IRC channel logs

2024-11-28.log

back to list of logs

<anemofilia>x8dcc: Do you still need a startx script? A Guix user sent me one once
<anemofilia>It works just fine, and is pretty simple
<anemofilia>(I'm on wayland now, but I'm keeping it for cases like this)
<x8dcc>No, no, that's not the issue. I wrote a very simple one: https://github.com/8dcc/linux-dotfiles/blob/main/scripts/startup/startx.sh
<x8dcc>It's just that I didn't understand why the 'startx' script from the 'xinit' package didn't work. Honestly, I kind of rather keep using this version because it's simpler and I know what goes on under the hood (unless there is something wrong with it)
<x8dcc>anemofilia: Thank you for trying to help me, anyway :)
<anemofilia>x8dcc: Guix also have a startx-command-service-type too, I never tried it tho
<anemofilia>You're welcome :)
<x8dcc>anemofilia: Yes, that service is what I was initially asking about
<anemofilia>Ah sure then
<x8dcc>But the main problem with the 'startx' script from 'xinit' is that, since it's not patched in any way, it tries to look for a '/bin/X' binary, but it can't find it in the '/gnu/store/...-xinit/' path
<anemofilia>Hmmm
<x8dcc>I have to leave now
<wolfdog>seems like the egg importer for `guix import' is slightly broken, doing `guix import egg spiffy -r' outputs this error for me: https://paste.debian.net/1337194/
<wolfdog>but non-recursive works fine
<ekaitz>wolfdog: open an issue please
<ekaitz>and cc me to it
<wolfdog>alrighty
<wolfdog>ekaitz: what would be the email I have to cc it to?
<wolfdog>ah, they quit oops
<wolfdog>ok i found the email (by searching through the archives), yay
<unmush>could a moderator confirm whether my guix-patches mail is in the queue?
<unmush>it's a bit tedious to wait several hours to get the issue number so that I can actually send the patch series
<podiki>woah, was meson updated directly on master?
<intermet>Hi there! I am trying to package a game with a copy-build-system. I use the directory of the game as the source. (source (local-file ... #:recursive? #t)). The game install dir is very large 12G+. I get the error guix download: error: fport_write: Connection reset by peer when I tried to build the package. Any advice?
<divya>Hello
<ekaitz>hi! wolfdog thanks for cc-ing me
<ekaitz>#74565
<ekaitz>looks pretty absurd but in the first look idk why that is happening
<civodul>Hello Guix!
<ekaitz>civodul: hi
<Rutherther>intermet: is your /tmp tmpfs? if so, your ram is getting filled with that 12 GB
<futurile>Morning morning
<ekaitz>weird question! if I have `(define* (hola #:key a b c #:allow-other-keys #:rest rest) (display rest))`
<ekaitz>what should `(hola #:a 1)` display?
<ekaitz>(#:a 1) right?
<ekaitz>hmmm ok yeah it looks that's what it should (it's a little bit antiintuitive)
<jaadu>ekaitz: I think this is why its frowned upon to use both keyword arguments and rest. Its semantically confusing.
<ekaitz>jaadu: pretty confusing, yeah
<intermet>Rutherther: How do I check that? It would make sense.
<intermet>I changed TMPDIR to a place somewhere in /home (with enough space) but I still have the same issue.
<Rutherther>intermet: "mount" will tell you what's mounted where
<ekaitz>civodul: what's error: remote unpack failed: unable to create temporary object directory
<ekaitz>i'm getting that when I try to push :(
<ekaitz>my first commit is being problematic hehe
<Rutherther>intermet: changing TMPDIR in your process environment is not going to do anything, you would have to change guix-daemon TMPDIR
<civodul>ekaitz: oh, could it be ENOSPC on Savannah or something like that?
<civodul>i haven’t tried to push anything today
<civodul>dunno maybe try again later? :-)
<ekaitz>civodul: i'm talking with efraim about it, it looks like he can push but I can't so it might be a permission issue
<civodul>oh
<efraim>I just rebased the rust-team branch without problems
<civodul>i do see you at https://savannah.gnu.org/project/memberlist.php?group=guix tho
<efraim>I added him there and to the keyring branch
<efraim>ekaitz: is your commit signed?
<civodul>ACTION risks a prediction: 2025 will be the year when we say goodbye to Savannah
<ekaitz>i think so
<efraim>git log -1 --show-signature
<ekaitz>savannah says successfully authenticated commit blablablah
<ekaitz>and yes gpg: Good signature from ...
<z572>Maybe the remote url is wrong?
<z572>ssh://git.savannah.gnu.org/srv/git/guix.git
<efraim>you;re missing a colon
<efraim>ssh://git.savannah.gnu.org:/srv/git/guix.git
<cbaines>I'm looking at how to reconfigure hydra-guix-130, should the configuration be set out similar to hydra-guix-129?
<efraim>wait, does it work with or without the colon?
<ekaitz>i have the colon
<ekaitz>and tried without it and fails too
<ekaitz>ahah this is a classic ekaitz issue
<ekaitz>idk man.. it's letting me connect and everything
<ekaitz>in fact it runs the checks and all
<ekaitz>ACTION is sad
<ekaitz>ACTION had fixed a bug today and wanted to push
<efraim>ekaitz: perhaps try changing the ssh:// to git://
<ekaitz>i tried that before
<ekaitz>oh but that doesn't work
<ekaitz>git works with no user
<cbaines>my savannah remote is: cbaines@git.savannah.gnu.org:/srv/git/guix.git
<cbaines>removing and then re-adding ekaitz as a member on savannah might also be something to try if that hasn't been tried already
<z572>ssh z572@git.savannah.gnu.org
<z572>Hello z572! You've successfully authenticated, but interactive shell access is not allowed.
<z572>Connection to git.savannah.gnu.org closed.
<ekaitz>cbaines: mine is ekaitz@git.savannah.gnu.org:/srv/git/guix.git
<ekaitz>and if i ssh i get that message that i'm successfully authenticated
<z572>maybe try ssh ekaitz@git.savannah.gnu.org
<rekado>ekaitz: oh, funny, that.
<ekaitz>yeah z572 that's what I meant, i can do it and i have the same result you had
<ekaitz>it's like i have access to connect but not to change the repo contents
<ekaitz>it has happened to me in the past when configuring a git repository with gitolite
<rekado>my changes to the r-build-system and the importer on the r-team branch lead to a lot of extra inputs --- many of them cannot be added, though, because that would lead to dependency cycles.
<cbaines>rekado, any suggestions for how to set out the configuration for hydra-guix-130? Currently I'm doing it similar to hydra-guix-129
<rekado>looks like the importers will need to gain the ability to check for cycles
<rekado>cbaines: that would have been my approach, too.
<efraim>ekaitz: can you clone the checkout with 'git clone ekaitz@git.sv.gnu.org:/srv/git/guix.git'
<ekaitz>efraim: cloning
<efraim>ok, I'll try removing you and then adding you again to the savannah group
<ekaitz>efraim: tried to push from the new clone and i got the same error
<ekaitz>maybe it's a savannah error that they are not handing new user permissions properly?
<efraim>perhaps
<jaadu>Are guix services available on foreign distros?
<Rutherther>jaadu: the guix home ones are. Guix system ones aren't
<efraim>ekaitz: looks like this? https://lists.gnu.org/archive/html/savannah-hackers/2024-11/msg00005.html
<jaadu>Rutherther: Thank you. That was very helpful.
<ekaitz>efraim: yes
<ekaitz>efraim: i'll send them another report
<efraim>ekaitz: to savannah-hackers@gnu.org?
<ekaitz>efraim: idk, probably in the savannah interface, there's a report tab
<ekaitz>get support, better said
<efraim>the savannah-hackers mailing list is where I found that bug report and a few like it
<civodul>you could try the #savannah-hackers channel also
<efraim>that sounds like a good idea
<ekaitz>they are the same, they come from the get support from in savannah
<ekaitz>opened #111156 in savannah thanks all for the help and specially efraim for catching the issue
<ekaitz>in the meantime if anyone wants to commit this patch... https://issues.guix.gnu.org/issue/74565
<ekaitz>hehe
<gabber>reconfiguring my system fails with the message: "guix system: error: no target of type 'postgresql' for service 'cuirass'". i have a cuirass service installed, but i fail to see the connection to 'postgresql'...? does anyone here know what's up?
<cbaines>gabber, cuirass uses PostgreSQL. This is probably due to the switching between having a default postgresql package for the service to use, and not having a default
<cbaines>gabber, you'll need to add a postgresql service to your system configuration, specifying the postgresql package version that you're currently using
<civodul>gabber: yes, check out ‘guix pull --news’, which mentions the change where postgresql-service-type no longer provides a default package
<gabber>thanks, you two! i did see the news headline about postgresql, but was unaware of the implications to cuirass (which of course makes sense)
<amano>Is there a way to make initrd execute dropbear service with a certain command?
<amano>I'm trying to look for a way to unlock a LUKS volume remotely through dropbear. How can I do that on guix?
<Deltafire>hmm.. i'm also using LUKS, but i noticed that initrd is on the encrypted partition - so the volume needs to be unlocked prior to initrd/dropbear
<Deltafire>on my ubuntu machine, initrd is on the unencypted /boot volume, so dropbear runs prior to unlocking /
<Guest78>Hi! I am unable to get cloudflare-warp working. Any help would be appreciated!
<Guest78>When using cloudflare-warp-bin package from rosenthanl repo, I get the error that "ERROR warp::root_ca: This OS does not support custom root CA management" and Firewall could not be started "ERROR firewall::linux: Failed to start firewall: Os { code: 2, kind: NotFound, message: "No such file or directory" }"
<cbaines>Guest78, given that package isn't in Guix, I'd talk to whoever manages the Rosenthal channel
<Guest78>Oh, ok! Is there any other thing like it?
<cbaines>I'm unsure, I don't know what Cloudflare WARP is
<Guest78>It is basically like a free VPN
<Guest78>My uni blocks a lot of websites for no apparent reason, like codeberg. And given that they are not willing to modify it. It is my only option.
<Guest78>It did work on nix though.
<Guest78>Or any other way to bypass these restrictions?
<cbaines>there's VPN things in Guix, wireguard for example
<Guest78>Yeah, but I am a student and not really able to spend money on these thing. T-T
<amano>Deltafire: I guess you can use something like JetKVM.
<amano> https://jetkvm.com/docs
<Deltafire>that would work, also similar projects based on raspberry pis
<amano>PiKVM requires a separate power cable. JetKVM gets its power from the usb cable connected to the controlled computer.
<amano>However, PiKVM can be connected to a multiport KVM.
<csantosb>Oh, librewolf is broken ... using faf8ba9 😥
<oliverp>Right so it was very to read the new Guix blog post from you the ohter day
<oliverp>Thans very much for shering the work progress with the community.
<civodul>csantosb: broken how?
<csantosb>civodul: guix pull && guix install librewolf
<efraim>on which commit? On master I was able to build and launch it
<efraim>369d2698b0bfc3726f8e6d232d43d0dda832225f
<csantosb>faf8ba9b43323dc61988a368690b20008611fc87
<civodul>works for me at 50d9e8b454906c0b95abfddf5c27ecfacf241cbe (from Sunday)
<efraim>the next 4 commits shouldn't have made a difference
<csantosb>I restart to check the log, I had to --do-not-upgrade=librewolf to upgrade my system
<csantosb>I hope my laptop survives the experience 😯
<csantosb>By bad. 16 GB is not enough to build it. Forget about.
<civodul>csantosb: looks green here: https://ci.guix.gnu.org/build/6713971/details
<csantosb>`guix weather librewolf' gives nothing available, is it normal ?
<civodul>it says OK for ci.guix.gnu.org here
<csantosb>Funny ... https://paste.sr.ht/blob/17490db491eb2c30f2bb4faeb8900a4e87f0fc09
<efraim>oh, looks like I got it from cuirass.genenetwork.org
<unmush>after preparing a big cover letter and formatting my 21 patches, I've discovered that my mail provider for this identity seems to be silently discarding outgoing mail...
<unmush>is there any other way I could send my patch series?
<andreas-e>Maybe change identity? The subdirectory with the content to send should still be there.
<andreas-e>You could edit the [sendemail
<andreas-e>[sendemail] block in .git/config or ~/.gitconfig
<unmush>I've discovered that hashbang.sh blocked sending of outgoing email some time ago apparently (it was working when I signed up...), so here's my patch series: https://unmush.de1.hashbang.sh/misc-files/mono-series.tar.gz and the gpg signature is at https://unmush.de1.hashbang.sh/misc-files/mono-series.tar.gz.sig and my gpg key is at https://unmush.de1.hashbang.sh/pubkey.asc
<unmush>in case anybody feels like looking at it
<look>Hi, did https://issues.guix.gnu.org/49102 come back? I can't seem to be able to hack on guix-daemon because of missing libz.so.1, is it just me (and I'm doing something wrong)?
<hapst3r>hi guix o/
<hapst3r>futurile: I've got a question, do you mind if we have a private chat for a moment?
<admmq>Hi guix! Is there a command to build ALL packages avaialable in the current guix checkout? Something like "guix build *"?
<futurile>sneek: later tell hapst3r sounds great - I'll be around to chat to :-)
<sneek>Okay.
<ekaitz>efraim & civodul now i can push
<ekaitz>and i did
<wolfdog>\o/
<wolfdog>thanks for fixing it ekaitz
<ekaitz>wolfdog: this one was easy
<ekaitz>wolfdog: de nada ;)
<efraim>ekaitz: yay!
<ekaitz>:)
<simendsjo>I'm wondering if I need to do something special to get eyes on a bugfix. A package upgrade I have from september has still not merged, but that's not really a hurry. But this is an annoying bug with a wrong url causing problems building dependent packages. Will someone look at it, or do I need to forward to some maintainers?
<nutcase>simendsjo: I can have a look at it, but I am not a maintainer
<simendsjo>nutcase: Thanks, issue 74359. But is it enough to simply add a patch to an issue and someone will look at it? Or is there just not enough manpower? I have the impression the review queue is quite large.
<nutcase>simendsjo: it could be true that more manpower could help with that latency. futurile puts some efforts in promoting the patch review process: https://libreplanet.org/wiki/Group:Guix/PatchReviewSessions2024
<simendsjo>Nice!
<nutcase>and now I'm just training it myself by reviewing your patch
<ekaitz>simendsjo: i could take a look to it
<ekaitz>simendsjo: the format of the commit message doesn't follow our standards
<ekaitz>if you improve it we could do it, it's a simple change
<simendsjo>ekaitz: Thanks, issue 74359. nutcase might already be on it though. It's not urgent, I just asked here to avoid it rotting forever ;) Looked it up, and it's affecting exactly 1 package ;)
<nutcase>simendsjo: I keep my hands off, I think ekaitz is of more help for you :-)
<ekaitz>haha
<simendsjo>ekaitz: I just looked at other commit messages in the same file and used the same format.
<ekaitz>did you? :)
<nutcase>I'm building, currently and I was about to set owner + under-review (not sure, if this is really required). But don't do that now
<nutcase>ekaitz: full stop at end of commit msg is missing or what is wrong?
<simendsjo>Ah, I didn't use a capital first letter and didn't end with a period. I'll submit an update.
<ekaitz>the body of the commit
<ekaitz>it's not only the title, but also the description
<nutcase>ok
<simendsjo>Sure, I can add that too, but I really want to include the part where it's a typo made in another unrelated commit by mistake. Where should I put that in this strict format?
<nutcase>ekaitz: If you find an additional minute, you could also check the commit msg of a patch, I just submitted.
<nutcase> https://issues.guix.gnu.org/74597
<ekaitz>simendsjo: the commits message have a first line that is considered a title, it should be followed by an empty line, and then the description comes
<ekaitz>in that description you can write whatever you like and later you have to add the change list like the changelog format says
<nutcase>ekaitz: the problem with #74597 is: guix weather -l bluez tells me "Building the following 1250 packages would ensure 2424 dependent packages are rebuilt...". I think, I don't want to do that on my notebook. And I am even not sure, if this is the way to go (update just one single package that affects that many others).
<ekaitz>this is the changelog format description https://www.gnu.org/prep/standards/standards.html#Change-Logs
<ekaitz>nutcase: i don't know very well how to deal with that yet... we had core-updates branch for that, but we don't now
<nutcase>ekaitz: nevermind, don't care about that now. I've to leave now, anyways. Have a nice day or night or whatever.
<simendsjo>ekaitz: I added an update (not up yet?), but I haven't read the changelog documentation here yet. Have to go to bed, so I can look more at it tomorrow if it's still not correct. I don't have an IRC bot, so I'm logging off and missing all discussions going forward :)
<ekaitz>nutcase: you are missing the title line in the commit btw
<ekaitz>it should be something like 'gnu: Update bluez to ...'
<nutcase>ekaitz: isn't that the subject?
<ekaitz>simendsjo: yeah, don't worry, i'm also in Europe, ping me tomorrow
<ekaitz>nutcase: oh true sorry
<nutcase>ekaitz: np
<ekaitz>nutcase: the patch is good, yeah, the only problem is the amount of deps that we need to know how to deal with
<nutcase>ekaitz: I have another such one (73428)
<ekaitz>nutcase: that's a long one, it's too late to take a look now
<ekaitz>ping me another time and i'll try to take a look
<nutcase>ekaitz: no worries, there is already some awareness for that issue, I think.
<nutcase>thakn
<nutcase>thanks
<ekaitz>:)
<nutcase>gn8