IRC channel logs

2026-10-21.log

back to list of logs

<Heterochromium>Is running `guix system reconfigure <current config>` sufficient to protect me from those compromised certs?
<flurando>hi, is there any way to --allow-downgrade for guix deploy? I some how messed things up and now guix kept saying my channel is not a descendant of commit blablabla. I need to enforce a
<flurando>"downgrade" for it.
<lechner>flurando: I have (allow-downgrades? #t) in my machine.scm.
<flurando>lether: thanks, I was only looking at command-line options, thus missing this.
<stephen>you could just have a newer root pull than user pull or neweer user than root. That will happen if you sudo guix pull vs sudo -i guix pull when you're used to using the one but not the other.
<Noisytoot>why is it taking so long to merge https://codeberg.org/guix/guix/pulls/11957?
<cnx>as mentioned in the linked issue, it was very late on hako's side of the world
<cnx>i'm gonna push the nss-* updates now, though do read through https://codeberg.org/guix/guix/issues/11954 for their scope of effect
<tusharhero-xmpp>untrusem: I do use Emacs on kde plasma wayland
<tusharhero-xmpp>Emacs GTK currently
<tusharhero-xmpp>Not pgtk
<tusharhero-xmpp>cnx: thanks!
<mra>hako: quick update on the ZFS install. I thought I'd try installing on my PC, but I can't boot into the installer on physical hardware. I get an error: "init[1]: segfault at 3fff00"
<mra>I meam I'm no expert but that seems bad to me
<cnx>captain tis thursday
<olivial>hmm... the 'guix import crate' tool only works on things from crates.io? there's no way to apply it to a cargo manifest from a git repo?
<efraim>olivial: there's also the --lockfile flag to import from an existing (or just created) Cargo.lock file
<olivial>that still needs the package to be on crates.io in order to get the metadata though, right?
<olivial>ACTION wonders how hard it would be to teach the importer about Cargo.toml
<jlicht>olivial: afaik you can "just" use it with local (e.g. your own) stuff
<olivial>hmm, wdym?
<jlicht>olivial: https://paste.jlicht.nl/local-cargo.html this is a snippet I have that "just works" for a local cargo project
<stephen>Offloading builds is awesome. But does that use the local channels and just move compilation to that system?
<kestrelwx>o/
<dariqq>any java exports know if there is an alternatice to keytool for producing a cert bundle in the java format? I know trust from p11-kit can but i first need to add the certs to it which fails due to readonly store
<untrusem>dariqq: tarcv comes in mind
<untrusem>also made a pr for php update, someone with a faster machine plz build it
<untrusem>t480 is not exactlp powerful
<jlicht>untrusem: which pr#?
<untrusem> https://codeberg.org/guix/guix/pulls/11972
<untrusem>jlicht:
<aldum>dariqq: I used KSE back in the day
<aldum> https://keystore-explorer.org/downloads.html
<dariqq>my issue is that i want it not to be in java because I want to avoid depending on an arbitry jdk just for certs
<aldum>ok, do you need a java keystore or not?
<dariqq>yes, I want to remove the java keystore from the guix jdks into a separate package to be able to update the certs without rebuilding java. So what java do I use to just prepare the certs?
<dlowe>ACTION hitting refresh to see when the php version goes up.
<untrusem>I have yet to disable one tent
<untrusem>test*
<untrusem>a pr is there
<untrusem>group are your here?
<untrusem>btw what's the merge policy for this, we wait for a week or two but in security incidents I guess we can merge asap
<lechner>It's ASAP if the fix is clear, but with a consideration for rebuilds or grafts. Probably best to have the security team sign off.
<user176462>Hello! I am looking to define a package for which I have written a pkcs11 module file. In my package definition this file is installed in /share/p11-kit/modules directory, as this seems to be the place where other packages also install their module files (gnome-keyring, p11-kit-trust). However, `p11-kit list-modules` only lists p11-kit-trust as a module and neither gnome-keyring nor my package.
<user176462>Does anyone have an idea as to why p11-kit does not load this module?
<user176462>It seems that in p11-kit definition, the variable P11_MODULE_PATH is defined pointing to share/p11-kit/modules. However, it does not seem to have any effect in this case?
<dariqq>user176462: You need p11-kit and your package in the same profile
<user176462>Ok I will try adding it to my system profile if it works. Thank you
<user176462>Hmm. Even with p11-kit in the same profile it does not seem to find the modules
<dariqq>can you try guix shell p11-kit your-package and see if the environment variable gets set?
<user176462>The environment variable is set both in a shell and in my system. But p11-kit list-modules does still not find the module not even in a guix shell
<nckx>stephen: Yes.
<dariqq>i guess then something in P11-kit does not work as expected
<untrusem>hi hako, could you if disabling those tests work on zfs?
<csantosb>untrusem: See recent "nss-certs: Graft".
<untrusem>pr number?
<csantosb>guix/pulls!11957
<csantosb>guix/guix!11957
<csantosb>In commit 7608a9954a7ba0a9ce756bf3135dc0ba51a40af9
<untrusem>ok
<untrusem>csantosb: what should i see in that commit?
<bavier>would love to figure our why my package updates routinely need to substitute 868MiB of qtdeclarative-debug every time.
<ieure>bavier, Grafts.
<Rutherther>and gc will remove it, making you redownload it again
<Rutherther>because nothing references it
<csantosb>untrusem: with respect to the merge policy with security related changes
<csantosb>in case it triggers a world rebuild
<untrusem>?? was this the answer to my, "btw what's the merge policy for this, we wait for a week or two but in security incidents I guess we can merge asap" question
<untrusem>because I don't have anything to do with that commit :??
<untrusem>csantosb:
<bavier>ieure: grafts doesn't make sense to me. the debug output would need to be in my manifest graph somewhere, right?
<Rutherther>grafts take effect on whole package, not per output, so they need all outputs of the package
<ieure>That.
<bavier>omg
<ieure>I agree that this is not ideal and could perhaps be improved.
<untrusem>hi ieure o/
<ieure>Hi untrusem.
<Rutherther>it definitely should be improved, I think there were some PRs, but they broke some stuff
<untrusem>ieure: long time no see
<ieure>untrusem, Yep. Saw the PR banning LLM changes and have been thinking about contributing again, but October is an extremely busy month for me.
<ieure>My oldest kid's birthday is at the start of the month, mine's at the end, and I spend a week volunteering for a retro gaming expo in between. Crammed a weekend trip with some friends in also.
<untrusem>:)
<untrusem>I have been made a committer this week
<ieure>Congratulations!
<untrusem>if you didn't saw the mail
<ieure>You earned that for sure.
<ieure>I didn't, I unsubscribed from the lists.
<j_importer>i really should start contributing, but i keep getting sidetracked and haven't spent much time poking around with guix :/
<untrusem>ieure: hehe thank you
<untrusem> https://lists.gnu.org/archive/html/guix-devel/2026-10/msg00116.html
<untrusem>j_importer: please do
<j_importer>i made the foolish mistake of saying "i'll try learning guix *and* a new WM at the same time"
<j_importer>always change one variable at a time!
<bavier>untrusem: congrats on committer!
<untrusem>bavier: thanks, its quite a responsibility
<dlowe>it doesn't look like it's happened yet?
<dlowe>oh, made a committer, I get it. Grats
<Noisytoot>is graphviz/dot broken for anyone else?
<Noisytoot>I'm trying to generate a dependency graph for gajim with "guix graph gajim | dot -Tsvg > gajim.svg" and dot is just hanging forever
<Noisytoot>strace shows that it's frozen on a brk syscall
<ieure>Noisytoot, It's because the graph is extremely large.
<ieure>Noisytoot, Try running `guix graph gajim > gajim.dot && ls -lh gajim.dot', it's probably hundreds of megs.
<ieure>IMO the `guix graph' feature is not very useful.
<Noisytoot>ieure: 636K only
<ieure>That is still large for a dot graph.
<ryanprior>large graphs are brutally challenging to represent in humane ways
<ryanprior>the best answer I know is to put a strict cap on how large of a graph you will show, which means grouping nodes based on rules and heuristics, and presumably providing some sort of interactive or iterative explanation of those groups