 2022-11-21 17:18:41 !41397 2022-11-21 17:19:02 I'd like this MR to be included in 3.17 so I need it pretty fast to be solved 2022-11-21 18:16:18 JuniorJPDJ: so for some reason it does not detect that a modified file has been modified on s390x? 2022-11-21 18:29:51 This is one issue which is new after rebase, second one is versioning 2022-11-21 18:29:53 ikke: 2022-11-21 18:30:09 this one I already left comment under MR 2022-11-21 18:41:06 JuniorJPDJ: py3-watchgod has a version.py file which is hardcoded to 0.0.dev0 2022-11-21 18:41:14 I assume on release they replace that 2022-11-21 18:42:27 py3-watchfiles has _rust_notify.py which seems to grab information from cargo 2022-11-21 18:47:42 what do you propose as a solution then? 2022-11-21 18:47:46 ikke: 2022-11-21 18:48:07 As I've no idea.. :x 2022-11-21 18:49:33 printf '__all__ = ("VERSIION")\nVERSION="%s"\n' $pkgver >watchdog/version.py 2022-11-21 18:49:35 in prepare() 2022-11-21 18:53:19 JuniorJPDJ: py3-watchfiles is going to be a bit more tricky 2022-11-21 19:00:27 Posted something in that thread for both 2022-11-21 19:05:20 ncopa: you probably missed riscv docker container release, there's edge for riscv but no 3.17_rc1 2022-11-21 19:05:35 ikke: thanks, I'll take a look in a few minutes and try to apply a fixes 2022-11-21 19:06:52 JuniorJPDJ: 3.17 still tracks master, so whatever is pushed for edge, is also built for 3.17 2022-11-21 19:07:44 note that we do not have a 3.17 release for riscv 2022-11-21 19:10:59 I mean the official docker image which got released here: https://github.com/docker-library/official-images/pull/13548/commits/fd8acfb753dba590ada38b915b84e1e47003c935# 2022-11-21 19:12:15 same, there is no 3.17 docker image for rv64 2022-11-21 19:12:34 so it was on purpose, ok, makes sense 2022-11-21 19:12:52 We don't consider riscv64 release-ready yet 2022-11-21 19:12:57 so we only build edge 2022-11-22 09:15:07 morning! lets do 3.17.0 today! 2022-11-22 09:16:40 \o/ 2022-11-22 09:43:40 hi all, what means -5 error in apk verify? 2022-11-22 09:44:06 apk add mariadb-common && apk verify /etc/my.cnf 2022-11-22 09:46:21 ncopa, \o/ 2022-11-22 09:46:36 \o/ 2022-11-22 11:14:41 \o\ 2022-11-22 11:15:05 /o/ 2022-11-22 11:26:55 indy: I think that it expects a package (*.apk) file 2022-11-22 11:27:58 it verifies the integrity of a package 2022-11-22 11:28:53 [donoban@localhost][/var/cache/apk]% apk verify openssl-3.0.5-r2.ab61e13f.apk 2022-11-22 11:28:55 openssl-3.0.5-r2.ab61e13f.apk: 0 - OK 2022-11-22 12:08:35 donoban, i hoped it will do something like "rpm -V packagename" 2022-11-22 12:10:28 indy: i believe that is apk audit 2022-11-22 12:14:52 indy, you're looking for apk manifest 2022-11-22 12:15:01 (maybe) 2022-11-22 12:16:02 apk manifest only outputs checksums of files that are part of the package 2022-11-22 12:16:15 it doesn't verify 2022-11-22 12:20:57 panekj, thresh apk audit prints packages with changed files, apk manifest is closer to what i want 2022-11-22 12:21:07 donoban, panekj thresh thanks 2022-11-22 13:10:37 so, anything else that should be mentioned in the release notes? 2022-11-22 13:17:53 oh... I almost forgot, openssl3 2022-11-22 13:39:19 perl 5.36 2022-11-22 13:40:53 are there any deprecations in alpine 3.17? 2022-11-22 13:44:14 @ncopa: Any suggestions would be highly appreciated on what package might be missing. I've made a list of available and installed packages of various names here: https://tpaste.us/BkP9 2022-11-22 13:49:28 WARNING: Ignoring https://dl-cdn.alpinelinux.org/alpine/v3.16/main: No such file or directory 2022-11-22 13:49:35 🤔 2022-11-22 13:50:41 EvTheFuture: im a bit busy now. How Can I reproduce here locally? 2022-11-22 14:00:46 @ncopa: No worries, I'm not in any hurry, just wanted to make sure we fix it if it's a problem with the upcoming 3.17 and not just with my installation. 2022-11-22 14:01:29 EvTheFuture: could you please create an issue in gitlab? With the details how I can reproduce. 2022-11-22 14:01:30 @ncopa: Install libvirt and qemu, create a VM configured for 3D acceleratoin 2022-11-22 14:01:44 what Do I install as guest? 2022-11-22 14:02:01 @ncopa: I use Ubuntu in this particular case 2022-11-22 14:02:04 how do I configure guest for 3D acceleration? 2022-11-22 14:02:23 @ncopa: I'll make an example XML for you 2022-11-22 14:02:23 eg, spice? vmware? qxl? 2022-11-22 14:02:44 woudl be great if you could post the xml in an issue 2022-11-22 14:02:52 i use libvirt myself 2022-11-22 14:03:07 so it shouldnt' be difficult to reproduce it for me 2022-11-22 14:05:10 Shall I create the issue for libvirt or qemu or 3.17 in general? 2022-11-22 14:07:22 issue in aports, qemu I suppose 2022-11-22 14:15:42 Can someone help me proofread? https://wwwtest.alpinelinux.org/posts/Alpine-3.17.0-released.html 2022-11-22 14:18:38 ncopa: I would say "via *the* openssl1.1-compat package", but apart from that, lgtm 2022-11-22 14:20:48 ncopa: Maybe add links to KDE as well? https://kde.org/announcements/plasma/5/5.26.0/ 2022-11-22 14:21:51 good suggestions. thanks! 2022-11-22 14:23:59 is "KDE Applications" correct? or should it be "KDE Gear"? 2022-11-22 14:31:04 ncopa: I wonder if we need to mention it at all? Isn't KDE Plasma enough and the applications are kind of implied anyway? I dunno since I'm not a KDE user myself. 2022-11-22 14:32:02 yeah, thats what I'm thinking as well. It was requested to add in the 3.16 release, so I just copied it over 2022-11-22 14:32:59 removed 2022-11-22 14:37:33 ncopa: 👍 2022-11-22 15:07:46 ncopa: I've created issue: 14364 Let me know if I missed anything. 2022-11-22 15:15:41 !41648 2022-11-22 15:17:51 Congrats everyone on the release! Thanks for all the hard work! 2022-11-22 15:19:35 ncopa: re the change to mkimg.standard.sh to add kea - was this intended to relate to DHCP *client? Kea is a server, ISC don't now have a replacement for dhclient (their old client) 2022-11-22 15:23:52 minimal: it is for dhcpd, the server, in dhcp package 2022-11-22 15:24:21 there is dhcpcd for client 2022-11-22 15:24:27 which is not from ISC 2022-11-22 15:25:33 I added a paragraph about migrate to Kea from dhcpd under "deprecation": https://wwwtest.alpinelinux.org/posts/Alpine-3.17.0-released.html 2022-11-22 15:27:55 ncopa: ah ok 2022-11-22 15:39:55 I have created https://wiki.alpinelinux.org/w/index.php?title=Release_Notes_for_Alpine_3.18.0. Please add stuff that should be mentioned there for next release. Thanks! 2022-11-22 17:05:26 oh, 3.17 has been branched offf. great work everyone \o/ 2022-11-22 18:55:04 will 3.17 be upgraded with the next linux-lts in december? 2022-11-22 18:55:49 why would it 2022-11-22 18:56:13 why wouldnt it 2022-11-22 18:56:32 you're asking so 2022-11-22 18:57:52 true 2022-11-22 18:57:57 I don't see why next kernel would be an exception to no non-security/bugfix rule 2022-11-22 18:58:28 new lts doesnt count as security? 2022-11-22 18:58:53 it's not security so no? 2022-11-22 18:59:02 alright 2022-11-22 18:59:43 assuming that by "next lts" you mean 6.x 2022-11-22 19:00:31 yeah 2022-11-22 19:00:34 previous lts will be maintained for next years so there's no need for upgrade 2022-11-22 19:01:07 anyway 3.18 is not so far away 2022-11-22 19:01:36 there is still 4.9 lts so I think we're safe 2022-11-22 19:12:13 https://pkgs.alpinelinux.org/ has no 3.17 yet, is it expected? 2022-11-22 19:12:30 DavidHeidelberg[m]: yes, it takes a bit before it has ingested all packages 2022-11-22 19:12:42 ikke: kk,thx! :) 2022-11-22 19:28:41 ncopa: I get this in the log file for the VM: "qemu-system-x86_64: warning: console: no gl-unblock within one second" and on th enext log line: "qemu-system-x86_64: warning: spice: no gl-draw-done within one second" 2022-11-22 20:15:28 is there a page for removed pkg for main, eg /posts/Alpine-3.17.0-changes.html ? 2022-11-22 20:16:38 No, we do not track that explicitly 2022-11-22 20:19:09 ok, thanks all for updated php-8.1.x, its helping me update my site 2022-11-22 20:20:02 kinda having fun with php-8.1.x, load full of changes to work on 2022-11-22 20:24:01 more fun note: I moved all my php related dev on a mobile phone running alpinelinux, :-) 2022-11-22 20:24:36 interesting 2022-11-22 20:27:45 my inverter served well for 8yrs, now don't feel like investing it in 2022-11-22 20:31:46 or fun way to put, ran out of big battery so moved to smaller one 2022-11-22 20:32:10 heh 2022-11-22 20:39:37 I lovingily call these devices 'black bricks' 2022-11-22 20:41:58 currently I don't think this might have been possible without alpinelinux 2022-11-22 21:41:34 are there problems with armv7 repo? 2022-11-22 21:41:34 build logs are available for my upgraded packages, but packages are not in repository 2022-11-22 21:41:51 example: https://build.alpinelinux.org/buildlogs/build-edge-armv7/community/py3-dirty-equals/py3-dirty-equals-0.5.0-r0.log 2022-11-22 21:42:38 another example is py3-watchfiles 0.18.1-r0 2022-11-22 21:45:29 JuniorJPDJ: caused by build failures holding the builder up 2022-11-22 23:43:29 gz on 3.17, nice work 2022-11-23 02:31:20 hello good night 2022-11-23 02:31:34 I just found that py3-slugify is giving error with the last update 2022-11-23 02:31:48 importlib.metadata.PackageNotFoundError: No package metadata was found for python-slugify 2022-11-23 02:32:37 that is the output taken from the /usr/bin/slugify 2022-11-23 02:33:20 and in python shell "import slugify from slugify" isn't finding module anymore 2022-11-23 06:35:59 Hello, now 3.17 has been released, I've got MRs that could use some reviews: !41641 2022-11-23 06:36:49 And also !41519 2022-11-23 06:38:24 If they are ok, I plan to create new MRs right after to move them to community 2022-11-23 08:24:49 Is there a faulty v6 server in the dl-cdn repository? 2022-11-23 08:25:21 for some reason sometimes fetching packages via it on v6-only networks just hangs 2022-11-23 08:29:51 dl-cdn is fastly 2022-11-23 08:30:58 so i should go scream at fastly instead i guess 2022-11-23 09:03:06 i wonder if it is our dns setup on fastly 2022-11-23 09:03:09 or our fastly config 2022-11-23 09:08:05 I have not noticed issues myself 2022-11-23 09:08:26 Not specific to ipv6 anyway 2022-11-23 09:08:51 i just noticed it here 2022-11-23 09:09:21 $ wget https://dl-cdn.alpinelinux.org/alpine/v3.17/releases/x86_64/alpine-extended-3.17.0-x86_64.iso.sha512 2022-11-23 09:09:21 Connecting to dl-cdn.alpinelinux.org ([2a04:4e42:38::645]:443) 2022-11-23 09:09:29 and there it hangs 2022-11-23 09:11:34 yeah 2022-11-23 09:11:39 it's random 2022-11-23 09:11:50 probs because there's multiple servers in rr? 2022-11-23 09:11:58 one of them works 2022-11-23 09:12:02 but pretty annoying 2022-11-23 09:20:49 Shouldn't python modules installed into `/usr/lib/python3.10/site-packages` depend on `python3>=3.10.0` and `python3<3.11.0`? 2022-11-23 09:21:45 sounds reasonable 2022-11-23 09:27:13 Also a lot of maintenance 2022-11-23 09:28:38 Couldn't `abuild` trace this as dependency like it does trace `.so` files? If an `/usr/lib/python3./site-packages` folder exists, automagically add `python3>=3..0` and `python3<3..0` as dep? 2022-11-23 09:33:44 it could but i am not sure if adding language-specific stuff is ideal 2022-11-23 09:34:01 maybe a py3 default function would be something 2022-11-23 09:34:06 or pyX 2022-11-23 09:36:59 I think if there would be a default function that `APKBUILD`s could call to add the deps and the linter would complain on that function not being called in any package that is named `py3-`, the `APKBUILD`s would add the deps when the modules are updated anyway over time. And when the logic to calculate the exact dependency version is in a central place rather than being reimplemented over and over in every APKBUILD, it should be 2022-11-23 09:36:59 maintainable, shouldn't it? 2022-11-23 09:37:27 yeah, sounds simple enough 2022-11-23 09:39:43 At least I would be extremely grateful if that would happen :) My system will remain a mess until the builders have completed all the python3 rebuilds ... :/ 2022-11-23 09:40:24 Would it be possible to maintain older version of ruby ? As we do for php 2022-11-23 09:40:41 or another approach is prefered and already discusses ? 2022-11-23 09:40:55 maribu: oh i know that feel 2022-11-23 09:43:22 staceee: someone needs to be willing to maintain it 2022-11-23 09:53:31 i updated my home router to alpine 3.17. seems like ipv6 does not work at all now. neither does dmvpn 2022-11-23 09:59:33 maybe default sysctl for v6 ra is off now? 2022-11-23 10:00:28 i have a feeling it is firewall or something 2022-11-23 10:00:49 oh, do you use awall? 2022-11-23 10:00:54 or nftables 2022-11-23 10:01:29 ping6 example.com 2022-11-23 10:01:29 PING example.com (2606:2800:220:1:248:1893:25c8:1946): 56 data bytes 2022-11-23 10:01:31 awall 2022-11-23 10:01:40 it hangs 2022-11-23 10:01:49 that is from the router itself 2022-11-23 10:01:56 oh 2022-11-23 10:04:03 is icmpv6 enabled? 2022-11-23 10:04:26 it's required for v6 functionality 2022-11-23 10:05:14 # ip6tables-save | grep icmp | tpaste 2022-11-23 10:05:14 https://tpaste.us/5kb0 2022-11-23 10:05:23 looks like it is 2022-11-23 10:05:43 -A OUTPUT -p ipv6-icmp -j ACCEPT 2022-11-23 10:05:57 also inbound 2022-11-23 10:06:06 but yeah 2022-11-23 10:06:09 seems fine 2022-11-23 10:06:20 i'm not sure what it could be to be fair 2022-11-23 10:06:45 good luck finding what caused it! 2022-11-23 10:06:45 inet6 fe80::fcaa:83ee:9075:****/64 scope link 2022-11-23 10:06:55 that's link-local 2022-11-23 10:07:05 i dont have link global then I guess 2022-11-23 10:07:05 do you have a global addr too? 2022-11-23 10:07:08 yeah 2022-11-23 10:07:08 nope 2022-11-23 10:07:12 that's it then 2022-11-23 10:08:01 how does your router get the subnet if its delegated from higher up? 2022-11-23 10:08:16 dhcpv6 2022-11-23 10:08:33 maybe that's what's missing 2022-11-23 10:08:40 or not working properly 2022-11-23 10:10:15 i had ipv6 enabled some time ago, but want re-enable it now 2022-11-23 10:10:24 might be it never worked properly 2022-11-23 10:12:28 well, the easiest way to do it is to get a address from your upstream, set the /64 on the router and distribute the addresses via RADVD 2022-11-23 10:12:48 i thought I had that enabled 2022-11-23 10:13:08 i remember fighting with this some time ago, but I dont remember how I made it work last time 2022-11-23 10:13:22 Nov 23 10:01:25 rtr-copa daemon.info dhcpcd[2241]: eth0: Router Advertisement from fe80::ae4b:c8ff:fe9a:f29e 2022-11-23 10:13:22 Nov 23 10:04:21 rtr-copa daemon.info dhcpcd[2241]: eth0: fe80::ae4b:c8ff:fe9a:f29e is unreachable 2022-11-23 10:13:26 well, i don't know what you do have enabled but that's the simplest way to do it in your case 2022-11-23 10:13:36 oh wow 2022-11-23 10:13:49 the upstream is unreachable? 2022-11-23 10:13:59 looks like that 2022-11-23 10:14:05 i wonder if it is my ISP doing weird stuff 2022-11-23 10:14:10 likely 2022-11-23 10:14:42 maybe they dont hand out 2 ipv6 subnets. I have the ISP provided WIFI router on a different port, for my home network 2022-11-23 10:14:52 my work network uses other port from ISP 2022-11-23 10:15:08 oh yeah 2022-11-23 10:15:14 that is likely the case 2022-11-23 10:15:45 in that case you can set a static address on the router that's identical to the one on the home network 2022-11-23 10:15:54 and forward all packets 2022-11-23 10:16:01 but your isp might do some filtering 2022-11-23 10:16:08 which would complicate this lots 2022-11-23 10:16:38 assuming that you can't reach their gateway 2022-11-23 10:16:44 this is likely the case 2022-11-23 10:16:58 let me disable ipv6 on the ISP router 2022-11-23 10:17:14 it might take a bit to get a lease though 2022-11-23 10:17:24 not sure if you can force invalidate the lease on their side 2022-11-23 10:17:30 so, my alpine router is not behind the ISP router, its "beside" from different ISP provided WAN port 2022-11-23 10:18:35 does your isp *support* that kind of setup at all? 2022-11-23 10:18:41 like in the technical way 2022-11-23 10:18:43 not physical 2022-11-23 10:19:07 for ipv4, yes 2022-11-23 10:19:14 ipv4 works 2022-11-23 10:19:17 hm 2022-11-23 10:19:50 or.... maybe they dont support ipv6 on all the ports? 2022-11-23 10:20:05 it could be 2022-11-23 10:20:57 the easiest way to check if it's supported would be pinging their gateway via link-local 2022-11-23 10:21:15 which likely results in the same case as the router-advertisement problem 2022-11-23 10:21:49 --- fe80::ae4b:c8ff:fe9a:f29e ping statistics --- 2022-11-23 10:21:49 5 packets transmitted, 0 packets received, 100% packet los 2022-11-23 10:21:54 weird you get the advertisements but sending the acknowledgement back is not working 2022-11-23 10:22:05 your isp is doing fancy things 2022-11-23 10:22:10 i remember fighting this some time ago 2022-11-23 10:22:34 and i remember i got it working somehow 2022-11-23 10:23:03 can you ping that address from a device in your home network? 2022-11-23 10:23:17 not now. I have disable ipv6 2022-11-23 10:23:24 on the isp router 2022-11-23 10:27:26 heh. the web interface on the isp router is weird 2022-11-23 10:27:31 Input value (fe80::ae4b:c8ff:fe9a:f29e) is an IPv6 address: ping6 2022-11-23 10:27:43 is the error message I get when clicking the ping6 button 2022-11-23 10:27:49 they apparently never tested it :) 2022-11-23 10:30:18 lmao 2022-11-23 11:09:32 lol! so I decided to give up ipv6 for now, and commented out it in my interfaces and dhcpcd.conf 2022-11-23 11:09:44 and when I restarted eth0 i got routeable ipv6 address 2022-11-23 11:09:55 what 2022-11-23 11:11:29 there was apparently 2 problems 2022-11-23 11:11:57 one was missing dhcpv6 support in awall. which I fixed last time i messed with ipv6 2022-11-23 11:12:01 https://gitlab.alpinelinux.org/alpine/awall/-/merge_requests/1 2022-11-23 11:13:30 the second issue was dhcpcd.conf. I had those two lines to handle out ipv6 in my lan: 2022-11-23 11:13:30 i remember when i wrote a script to merge stuff from nftables in a save script 2022-11-23 11:13:34 ia_na 1 2022-11-23 11:13:36 iptables* 2022-11-23 11:13:40 ia_pd 2 eth1 2022-11-23 11:14:05 not fun to work with 2022-11-23 11:20:26 does gitlab also suffer from the same fastly v6 problem? 2022-11-23 12:59:24 dunno 2022-11-23 12:59:42 im giving up on ipv6 ia_pd for now 2022-11-23 13:00:02 will try debug the qemu 3d graphics issue 2022-11-23 13:00:13 virgl? 2022-11-23 13:00:23 https://gitlab.alpinelinux.org/alpine/aports/-/issues/14364 2022-11-23 13:00:34 buti dont seem to have glxgears working on my desktop 2022-11-23 13:00:40 $ glxgears 2022-11-23 13:00:40 Error: couldn't get an RGB, Double-buffered visual 2022-11-23 13:02:16 no mesa-dri? 2022-11-23 13:04:45 !14364 : huh, this might be because of mismatching mesa versions on host and guest 2022-11-23 13:04:51 ah damnit 2022-11-23 13:05:06 #14364 * 2022-11-23 13:06:45 i have used a similar setup except i used xorg instead of wayland recently without problems 2022-11-23 13:06:48 on edge 2022-11-23 13:07:22 it might be that wayland allocates the framebuffer differently and that is a variable 2022-11-23 13:07:45 but the mesa versions seem like the entrypoint in debugging this 2022-11-23 13:26:37 humpf... vainfo shows I hoave no acceleration 2022-11-23 13:26:55 vainfo seems to not work yet 2022-11-23 13:26:59 with virgl 2022-11-23 13:27:06 this is on the host 2022-11-23 13:27:10 oh 2022-11-23 13:27:19 mesa-va-gallium? 2022-11-23 13:27:19 i dont even have 3d acceleration on my alpine host 2022-11-23 13:27:24 lmao 2022-11-23 13:27:40 $ glxinfo 2022-11-23 13:27:40 Error: couldn't find RGB GLX visual or fbconfig 2022-11-23 13:27:40 name of display: :0.0 2022-11-23 13:27:54 no idea why 2022-11-23 13:28:02 do you even have render nodes? 2022-11-23 13:28:22 render nodes? 2022-11-23 13:28:27 i have /dev/dri/* 2022-11-23 13:28:30 /dev/dri/renderX 2022-11-23 13:28:34 yup, that i have 2022-11-23 13:28:45 what card do you have? 2022-11-23 13:28:48 renderD128 2022-11-23 13:28:54 i9 comet lake 2022-11-23 13:28:59 i9 10900 2022-11-23 13:29:27 xorg uses modesetting 2022-11-23 13:29:36 xf86-video-intel 2022-11-23 13:29:41 do you have it? 2022-11-23 13:30:10 i think i disabled it due to xf86-video-intel not making releases for a half decade? 2022-11-23 13:30:33 no, i dont have xf86-video-intel 2022-11-23 13:31:14 i guess they consider it stable 2022-11-23 13:31:24 or maybe stuff's integrated into gallium now 2022-11-23 13:32:01 GLX is about acceleration afaik 2022-11-23 13:32:21 [ 6.767] (II) xfree86: Adding drm device (/dev/dri/card1) 2022-11-23 13:33:09 yeah seems like the new intel stuff is integrated as part of gallium/i915 2022-11-23 13:33:19 so mesa-dri-gallium should work better 2022-11-23 13:33:38 i have mesa-dri-gallium installed 2022-11-23 13:34:05 and glxinfo fails still? 2022-11-23 13:34:09 that's odd 2022-11-23 13:34:18 yup 2022-11-23 13:34:32 my Xorg.0.log my X https://tpaste.us/vyYL 2022-11-23 13:35:07 [ 6.821] couldn't get display device 2022-11-23 13:35:07 [ 6.821] (II) modeset(0): glamor initialization failed 2022-11-23 13:35:18 i dont know if glamor is relevant anymore? 2022-11-23 13:35:25 they change tech ever year 2022-11-23 13:36:58 [ 7.080] (II) AIGLX: Screen 0 is not DRI2 capable 2022-11-23 13:37:14 huh 2022-11-23 13:37:31 it seems to pick glamor because you are missing the intel module 2022-11-23 13:38:01 so my best guess is still using xf86-video-intel 2022-11-23 13:40:16 to my understanding that driver is no longer maintained? 2022-11-23 13:40:58 yeah but i guess there's no alternatives 2022-11-23 13:41:07 feel free to correct me if i'm wrong 2022-11-23 13:41:14 https://www.phoronix.com/news/Fedora-Xorg-Intel-DDX-Switch 2022-11-23 13:41:39 oh 2022-11-23 13:42:46 no commits since jan 2021 https://github.com/freedesktop/xorg-xf86-video-intel 2022-11-23 13:53:12 ok. i have no idea why i dont have any 3d accel or va 2022-11-23 14:13:11 look on the bright side, you have working 2d Xorg. 2022-11-23 14:13:37 the little eee-pc didn't even get that far after 3.17 upgrade 2022-11-23 14:16:31 Fwew, they recommend xf86-video-modesetting. 2022-11-23 14:25:41 Is there a way to view what service started another service using rc-status? 2022-11-23 14:25:48 I am trying to figure out what is bringing up networking 2022-11-23 14:26:19 I have also disabled it from the default runlevel so something is bringing it up separately 2022-11-23 14:36:00 dzilvys: its probably something that depends on "net" which "networking" provides 2022-11-23 14:36:16 dzilvys: how about grep "need" /etc/init.d/* | grep net ? 2022-11-23 14:39:36 minimal: Oh my god, thank you this has helped a lot 2022-11-23 14:40:13 ncopa: Right, so this will be a pain in the ass to stop networking from starting before it starts executing local.d scripts 2022-11-23 14:43:39 Is it possible to see when iptables get started then? I am worried that the firewall would be down before networking gets started 2022-11-23 14:46:12 dzilvys: which firewall? awall? iptables? nftables? they're also started by init.d script, iptables for example defines "before net" 2022-11-23 14:46:23 iptables 2022-11-23 14:46:30 Ah right wonderful 2022-11-23 14:47:43 is there a way to see the build queue in http://build.alpinelinux.org/ ? 2022-11-23 14:47:48 dzilys: iptables also defines "provide firewall" and some services such as chronyd define "after firewall" 2022-11-23 14:49:30 minimal: Cool, this helps a lot. 2022-11-23 15:14:54 if a CVE is assigned to the wrong package on security.alpinelinux.org should I declare the CVE as not applicable ("0:") in the APKBUILD or is there another way to "fix" this? 2022-11-23 15:15:25 https://security.alpinelinux.org/vuln/CVE-2020-36455 for example is about the slock Rust crate (which we don't package directly anyhow) but is assigned to the slock screen locking package 2022-11-23 15:37:16 lul 2022-11-23 15:42:22 nmeum: I think that there is a 'exceptions' repository 2022-11-23 15:42:42 https://gitlab.alpinelinux.org/ariadne/security-rejections 2022-11-23 15:46:00 Ariadne: ^ would it maybe make sense to move that repository to the /alpine/ namespace? 2022-11-23 15:46:12 yes 2022-11-23 15:46:37 Will arrange 2022-11-23 15:46:56 I think a rewrite is more applicable here 2022-11-23 15:52:11 rewrite? 2022-11-23 15:56:32 https://gitlab.alpinelinux.org/alpine/infra/docker/secfixes-tracker/-/blob/master/config/prod.settings.py#L44 2022-11-23 15:58:06 ah, sure that could work too 2022-11-23 15:58:46 though we don't package the slock rust crate so there isn't really something we could rewrite too 2022-11-23 15:58:54 s/too/to/ 2022-11-23 15:58:54 nmeum meant to say: though we don't package the slock rust crate so there isn't really something we could rewrite to 2022-11-23 15:59:44 also: if you move the security-rejections repo you might want to adjust the paths in SECURITY_REJECTIONS in the referenced file (hopefully gitlab does some sort of redirect too) 2022-11-23 18:52:19 are there plans for releasing risc-v in some stable version? is it documented somewhere? what is probable version which will see risc-v support? :D 2022-11-23 18:57:58 JuniorJPDJ: From what I gather, not any time soon 2022-11-23 18:58:14 At minimum, we want to have a native builder, which means we need some server-grade rv64 hardware 2022-11-23 19:11:18 ugh, will be hard to have server grade risc-v hardware 2022-11-23 19:11:40 the best we can get are 80$ 4core singleboards now 2022-11-23 19:12:14 yes 2022-11-23 19:34:03 JuniorJPDJ: I created a tracking issue a while ago #13269 with items that are important in my opinion, the most important one being the lack of check() runs (which likely requires native hardware to be resolved as pointed out already) 2022-11-23 19:35:10 ikke: regarding the slock cve: should I create an MR for the infra docker repo with a rewrite rule? if so what should I rewrite to? 2022-11-23 19:36:17 nmeum: I'd rewrite it to rust-slock 2022-11-23 19:36:25 even though it does not exist, it conveys what it means 2022-11-23 19:37:02 nmeum: and yes, feel free to make an MR 2022-11-23 19:37:09 ok, will do 2022-11-23 19:38:50 nmeum: nice! thanks for the pointer! 2022-11-23 20:39:22 ikke: ping 2022-11-23 20:39:39 ikke: I just got an UPS notification - is there any chance something fun is arriving in the name of Alpine Linux? 2022-11-23 20:39:50 telmich: might sure be 2022-11-23 20:39:56 We are expecting arm servers 2022-11-23 20:40:08 ikke: but you did not get a "has been sent" notification from anywhere, or have you? 2022-11-23 20:40:13 2 of them? 2022-11-23 20:40:36 And potentially involving AVANTEK? 2022-11-23 20:41:06 No, I have not received a notification myself 2022-11-23 20:41:12 at least, not a shipment notification 2022-11-23 20:42:31 I got a UPS notification that I cannot sort/put into any planned incoming deliveries, so those servers are the only ones that I have in my mind that could be them 2022-11-23 20:42:45 But maybe somebody else is sending an unannounced package 2022-11-23 20:43:02 Last I received was: "Shipment of servers is going through the paperwork. Will keep you posted on when it is done." 2022-11-23 20:43:18 That was Oct 5th 2022-11-23 20:43:56 ok 2022-11-23 20:44:28 I can respond and ask if they have happened to have shipped it already 2022-11-23 20:45:09 that would be much appreciated 2022-11-23 20:45:17 In case they give you an UPS tracking number, we can then compare 2022-11-23 20:46:21 I copy you in the thread 2022-11-23 20:46:24 You were already involved before 2022-11-23 21:04:28 Nice 2022-11-23 21:04:45 confirmed telmich 2022-11-23 21:05:34 Yes, those servers must be shipped by Avantek. 2022-11-23 21:06:14 No tracking number though 2022-11-23 21:07:26 Yes but shipper is same 2022-11-23 22:27:12 Tracking Number: 1ZY8662R6890799167 is what I have, clandmeter & ikke (ups: https://www.ups.com/track?loc=en_CH&Requester=NES&tracknum=1ZY8662R6890799167&AgreeToTermsAndConditions=yes&WT.z_eCTAid=ct1_eml_Tracking__ct1_eml_qvn_eml_resi_5scn&WT.z_edatesent=11232022/trackdetails) 2022-11-23 22:27:43 telmich: ok, I'll ask if they can confirm 2022-11-24 00:03:21 27.00 KGS i guess it could be some very big chocolates 2022-11-24 08:19:18 its probably the weight of our 3.17 release coming in 2022-11-24 08:29:58 :) 2022-11-24 09:21:03 morning. today my 3d accel works. a reboot was probably all that was needed after installing some random package 2022-11-24 09:58:51 I added python 3.11 to alpine 3.18 release notes: https://wiki.alpinelinux.org/wiki/Release_Notes_for_Alpine_3.18.0 2022-11-24 10:17:25 Some python packages looks outdated on edge repos (rebuilds). Known ? (sorry if you spam the answer her) 2022-11-24 11:28:46 staceee: the builders are still working on community, only once it's finished, the packages will be uploaded 2022-11-24 12:01:53 staceee: seems like community for most arches has been uploaded 2022-11-24 12:03:08 nice ! 2022-11-24 14:07:44 Hello. I am trying to create a custom alpine iso (aports mkimage) with a custom script that runs on boot, but have no clue on how to inject the script. Ant pointers? 2022-11-24 14:08:22 ACTION meant to say "Any pointers?" 2022-11-24 14:23:29 light-lead: runs on boot of the ISO? or on boot of an install made from the ISO? 2022-11-24 14:42:52 My script uses rclone, borgbackup and partclone to restore full computer images. 2022-11-24 14:43:57 The intention is to use alpine boot from ISO or initramfs, and run the script from ram. 2022-11-24 14:44:43 It should work kind of like clonezilla, but with some optimizations for our networks. 2022-11-24 14:51:04 apkovl may be useful here 2022-11-24 15:08:51 @Shiz, looking into it... THX 2022-11-24 16:46:47 you can use /etc/init.d/local service. `rc-update add local` and add your script to /etc/local.d/*.start 2022-11-24 19:32:57 hello good afternoon, anyone was able to compile and try labwc 0.6.0 on alpine? 2022-11-24 23:14:04 looks build-edge-ppc64le builder hangs 2022-11-25 07:29:18 hi all, off-topic question - any estimation when we can expect alpine 4.x ? :) 2022-11-25 07:29:36 what would it include? 2022-11-25 07:30:07 good question 2022-11-25 07:35:29 Only reason for a major version increment would be apk-3.x, that changes 'the thing that makes alpine alpine'. 2022-11-25 07:37:44 it isn't that much different 2022-11-25 07:38:03 And I hope that is still far away, as it apparently departs from 'an apk is just a compressed tar that can be handled by standard tools' 2022-11-25 13:09:35 I'm having trouble with apk. Installing a package like aspnetcore7-runtime works fine. When I then subsequently install nginx, apk will then install both nginx and aspnetcore6-runtime AGAIN. Does anyone have any idea why that is? Maybe /lib/apk/db/installed is not being updated? 2022-11-25 13:22:27 Is it possible that with the -p dirname switch, apk forgets to look at dirname/lib/apk/db/installed and instead looks at the hosts? 2022-11-25 14:48:16 sounds weird 2022-11-25 17:08:17 aspnetcore7 vs 6 2022-11-25 18:51:42 There are currently two supported versions of the dotnet, 6 and 7. Since dotnet7 just got released, AFAIK no packages directly depend on dotnet7 (and runtimes coming out of it) yet. 2022-11-25 18:53:20 I don't quite understand the reason why installing nginx would pull aspnetcore though, as their shouldn't be a dependency. Are you installing jellyfin, for example? 2022-11-25 18:53:55 But anyways, installing both versions at once is supported. dotnet projects that require version 6 will use that one even if they are both installed. 2022-11-25 19:12:39 oh right hcs is gone 2022-11-25 20:18:38 so, i updated testing/galera (latest version, now builds on all archs), and then moved it from testing --> community... the APKs have disappeared from /alpine/edge/testing/... how long before they show up in /alpine/edge/community? 2022-11-25 20:21:08 tomalok: I'd expect that to happen at the same time 2022-11-25 20:24:01 tomalok: you left out the arch field 2022-11-25 20:24:09 Which means the package is disabled 2022-11-25 20:25:21 You want `arch="all"` 2022-11-25 20:25:59 🤦 ikke: thanks 2022-11-25 20:33:14 was amazed how quickly the merge pipeline got through the builds... 2022-11-25 20:36:13 Yeah, it can do nothing quite fast :p 2022-11-25 20:43:07 looks like there's still work to be done on archs other than x86, x86_64, and s390x (at least this one's newly working) -- wordsize issues. 2022-11-27 00:14:28 Is known issue that Alpine 3.17 sometimes "freeze" (at least in gitlab runner)? I had no issue with 3.16 2022-11-27 00:15:13 it's not happening very often, but usually while building 2022-11-27 02:04:42 Thanks to whoever ported getent(1) from NetBSD and to whoever patched it to add getent ahosts. Has anyone ever tried submitting this patch back upstream to NetBSD? https://git.alpinelinux.org/aports/commit/main/musl/getent.c?id=89a718d88ec7466e721f3bbe9ede5ffe58061d78 2022-11-27 07:06:30 DavidHeidelberg[m]: haven't experienced / heard about it myself 2022-11-27 07:58:42 I'm going to upgrade our gitlab instance in a couple of minutes 2022-11-27 10:56:59 ayakael: dotnet7-build failed on armv7: error NU1101: Unable to find package Microsoft.DotNet.ILCompiler. No packages exist with this id in source(s): prebuilt, previously-source-built, reference-packages, source-built 2022-11-27 15:46:36 does anyone happen to know why community/fennel has lua 5.2, 5.3, and 5.4 support but not lua 5.1? 2022-11-27 15:50:29 The APKBUILD only specifies those versions: `_lua_versions="5.2 5.3 5.4"`, but not sure if there is a specific reason why they never enabled 5.1 2022-11-27 15:51:16 2b24d3b0b8c760932449a333fd31139ffdd4df27 changed it from 5.1 to 5.3 2022-11-27 15:51:22 right, my question was why not also 5.1, since 5.1 is the only version that luajit works with 2022-11-27 15:51:55 You'd have to ask Will Sinatra 2022-11-27 15:52:18 But they're not here 2022-11-27 15:52:21 I was a bit hoping they would be, rats 2022-11-27 15:52:23 I'll mail them 2022-11-27 16:03:09 alternatively you could create an MR to enable it 2022-11-27 16:04:56 ya, I may 2022-11-27 16:05:08 although I think a bit more work would need to be done for it to actually work with luajit and I 2022-11-27 16:05:13 er, and I am unsure what that work is 2022-11-27 17:52:41 I successfully got alpine running on my Radxa Rock 5b 2022-11-27 18:00:23 rather than commit 301074dd, should chromium fix have been to switch chromium to its own internal version of libpng? 2022-11-27 18:50:15 ikke: Thanks for the heads up, fix is being implemented here https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/41777 2022-11-27 18:51:03 👍 2022-11-27 18:51:10 ikke: if I keep the pkgrel the same will this break builds on other arches? 2022-11-27 18:51:32 No, it will just no rebuild it there 2022-11-27 18:51:37 just not* 2022-11-27 18:51:52 If you think this does not affect the other arches, then it should be fine 2022-11-27 18:53:57 Awesome. Apparently dotnet7 on armv7 does not support ahead-of-time and build looks for ahead-of-time tools. Fix just tells it to not need Aot on arm, and fix expresses itself only on that platform. I'm letting it build in pipeline, and then I'll come back to you. 2022-11-27 18:54:19 aack 2022-11-27 18:54:25 s/a// 2022-11-27 18:54:25 ikke meant to say: ack 2022-11-28 03:26:32 ikke: Okay fix works, Ready to merge. 2022-11-28 06:39:59 Merged 2022-11-28 07:50:11 morning 2022-11-28 07:50:57 post-deinstall of inetutils-syslogd does potentially harmful things. enables busybox syslogd openrc service 2022-11-28 07:52:30 Oof 2022-11-28 07:55:34 im doing a grep. seems like there is another 2022-11-28 07:56:16 sxmo-xdm-config.pre-deinstall does: rc-update del xdm 2022-11-28 12:57:28 is busybox syslogd not ok ? starting crond also starts it 2022-11-28 13:01:24 Install scripts should touch services 2022-11-28 13:41:42 Cogitri: shortwave 3.1.0 is out 2022-11-28 13:44:01 Piraty: cogitri is not really active on Alpine Linux anymore 2022-11-28 13:44:10 oh 2022-11-28 13:44:24 sorry for that 2022-11-28 13:55:05 Busy with life 2022-11-28 14:54:26 No worries :) 2022-11-28 14:54:59 Yeah, who would’ve guessed that one is busy when working part time and doing the masters at the same times :D 2022-11-28 14:55:46 I certainly didnt :p 2022-11-28 15:11:06 Cogitri You and I, we get each other haha 2022-11-28 15:21:28 can this be enabled if possible, https://www.sqlite.org/csv.html 2022-11-28 16:08:57 Any clue what this could be? https://gitlab.alpinelinux.org/Nulo/aports/-/jobs/913943#L1563 2022-11-28 17:03:45 Nulo: Memory serves, I had a similar error with dotnet due to new version of clang. Building with: -Wno-sign-conversion fixed it for me, but I'd share this with upstream as eventually clang / gcc will tighten what's allowed. 2022-11-28 17:06:46 gint64 =? int64 =? uint64 =? guint64 2022-11-28 17:07:44 They use both Qt and GLib ? 2022-11-28 18:02:11 My Python is not being upgraded to python3 but some packages depending on it have updated to 3.11 and are thus unusable now. What is the command again to check what is still depending Python 3.10? 2022-11-28 18:14:41 PureTryOut: try: apk add -s python3=3.11.0-r2 2022-11-28 18:59:51 sadly that doesn't tell me much 2022-11-28 19:00:23 ACTION sent a code block: https://matrix.org/_matrix/media/v3/download/matrix.org/bAHuEABjBqEikSIFOzSwnBfK 2022-11-28 19:09:56 ayakael: why does this happen only in armv7? 2022-11-28 19:13:22 Nulo: Can't say for sure. More experienced eyes might glean more, but I know for dotnet it only occured on x86 and ppc64le as those platforms had special code for them, and some of that code had the problematic sign conversions. 2022-11-29 00:53:02 vkrishn: do other distros compile it 2022-11-29 10:18:59 Hello71: https://packages.debian.org/bullseye/libsqlite3-mod-csvtable 2022-11-29 10:21:24 and others too, https://packages.debian.org/search?suite=stable§ion=all&arch=any&searchon=names&keywords=libsqlite3 2022-11-29 10:23:07 spatialite could be worth looking 2022-11-29 10:24:46 and tcl, if jimtcl-sqlite3 is different pkg 2022-11-29 11:58:31 someone happy to take a look at a new package of mine?: https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/41133 2022-11-29 13:30:39 I think have found a SEVERE bug in Alpine 3.17 w QEMU: win10 VMs DO NOT start, they hang and lockup on the "windows logo" , and no "rolling balls" :) appear 2022-11-29 13:31:04 I have two exact same machines: one with Alpine 3.16 boots, same file on Alpine 3.17 hangs 2022-11-29 13:31:48 ancient windows XP boots, centos boots win10 HANGS (all on qemu) 2022-11-29 13:32:02 all win10 are on on CSM/legacy BIOS boot 2022-11-29 13:33:39 same machines, same options same everything, Alpine 3.16 qemu win10 boots regularly, 3.17 win10 HANGS 2022-11-29 13:34:20 128GB ram machines, Ancient 10-core Ivy Bridge XEON processor 2022-11-29 13:35:34 qemu options same from same script, kernels opition standard/default 2022-11-29 13:36:27 Qemu was upgraded from 7.0.0 to 7.1.0 2022-11-29 13:36:57 @ikke I see 2022-11-29 13:40:39 Not sure if that's the cause, but it's one option 2022-11-29 13:42:22 it's puzzling because it's a total lockup, no logs nothing, just everything freezes, cpu stays at 0% 2022-11-29 13:43:19 exported the exact same qcow2 file via nfs, with exact same script to launch qemu on 3.16: boots OK 2022-11-29 13:50:05 wky72: sounds like a Qemu issue, I see 1 issue on Qemu gitlab (#1115) that sounds like same/similar problem (though with Qemu 7.0.0) - they say switching to UEFI and GPT resolved it 2022-11-29 13:52:17 wyk72: do you have secure boot enabled? 2022-11-29 13:52:26 can someone make sense of the fact that the module alias for scsi devices in linux-lts x86_64 has the form `alias scsi:t_` while the module alias for linux-virt x86_64 is `alias scsi:t-` (`-` vs `_`)? Check with grep -E 'scsi:t[-_]' /lib/modules/**/modules.alias 2022-11-29 13:52:47 i know secureboot is broken in current efi firmware from qemu 2022-11-29 13:53:48 nmeum: i noticed that years ago. Seems like either - or _ can be used in module names 2022-11-29 13:53:49 Does anyone know how to load set ENV=${HOME}/.ashrc in a gnome-wayland session without hacks? 2022-11-29 13:54:01 ncopa: he said legacy/CSM boot, so no UEFI 2022-11-29 13:54:04 as per https://github.com/torvalds/linux/blob/master/include/scsi/scsi_device.h#L615-L617 the correct module alias would be `alias scsi:t-` and I just don't understand why it is `alias scsi:t_` instead on linux-lts x86_64 2022-11-29 13:54:38 ncopa: really? by why is it inconsistent across different kernel flavours? 2022-11-29 13:55:23 nmeum: in general i've noticed a "-" / "_" mismatch with some modules filenames vs the name that modinfo shows 2022-11-29 13:56:00 I just ran into this because this breaks automatic module loading in booster 2022-11-29 13:56:48 minimal that is is what i have observed too, years ago. I have never seen it documented anywhere as far i can remember 2022-11-29 13:56:55 an "alias" is an alias though, not the definitive name of something 2022-11-29 13:57:02 $ modinfo af-alg 2022-11-29 13:57:02 filename: /lib/modules/5.15.79-0-virt/kernel/crypto/af_alg.ko.gz 2022-11-29 13:57:34 both modinfo af_alg and af-alg works 2022-11-29 13:57:39 same with modprobe 2022-11-29 13:57:48 it is weird because even on linux-lts /sys/devices/**/modalias reports scsi:t-0x00 while it is named scsi:t_0x00 in /lib/modules/.../modules.alias so there is a mismatch 2022-11-29 13:58:04 ncopa: ah, interesting so they have special handling for _/- in modinfo 2022-11-29 13:58:20 I will look at the modinfo code, maybe I will find more information there 2022-11-29 13:58:24 thanks for pointing this out 2022-11-29 13:59:10 "Hyphens (-) and underscores (_) are not differentiated in module names." 2022-11-29 13:59:29 @ncopa no it's a BIOS/LEGACY win10 machine, 440fx ancient emulated chipset (exported from libvirt) 2022-11-29 13:59:54 ncopa: do you have link to the document where you found that quote? :) 2022-11-29 14:00:13 @minimal thanks I'm looking into it, it's a kind of huge task to switch a win10 machine from MBR to UEFI 2022-11-29 14:00:24 nmeum: google gave me this: https://renenyffenegger.ch/notes/Linux/kernel/module/index 2022-11-29 14:00:40 i also found this: https://lkml.iu.edu/hypermail/linux/kernel/0412.3/0476.html 2022-11-29 14:01:04 ty 2022-11-29 14:01:16 ah, it is actually documented in man modprobe. https://man7.org/linux/man-pages/man5/modprobe.d.5.html 2022-11-29 14:01:52 ohhhh 2022-11-29 14:02:10 ok, so this is a bug in booster then sweet 2022-11-29 14:03:26 wyk72: maybe try in #qemu they have been pretty helpful, at least with me 2022-11-29 14:08:36 IDENTIFY 2022-11-29 14:08:47 IDENTIFY kal 2022-11-29 14:10:27 ncopa: when you say secureboot is currently broken in EFI firmware are you referring to Alpine not packaging the separate OVMF file for secureboot (OVMF_CODE_4M.ms.fd)? or to an upstream EDK2 issue? 2022-11-29 14:19:46 Does anyone know how to load set ENV=${HOME}/.ashrc in a gnome-wayland session without hacks? 2022-11-29 14:22:49 I've tried booting the win10 imge augmented with an ancient GRUB4DOS instead of the standard win10 bootloader - it boots! 2022-11-29 14:23:08 plain horrible but it's a workaround 2022-11-29 14:42:51 minimal: https://github.com/stefanberger/swtpm/issues/669#issuecomment-1278890027 2022-11-29 14:43:28 its the edk2-x86_64-secure-code.fd shipped with qemu that is broken due to https://github.com/tianocore/edk2/commit/4de8d61bcec02a13ceed84f92b0cf3ea58adf9c5 2022-11-29 14:44:24 this is needed to be applied in upstream qemu: https://lists.nongnu.org/archive/html/qemu-devel/2022-06/msg01441.html 2022-11-29 14:44:38 but i dont think they care, because most distros shit their own ovmf build 2022-11-29 14:44:48 s/shit/ship/ 2022-11-29 14:44:48 ncopa meant to say: but i dont think they care, because most distros ship their own ovmf build 2022-11-29 14:44:54 :D 2022-11-29 14:45:13 ncopa: our own ovmf package doesn't ship a separate secure boot OVMF file 2022-11-29 14:45:32 no but our ovmf works: cd159705145e79ed990c7a365cc9d1135b1afbbc 2022-11-29 14:45:50 also it would be good to package up some of the EFI "basic" apps like a Shell and the app for enrolling secure keys (MMX64.EFI?) 2022-11-29 14:46:22 we should probably build our ovms similar to how qemu build it and make qemu depend on it 2022-11-29 14:46:38 so we dont need to use the (broken) prebuilt qemu ovmf 2022-11-29 14:46:51 ncopa: ah, so we're shipping a single OVMF file with it builtin? ok, I didn't realise, many distros ship multiple OVMF files 2022-11-29 14:47:17 yeah, we probably should do that 2022-11-29 14:48:04 also from memory we're shipping the QEMU provided files for BIOS and UEFI plus also have edk2 package (and seabios?) providing different versions of the same files 2022-11-29 14:48:46 and iPXE also from memory 2022-11-29 14:51:35 @minimal: I can confirm that: in /usr/share/qemu I can a complete "qemu" set of bios, but if I install the package "seabios" i find a different version in /usr/share/seabios 2022-11-29 14:51:54 @minimal: I can confirm that: in /usr/share/qemu I can FIND a complete "qemu" set of bios, but if I install the package "seabios" i find a different version in /usr/share/seabios 2022-11-29 14:55:55 wyk72: wondering which one Qemu uses by default 2022-11-29 14:57:07 ncopa: with secureboot on Qemu is the enrollment "app" builtin? 2022-11-29 15:23:49 ikke: How does alpine generate https://alpinelinux.org/releases.json? 2022-11-29 15:26:55 EOL date for community repo is "2022-05-01", which I assume should be "2023-05-01" instead 2022-11-29 15:28:35 P 2022-11-29 15:29:07 PabloCorreaGomez[m]: it's generated in alpine-mksite from a yaml file 2022-11-29 15:34:56 ikke: thanks, I just sent a MR https://gitlab.alpinelinux.org/alpine/infra/alpine-mksite/-/merge_requests/57/diffs :) 2022-11-29 15:36:45 I finally switched the WIN10 machines to EFI/TianoCore boot, and they boot normally 2022-11-29 15:36:56 I filed a bug to qemu gitlab 2022-11-29 15:45:15 PabloCorreaGomez[m]: thanks! 2022-11-29 15:46:12 wyk72: so that fits with the similar issue reported on QEMU gitlab 2022-11-29 15:50:35 #minimall it's same issue as you reported #1115 ( https://gitlab.com/qemu-project/qemu/-/issues/1115 ) 2022-11-29 15:50:44 @minimall it's same issue as you reported #1115 ( https://gitlab.com/qemu-project/qemu/-/issues/1115 ) 2022-11-29 15:54:39 ncopa: cool! now the appstream generator for v3.17 will work again :) 2022-11-29 21:24:54 I see that for community packages are only supported until next stable release. Does that mean I should not update secfixes for my packages in 3.16? 2022-11-29 21:26:40 It's a bit at the discretion of the maintainer, but we also want to avoid raising the expectation that the packages will be maintained longer 2022-11-29 21:30:24 Copy that. An internal dillemma if I should keep dotnet6 up-to-date on 3.16. 6 months support sounds like too little, but 24 months sounds like too much. 2022-11-29 21:31:48 You could decide based on the severity of the issues 2022-11-29 21:32:21 Some projects decide to backport further when sever security issues are discovered 2022-11-29 21:32:40 Quite right, not every minor release has secfixes. Good call. 2022-11-29 21:34:18 What's current practice for other compilers in community? Do y'all only backport secfixes or try to keep up with minor releases when release is under 'sec fixes' support? 2022-11-29 21:34:55 I haven't seen it happen often 2022-11-29 21:35:08 either case 2022-11-29 21:36:03 But I have personally backported a minor release for an alpine release that was just out of support 2022-11-29 21:37:20 Kekay thanks. Current practices seems to backport on a need-to basis. dotnet6 is EOL in may 2024, so it feels right to backport a minor release when they are sec fixes until that date. 2022-11-29 21:38:08 Falls where alpine 3.16 is EOL anyways so good timing 2022-11-30 09:58:31 Ok, after disabling the Mold the flaking didn't disappear, so it's Alpine 3.17 issue :( I'll fill the bug and drop docker link there