IRC channel logs
2025-01-15.log
back to list of logs
<lfam>Yes, sounds nondeterministic based on upstream. Good thing we will disable these tests shortly <graywolf>lechner: Hi! Quick question about #71613. Did the fix actually work? I does not work for me, and blaming the smtpd.h I do not really see any changes. <graywolf>So I am not sure why it would stop working... <graywolf>lechner: I took the liberty to produce #75571 which seems to fix it on my system, feel free to review (since you fixed it originally). <unmush>hash mismatch with libgcrypt-1.10.1.tar.bz2 <unmush>it gave me 0mjmcijkjrhczgwinwfmxls0hx9xb7ah5lip1fk4wqzv3f00bbpn <vhns>Is the Guix Build Coordinator what I'm looking for if most of my devices are very slow and I want builds to happen on a faster server I have? <unmush>hmm, guix download of that now gives me the correct hash also <mange>Maybe. Also could be a spurious error page of some sort? <unmush>I wonder if there should be an option to try various fallbacks in case of hash mismatch e.g. other mirrors, software heritage, etc <unmush>currently it only does that in case of getting an error response or no response at all I think <mange>vhns: It depends what you want. You may want offloading (see "(guix) Daemon Offload Setup" in the manual), or you may want something like Cuirass and/or the Build Coordinator. You haven't said much, but I suspect you want offloading. <vhns>mange: offloading was the right one, thanks! <unmush>... now I've got a hash mismatch for libgpg-error-1.45.tar.bz2 <mange>I got the same, but the contents are HTML. <mange>If I open that URL in my browser it says "Error :( We can't seem to find the page you're looking for." <unmush>I suppose part of the reason to not automatically try alternate sources on hash mismatch is that it would require hashing the file twice for each successful download, once by the builder and once by the daemon <efraim>unmush: crysys.hu and some other defunct mirrors were removed at the end of october, not sure why it's still trying to download from them for you <Rutherther>arjan: that seems unnecessarily complicated, it suffices to use the startx command service in your system, it gives you startx command that is wrapped with all the modules needed. In case you are not on guix system, there is the procedure it uses xorg-start-command-xinit that gives you that command <arjan>I recall that only works when xorg-server and modules are installed in the system profile, but I am using guix home <khimaros>is there a shorthand for installing pip packages (or executing arbitrary shell commands) in a guix shell (using eg. manifest.scm)? i'm looking at using guix shell to replace my use of `mise` <Rutherther>arjan: no, you don't have to have any modules installed in any profile at all, definitely not in the system profile <khimaros>it looks like i can define a pip package in my manifest.scm, but i'm curious if there's an easier way. i'm okay if the packages get installed into .venv instead of as actual guix package. <Rutherther>khimaros: no, guix shell gives you packages managed by guix. It doesn't give you pip packages, you need to get those yourself <khimaros>my thinking is... this is a potential path toward gradually replacing the bespoke shell commands with guix packages, but i want an easy on-ramp for existing tooling even if it's not 100% reproducible at first. <arjan>hm at least I had these problems some time ago because of a security feature in xorg-server, might want to retest then <jakef>hi efraim, how goes the rust packages? <efraim>not bad. I have sequoia-chameleon-gnupg locally waiting to be pushed, and otherwise taking a bit of a break until after FOSDEM <jakef>very good. are you presenting there? <efraim>I have a talk about using gawk and extending it, and I'm sure I'll be very busy during guixdays <efraim>I'm keeping myself distracted with my poor C skills for now while working on the gawk presentation, since the upstream examples of extending gawk use C <arjan>efraim: good luck with preparing your talk! <khimaros>Rutherther: so there is no way to invoke shell commands in the shell upon startup or during build? <Rutherther>khimaros: when you are invoking commands during build, you are packaging something. During startup, sure, you can just append "--" and after that what you want to execute in the shell. I didn't think you were asking for that, since I would classify this as normal usage of a shell <dariqq>it only took 8 attempts but i was able to overcome the flaky tests of python-pydevd <khimaros>Rutherther: and at build time? it seemed like this might be a supported thing: " In some unusual cases though, you might want to create manifest entries for things that are not packages" <Rutherther>khimaros: I don't know what you mean by that. The type may not be a "package", I would still call it process of packaging, albeit it can be a gexp etc. <Rutherther>khimaros: the process of building with guix comes with all the advantages and disadvantages. Most prominently for this, you can't use the internet during builds, so you cannot just execute pip and call it a day. You would need to make it a fixed output derivation, and at that point, if you want it to work reliably, will have to make sure pip produces completely reproducible environments, and will probably spend much more time on this than if you just... <Franciman>hi, is there a list of wifi cards that are compatible with linux libre anywhere? <Franciman>divya: uhm i feel a bit awkward to ask this. I checked that website, and it shows some intel cards from 2024 as working with free software <Franciman>wtf? There is Wi-Fi 6 cards supported by linux libre? <divya>Franciman: Indeed the driver and module of some new intel chips is libre. It doesn't have any blobs. <divya>Franciman: And yeaah there's now a bunch of cards supporting linux-libre <divya>Unfortunately they're all too expensive for where I live <efraim>divya: About once a year I go through and try to purge some of the unneeded rust packages, but it's not an exact science <divya>efraim: Got it, you're working on fixing them? <efraim>I'm not sure that anything is broken, bottom builds just fine <futurile>Q: is there a specific email alias to notify adding to the Web site? (I'm sure Simon told me one in the past but I can't find it) <sleepy68>hey, new user installing guix from the binary tarball. I seem to be missing the guix-gc.timer service. Anyone hit this issue before? <Rutherther>sleepy68: what makes you think you need that service? <sleepy68>well just that its in the install docs and i assumed there was a reason they gave that option <sleepy68>I have guix-gc.service just not the timer <Rutherther>sleepy68: you can of course create it,if you want to periodically run guix gc, but it's not required, you can just gc manually whenever you want to remove older stuff you don't use anymore <sleepy68>Rutherther thanks for the clarification! I had thought as much. Should the docs maybe be amended? just seems a tad misleading to provide a command that only half works. The comments at the top of guix-gc.service also mention this file so maybe that too. I'd be happy to submit a patch or whatever if you could point me in the right direction (unless <Rutherther>sleepy68: I haven't used this installer, so I cannot say whether this is a bug or an intention <sleepy68>ah okay well thanks anyway, ill check some older commits and see whats what <divya>How can I work with a library that I installed using guix shell? For example to make it available to gcc? <Rutherther>divya: put gcc-toolchain to that shell as well. Then you will have it in LIBRARY_PATH and INCLUDE_PATH. That's all to provide it somehow. Of course if the library uses non-standard include location, you will need to use other tool to get it, ie. pkg-config, again, by adding it to the same shell, then you will get proper PKG_CONFIG_PATH. <unmush>efraim: because I haven't rebased my local repository in quite some time <lilyp>is that me or was rust-team merged? <lfam>lilyp: "Merge remote-tracking branch 'origin/rust-team'" on January 9 <lfam>I wonder if the merges of team branches should be mentioned in etc/news.scm <lilyp>maybe not, I was just looking at librsvg and seeing a vaguely familiar version <lilyp>my previously last pull was three days ago, so it wasn't merged "in the mean time" <Rutherther>lfam: that's interesting that there is really a merge commit, for example if I remember right, the core updates merge few months ago happened by directly appending the commits instead of a merge commit <lfam>Yes, there are two ways to do it (rebase and merge) and it's a matter of taste <lfam>Of course there is a correct choice, as with all questions of taste ;) <Rutherther>I personally prefer a merge commit, since it has two parents, you can track where the previous master pointed to <lfam>And I prefer a rebase. But ultimately it's up to whoever is in front of the keyboard <Rutherther>I think it costs people time, for example someone had an issue after core updates merge, and wanted to do back. They tried a few commits and they had no substitutes available, if there was a merge commit they could just go back easily <lilyp>FWIW, a rebase is cleaner, but requires that only one person signs off commits on any given branch (or else you're signing commits that someone else signed before and have to update the trailer… not ideal) <lfam>It doesn't matter who signs the commits as long as they are authorized. That's the security model we have <dariqq>I just sent the 7th rebase of #71968 and I am increasingly losing motivaton to keep rebasing it . If someone wants to take a look at it, Id greatly appreciate it <maxz>What's the canonical way to discard the output of a Shepherd service? (e. g. not having it go to the Shepherd log or to a log of its own) And would providing /dev/null as log-file work or cause problems with log-rotation? <Rutherther>PuercoPop: on the link you sent I see exact syntax defined, showing select, prefix and renamer, so I think that there isn't opposite of a select <lilyp>R6RS libraries have "except" <lilyp>resolve-interface (and likewise define-module) has #:hide <PuercoPop>lilyp: Thanks! #:hide is what I was looking for <podiki>weird, i used the'ensure-no-mtimes-pre-1980 phase from python-build-system and yet errors on pre-1980 timestamps :( <podiki>maybe build is trying to do something from a store file...? <lfam>podiki: Does it say which file it is? <lfam>There could also be a nested zip file? <lfam>(Check the code comment on the ensure-no-mtimes... procedure) <podiki>trying to disentangle the error, python-hatchling and other build system files... <lfam>Is it trying to zip this file up? <lfam>Or, is it complaining about the current date, not a file mtime? <lfam>I see other packages with that error in a code comment, and they change the SOURCE_DATE_EPOCH so that python can build wheels <lfam>I guess it's a rare package that builds a wheel in a build phase? <podiki>don't even need the mtime change phase <lfam>Maybe add a code comment in your package for the next person who sees this problem <podiki>this is for something that probably can't go in guix, though i'm actually not sure (is a script/launcher to use valve's steam runtime/proton tools) <podiki>maybe it can, though maybe proton itself has some binary bits from windows stuff? or it can (like you can with winetricks) <lfam>I wouldn't make the effort if I had written the package. Can't imagine the arguments <podiki>i got stuck with some difficult debugging (containers in containers, something works in a shell in the container but not launched from a program in the container) so i package up the tools it uses instead, maybe make a better guix version <freakingpenguin>With the gnupg 2.2->2.4 update gpg-agent can no longer connect to my smartcard. Has anyone already encountered this and found a solution? <freakingpenguin>I see some changes on the Arch Wiki since I last set this up, but no luck so far. <podiki>i use a smartcard but i don't think have updated since that change <podiki>gpg --card-status shows nothing i guess? <freakingpenguin>From the gpg --card-status output and AW changes I think it's because gpg is insisting on using scdaemon (their own smartcard daemon) and not pcsc, which I use via pcscd-service-type. <podiki>hm. i also use pcscd so this might be a problem for me too <freakingpenguin>disable-ccid in scdaemon.conf used to be enough to fall back to pcsc but I might need something more now. <freakingpenguin>Apparently the fallback mechanism was removed due to their improved support for multiple readers. <podiki>i don't think i did anything to scdaemon.conf (and i do see scdaemon running) <civodul>futurile: hey! the blog post looks great to me! i’m ready to push but do you prefer to have it pushed tomorrow at a time when people are more likely to be around? <freakingpenguin>podiki: Apparently I'm just special. This entire time scdaemon.conf had an invalid line after disable-ccid, so I was still using the old fallback mechanism and not disabling scdaemon's default behavior. <podiki>and now all good with that invalid line removed? <freakingpenguin>After fixing that, removing an old reader-port line and keeping disable-ccid alone is enough. <podiki>why do you disable-ccid? i saw that in the guide i followed so gpg doesn't keep asking for an inserted card, but i don't havethat issue <freakingpenguin>I think it works like this: At <=2.2, gpg will fall back to pcsc automatically if ccid-based access fails. With 2.4, the auto fallback was removed. <freakingpenguin>But >=2.4 you likely will need to add disable-ccid so gnupg will continue using pcsc. <ieure>Why pcscd over scdaemon? Better support for >1 smart card? <freakingpenguin>Oh, I think I added disable-ccid in 2.2 because I had an issue when Firefox would "take over" the smart card, locking gpg out. <podiki>the only perhaps minor issue I've had is when something makes it so that i need to unlock the card again, maybe after an oath code? but that might be a yubikey thing specifically <freakingpenguin>I notice I have to unlock my yubikey multiple times when I perform different operations, i.e. signing vs decryption. <podiki>it is either for oath codes or when a site wants security key touch, but pretty sure it is the codes. maybe something about the different slots/modes of yubikey? not had to unlock when doing other gpg stuff <xaltsc>Hey, can I install guix on ZFS via a chroot ? <Kolev>How do I get Jitsi Meet (screen sharing) working in LibreWolf? I don't want to use Firefox...