IRC channel logs

<folaht>I created a python symlink to python3 in ~/.local/bin and appended $HOME/.local/bin to PATH in ~/.bash_profile. Is this guix compliant?

<mange>Yeah, that's fine. You could also install the python-wrapper package (instead of python) which does something similar.

<dajole>Is there a reason the latest release is signed by Efraim Flashner instead of Ludovic Courtes?

<vagrantc>because Efraim was part of the Release team and Luduvic was not :)

<dajole>I see. Is there a place to submit a suggestion to add this to the documentation?

<vagrantc>codeberg.org, although it has been a bit unreliable today ... they might be doing some upgrades, or getting attacked by another botnet

<seres_>there is a PR open to reflect this change in the manual already

<dajole>It looks like they are getting DDoSed: https://fosstodon.org/@codebergstatus@social.anoxinon.de/116173195095377866

<dajole>Thanks! So for now it's save to just import Efraim's key to verify the signature?

<vagrantc>i wonder if in the future we shouldn't have a signed team transition statment or something

<vagrantc>or... wouldn't even hurt to create that retroactivelyu

<vagrantc>dajole: as long as you get the right key :) ... key trust issues are tricky.

<folaht>I get a "Git error: failed to connect to codeberg.org: Connection refused" when trying to run 'guix system reconfigure /etc/config.scm'

<folaht>Is that due to the DDOS attack or have I made an error?

<vagrantc>probably the ddos

<dajole> https://status.codeberg.org/status/codeberg

<vagrantc>ACTION is curious what system configurations need to connect to codeberg ...

<vagrantc>i guess a new guix-daemon version? although that should at least also try one of the fallbacks

<vagrantc>some configurations require running guix pull ... but i try to avoid that

<mange>I think it does a clone to build the guix package?

<folaht>vagrantc, my system configuration is pretty basic.

<folaht>So unless virtualization services need to connect to codeberg, all of them do?

<vagrantc>mange: yeah, but that should fall back to softwareheritage or other locations

<vagrantc>and should also generally pull the guix package from substitutes...

<vagrantc>or, to avoid confusion, and should also generally download the guix package from substitutes... rather than running guix pull

<ieure>I've noticed that system reconfigures over the last ~3 days have all compiled guix for some reason.

<vagrantc>not too uncommon for inputs changing to trigger that ... but yeah, it happens ...

<vagrantc>directly or indirectly... because this is guix :)

<apteryx>hm, not sure what this means: https://paste.guixotic.coop/_shell_-7645691-7645969.html

<apteryx>there *is* a remote named origin with a HEAD ref

<ieure>Oh damn, Codeberg is still down? Unfortunate.

<podiki>it was up a bit, and ssh was working, but seems the ddos continues (and now they've gone to sleep)

<podiki>sneek: later tell efraim I've pushed your mesa changes (and updated to latest release) to mesa-updates to build for now while discussions continue on the 2 approaches

<sneek>Got it.

<amano>Is `guix deploy` a good alternative to terraform?

<amano>Or, is `guix deploy` not a full substitute for terraform?

<csantosb>Good morning Guix ! Noticed that qa.guix.gnu.org has a serious something going on

<janneke>o/

<janneke>prolly codeberg ddos related

<darthlukan>I think there was something about a DDOS on the Status page - I don't have the Mastodon message link, but did see ity

<darthlukan>Ugh, sorry, autocomplete picked up someone's name

<GuixSurKvm>Bravo seres_ et merci !

<GuixSurKvm>Tu as une sacré efficacité pour trouver les bonnes infos aux bons endroits. Malheureusement je n'étais pas passé par le wiki ne NixOs, je retiens.

<GuixSurKvm>C'est le GG !

<efraim>podiki: thanks

<sneek>Welcome back efraim, you have 1 message!

<sneek>efraim, podiki says: I've pushed your mesa changes (and updated to latest release) to mesa-updates to build for now while discussions continue on the 2 approaches

<efraim>sneek: botsnack

<sneek>:)

<futurile>what sort of muppet DDOS's a source code hosting site, like what is the point!

<folaht>I'm trying to run 'nix flake init', but recieve the error: '/home/folaht/.local/share/nix/root/nix/store/f9c35cwj3asd78n5rfyriqdhb6d0ykil-source/trivial' was not found in the Nix store

<folaht> If you've set 'defaultTemplate.path' to a string, try using a path instead.

<futurile>morning all

<folaht>Good morning futurile

<folaht>futurile, competition?

<folaht>Maybe it's microsoft wanting to force everyone to use github

<futurile>folaht: sort out your image/webm problem?

<folaht>futurile, I filed it as an issue

<futurile>folaht: I don't like Microsoft particularly, but that seems unlikely to me. Codeberg isn't even worth bothering about as far as GitHub is concerned honestly.

<futurile>folaht: no it will be some maladjusted idiot who thinks they're cool

<folaht>Is the DDOS attack still going on? I want to show futurile the issue, but can't access codeberg.

<futurile>seems like they are still down according to their status page

<gabber>is it just me or could we improve the "how to write (simple) services" section(s) in our documentation?

<folaht>As a user, I would say definitely yes.

<folaht>The documentation is good, but it's lacking a manual.

<folaht>manuals

<gabber>and also, contrary to the documentation, provenance-service-type was not addad automatically upon system reconfigure (did this change come with 1.5.0? i'm currently stuck in january with my pinebook pro [aarch64])

<gabber>folaht: are you by chance interested in helping to improve guix documentation?

<folaht>Uhmmm... I'm a new user. I would be if I knew how.

<folaht>Oh, I could help by start asking where the guix Howto's should be located at?

<folaht>Because there's Arch that puts everything in the documentation.

<folaht>And there's Manjaro Linux that usually puts it's howto's in the forum.

<folaht>As a category.

<futurile>overall I think services is really confusing, but I guess we're stuck with the naming now heh

<efraim>I didn't realize that mesa has a virtio vulkan driver

<futurile>virtio meaning can be used in vms?

<efraim>I would assume so

<futurile>that would be cool

<gabber>folaht: we in the documentation team could definitively use input from new users

<folaht>Why does the systemcrafters forum only have four categories by the way?

<efraim>heh, I have a mesa git checkout already, time to look around there

<gabber>folaht: you could join our documentation team (:

<folaht>gabber, well, my input would be to add a category in systemcrafters to add howto's.

<folaht>How do I join the documentation team?

<gabber>folaht: we're somewhat in the process of restructuring a bit our documentation. from the "almost everything goes into the manual" towards a bit more structured "there's the reference, there's how-tos, there are tutorials and finally, there are some in-depth explorations"

<folaht>What's the difference between how-tos and tutorials?

<gabber>folaht: by opening a pull request on codeberg adding yourself as a team member to the documentation team. but if you don't feel comfortable to do so, i can gladly do it for you

<gabber>folaht: a tutorial teaches people specific skills, how-tos explain how one can do stuff (the difference is mainly in the format that is written IIRC)

<gabber>see https://diataxis.fr

<efraim>virtio looks like it's integrated in (all? many?) other drivers, so it looks like it's not something we should skip

<efraim>for comparison, Debian enables it on all architectures built with llvm support, so almost everything

<folaht>gabber, I'm willing to write howTo's.

<gabber>folaht: cool! apart from writing, it's also mainly about reviewing what other people write and commenting on whether it is understandable and well written

<efraim>the vdpau directory disappeared between mesa-25.2.3 and 26.0.1

<futurile><sigh> my leg is really painful today, it's making me very irritable heh

<gabber>futurile: i feel you. my body is a wreck

<gabber>can't wait to upload my conscience to the cloud and be finally free of the pain /s

<futurile>it's funny when you realise the connection between brain<->body

<futurile>I expect that would mean watching more Ads ;-)

<jakef>just got a guix pull through!

<gabber>futurile: it'd be much funnier if there were ever such thing as relief

<gabber>jakef: \o/

<futurile>whoop!

<futurile>codeberg.org status show the web site as back up

<csantosb> https://status.codeberg.org/status/codeberg

<folaht>gabber, do I have to sign up anywhere?

<efraim>sneek: seen sughosha

<sneek>sughosha was in #guix 5 months ago, saying: Hi Guix maintainers, I recently got commit access to Guix repository. I have my GPG key added to my Codeberg account (via Codeberg's user settings). Will merging from the web interface also sign the commits?.

<efraim>sneek: botsnack

<sneek>:)

<cbaines>the nar-herder on bayfront has been broken for most of the last day

<cbaines>I'm trying to restart it, but it hasn't started working just yet...

<gabber>folaht: yes, on codeberg. and clone our repository. and open a pull request.

<gabber>the first step is necessary (account on codeberg), the others i can do for you

<gabber>it's easiest if you message me your (codeberg) username and email address

<gabber>does SDDM turn on numlock when showing the login screen?

<cbaines>right, bordeaux.guix.gnu.org should be back now

<folaht>username: folaht

<folaht>email: folaht@protonmail.com

<gabber>efraim: are you looking for sughosha?

<efraim>gabber: I was unhappy with the description for qtvirtualkeyboard, but then I saw it was there also for the Qt5 version, so I just changed both myself

<gabber>i see

<efraim>gabber: sddm turns on numlock by default

<gabber>and we got rid of numlock as of 1.5.0, right?

<gabber>i just crafted this "disable-numlock" service but i guess that was for nothing

<efraim>I have no idea. I just know when I boot my keyboard has the light on but it doesn't do anything, and then it turns off when I login

<efraim>there's also a numlock option for the sddm-service-type

<gabber>on the pinebook the "numpad" overlays the default keys so my password wouldn't work

<gabber>i guess that'd be the way to go, then (:

<gabber>thanks!

<efraim>I turn mine off on the pinebook pro in the service

<gabber>ACTION could have just checked efraim's configs from the beginning

<efraim>😜

<efraim>if you can figure out why the qemu-binfmt-service-type causes boot to hang I'd be very happy

<efraim>what desktop are you using? I've mapped some of the keys for sway

<gabber>sway

<gabber>did you recently pull and reconfigure?

<gabber>i'm stuck on some beginning-of-the-year commit

<efraim>I actually have barely used it in the past year since I got my X1 carbon

<gabber>and you use qemu-binfmt-service-type on the pinebook?

<efraim>I tried to

<gabber>this could be out of reach for me to investigate today

<gabber>what do the logs say?

<efraim>I don't remember, it's been a while

<efraim>I don't remember there being logs

<gabber>IIUC we had quite some updates to aarch64 with 1.5.0, maybe a reconfiguration solves your problem(s)

<gabber>ahh

<efraim>I don't think I ever got it working

<gabber>even through serial?

<efraim>I know I didn't try that

<gabber>efraim: my i ask what you were trying to achieve? IMHO the pinebook's already tortuously slow to compile native stuff

<efraim>probably just feature parity with x86_64

<gabber>csantosb: are you here?

<gabber>should i craft (a third) PR moving pybind11 to pybind11-2 and using your definition for the new pybind11 (v3)?

<gabber>it feels kinda wrong to have 3 PRs addressing (almost) the same thing but at the end we also get to close 3 PRs

<jab>morning guix people!

<gabber>jab: \o

<jab>gabber: how's what the PR pybind11 thing that you are talking about? What python problem does it solve ? I'm completely unfamiliar.

<gabber>we're trying to package librelane. this relies on pyosys which in turn depends on pybind11 (version 3). we currently have pybind11 (v2)

<gabber>now there is 1 PR addressing the pybind11 update for the python-team branch, one from csantosb upgrading the package (for master) and now the third variant, that moves the current definition to the variable pybind11-2 (a pinned package) and adds the new package definition (v3) under the variable pybind11

<jab>what's librelane ?

<folaht>If I want to do a feature request, do I have to join the developers mailing list?

<jab>folaht: nope!

<jab>You can just send an email to guix-devel@gnu.org ... but may I give you some advice before you do so ? It'll give you a better chance of getting it added.

<folaht>Okay, gimme advice!

<folaht>jab, I'd love to hear your advice. I want to see my feature request implemented.

<jab>sweet.

<gabber>folaht: what's your request?

<jab>So guix-devel@gnu.org is usually meant for developers to propose a new feature, but you can of course request a new feature. I would encourage you to say something like, I would love this feature to be added to guix, and I'm willing to $xx dollars to see it added. Is anyone willing to add it for me ?

<jab>and/or you could contact the guix corporations directly that hack on guix. guixotic is REALLY nice. My friend is currently paying them to package sparrow wallet for guix.

<jab>folaht: and maybe ask for the feature here in irc. possible gabber can tell you that we already have it.

<untrusem>guix is not a corporation

<folaht>Is guix an NGO?

<untrusem>its a non profit

<untrusem> https://foundation.guix.info/

<folaht>Well the feature is what we've just discussed. Make this page: https://guix.gnu.org/manual/1.5.0/en/ look more like this -> https://docs.python.org/3/, so enlarge the manual from just a reference, to also have a tutorials and howtos

<folaht>*tutorials and howtos section.

<untrusem>there is guix cookbook

<untrusem>that have tutorials and how to

<untrusem>we need to expand that

<gabber>folaht: aren't you on the documentation team?

<untrusem>I also had an idea of unofficial community wiki

<gabber>the goal is https://diataxis.fr

<gabber>but getting there is quite the path

<gabber>let's make it work (together)

<untrusem>yeah i remember you sharing it before

<folaht>Oh, I see now. I missed the cookbook.

<futurile>it also needs expansion heh

<jab>untrusem: guix is not a corporation.

<jab>But there are people that work on guix for $

<jab> https://guix.gnu.org/en/commercial/

<flurando>what's going on, any compilation of configs requires building of guix-1.5.0 which just fail saying no build log.

<untrusem>jab: yep I know about that, either way thanks but those are not corporatiots either but cooperatives

<untrusem>flurando: could you elucidate?

<flurando>what does elucidate mean, I'am sorry not understanding this word.

<untrusem>my bad

<untrusem>can you explain more

<untrusem>also, logs would help

<flurando>the guix pull of 8afd7d3 could not build guix-1.5.0, which is triggered in any thing like guix system reconfigure or guix deploy.

<flurando>now trying the latest guix, 13346ec

<flurando>Well, I guess I have to try again hours later, as I currently have stuff to do without my laptop, which is not safe without myself watching.

<flurando>By the way, It is /gnu/store/cppz0w4hz4qjd1vlmp32k3kbyc19hrma-guix-1.5.0.drv that could not be built, failing at 100% check saying no build log found. Unfortunately this is triggerred by any recent system reconfigure or guix deploy, evaluating unpinned system configurations.

<flurando>trying again with guix-daemon restarted (to use the new guix pull), sorry if this works

<jab>folaht: the cookbook could use some work. I think the "automatic login to virtual console" has been broke for quite some time.

<jab>it's been a while since I've tried that feature though.

<gabber>flurando: quick fix: revert your guix pull (e.g. `guix pull --roll-back`) and reconfigure your system

<jab>I do like this page's https://docs.python.org/3/ API reference. Seems pretty clear.

<gabber>flurando: once your configs work with the "old" checkout you can try "fancy" stuff, e.g. `guix time-machine -- system vm your/sys/config.scm` to see if reconfiguration would work

<folaht>jab, I just read chapter "5 Virtual Machines". I would not have been able to set up a vm in virt-manager with this manual. But I'm a casual user. I'm not sure who the audience is.

<flurando>gabber: I understand, but I do guix pull because the previous guix I was on has another bug with guix deploy, missing '' or "" in the remote exp, which is fixed by the current guix pull. Personally have not tried guix system vm stuff, thanks for the tip.

<untrusem>folaht: can you gather your feedback in a document or something, so that it can be addressed without its getting lost

<untrusem>maybe as an issue on repo

<flurando>folaht: me as a casual user, just using gnome-boxes...

<gabber>folaht: for each and every stumbling stone you see, can you please open an issue on codeberg with the documentation-team label?

<gabber>we could do a Milestone (some issue/pr abstraction) where we gather all the issues we currently have and work them

<flurando>gabber: actually it is confusing how to contribute to documentation, as seemingly we have to have a development folder of guix setup and try to compile the documentation before sending a patch. Why not just change the texi files then use some CI to check the syntax?

<gabber>flurando: you are right!

<gabber>major parts of the guix contribution ecosystem rely on manually doing (and checking) stuff, which has come to be a (current) bottleneck in our project

<folaht>gabber, how does one add a label? Is that only after the post has been made? The brang/tag button is greyed out for me.

<gabber>the intention of keeping the (reference) manual in sync with the codebase seemingly makes sense, but i am not sure how i feel about other parts of the documentation (which are currently either incorporated into the manual, the cookbook or in the webs distributed on some blogs)

<gabber>folaht: if you tell me the PR number i'll add it

<gabber>i think one needs codeberg project (guix) privileges for that? i am not really sure how i could grant that (and if i should)

<flurando>gabber: true, and guix actually has its own CI tooling, the curiass. However, it is not clear how to implement traditional GitHub Action style CI to curiass that functions like the linux project's CI stuff.

<gabber>i don't know how GitHub Actions work, but we have our neat little cuirass-build-bot that does cute little builds and requests changes from time to time

<gabber>if you have guix installed, you can check if your changes break the manual by invoking `guix shell -D guix -- make doc/guix.info`

<flurando>gabber: but in many cases of CI usage, simply use curiass to compile and build the packages is not enough, which only ensures stuff works instead of stuff nice.

<gabber>WDYM by "stuff nice"

<flurando>gabber: for example there is coding guidelines and somehow the community wish all the source code obey them. It is duplicate to write checks in each package wrapping, which would be much better if some checks could be applied to a group of definitions in curiass.

<gabber>flurando: WDYM "duplicate to write checks in each package wrapping"? the check phase is part of (most if not all) build systems, we usually do not add extra checks

<flurando>make doc/guix.info could only say whether it builds, but it does not say whether the change is nice, in a sense without inapproperate words or phrases or similar, which requires manual check later.

<gabber>flurando: well, we heavily rely on peer-reviewing our changes. and i wouldn't know how Microsoft could do this magic adequately, appropriately and with sufficient reliability

<gabber>but i agree: if we could just push changes and there were this giant green button some of us could push and stuff just happened that would improve the general situation that'd be great

<flurando>gabber: well, "duplicate to ..." means nearly every definitions include the obey-code-guideline in their check-phase

<gabber>also kinda scary, but still very cool

<gabber>flurando: i don't think i understand... could you point me to an example?

<flurando>it is scary in codes, but in the area of documenting, it is not scary at all.

<flurando>gabber: sorry, I was just dreaming how things (document contributing) could be easier instead of comparing guix to some existing project (Linux is just used to prove CI could be very complex).

<gabber>don't be sorry for dreaming!

<gabber>dreaming means inspiration and inspiration can lead to improvement, which is usually nice (:

<flurando>fine:)

<flurando>By the way, if someone else here is insterested in getting a server on guix, see https://fanbox.xn--x9pw31j.com/tutorial/install-guix-on-full-root-vps.html for VPS and https://fanbox.xn--x9pw31j.com/tutorial/setup-home-server.html for minipc at home. This is my personal site without ads, but I suggest visiters disable javascript as it does have cloudflare scripts logging!

<flurando>well, restarting guix-daemon makes the guix-1.5.0 build pass, so it was not guix commiters' fault, my bad.

<flurando>Actually I would be happy if my two flashing guix tutorial could be on the official guix cookbook, with some warning as it uses third party scripts on github. Give me feedback with contact info listed at https://blog.xn--x9pw31j.com/page/about.html, chinese site this time but this page is in pure English.

<gabber>flurando: If you craft a commit and open a PR chances are quite high that this will happen ;)

<gabber>sorry, my stack is currently almost overflowing, so i don't have time for extra steps. but if it lands on my (or the team's) desk, it will (eventually) be reviewed

<flurando>gabber: really, I would like to do that for the vps one, since minipc one only applies to scenario where one does not want to use Udisk to install Guix. There might be some modifications as it is not approperate to include my ssh pub key in guix cookbook...

<flurando>As my fanbox site is consisted of org files actually, I would like to send the modified article just in org by email.

<untrusem>+1

<flurando>However, my article consists of links to my personal github and codeberg projects, leaving me wonder whether these are acceptable in guix cookbook.

<untrusem>+1

<untrusem>flurando: you can edit those parts out

<flurando>untrusem: Yep, however, the links are not social links, whose content actaully required. You can check it out in spare time https://fanbox.xn--x9pw31j.com/tutorial/install-guix-on-full-root-vps.html with js disabled.

<flurando>the first link is to some script on github used to flash the vps, the second link is in the end, which readers could follow the article with the repo in hand, like a simple codelab.

<flurando>I can assure you that the functionality used in the article is personally checked and tested by me, and any further check from community or personal forks are welcomed.

<redacted>That sounds like you're offering a program to do something. That doesn't sound correct for cookbook content.

<flurando>redacted: Right, that is exactly my concern.

<flurando>However, I offered detailed guides in the article, and a corresponding repo to really do it.

<flurando>Which, a common VPS running guix system, would otherwise be impossible for newbies given they mostly would not happen to have found my article.

<flurando>Not a program actually, a set of scripts, too much to be included in the article.

<flurando>But if anyone would like to examine the github script link and apply the guides (or happen to trust me), they can use guix system with any vps providers with full root access, not matter whether guix is supported at all.

<untrusem>will check later

<redacted>Maybe you should package the reinstall script for Guix. If it's in the official channel, there'd be no need for a link anymore.

<flurando>I think this is very useful if the external "program" concern could be handled. I wrote this because someday I really wanted to use Guix system on Recknerd but found nothing, so desperate until I found the script, forked and tested it until everything works in the end. By the way, I only checked the dd functionality in my fork, so it is possible for viruses elsewhere.

<flurando>redacted: No, this script only works on Debian... I have no clue how to run it on Guix.

<flurando>It would be really strange for a guix package that contains "blobs" in textual form, not assumed to be used here.

<flurando>My idea is everyone either trust someone in this community that has forked and examined the script (like me), or fork and examine the dd functionality themselves.

<flurando>After then, just clone my codeberg repo GuixCloudInit and follow instrustions there. The codeberg repo, instead, is tiny and cleaned in my view.

<flurando>Also, if guix cookbook is not the most approperate place, this is welcomed to be reposted on anyone's personal blogs, or even better, the blogs part of the official guix website as a new blog.

<flurando>Well, the remote-exp missing '' after -c is not fixed in the latest Guix 13346ec, I would fire an issue on codeberg later when I have time.

<futurile>gah! second force push of the day

<flurando>Does anyone know where in the guix src code is generating remote exp command located, I checked guix/remote.scm and guix/deploy.scm, no finding. Or is there anyone else facing issue when doing guix deploy? I mean the issue that remote-exp missing '' around the exp, causing the deploy and ssh staled, then only force reboot could let me ssh into it again.

<flurando>interestingly, only the remote exp containing elogind service type info would miss the '', I am wondering why.

<jakef>do we no longer require contributors to sign commits in guix-science?

<jakef>oh i just can't read codeberg, carry on

<flurando>Really confusing when adding elogind service makes remote exp missing '', errors are expected, but not this.

<civodul>comrades, i have created a milestone on avoiding tarballs with pre-built artifacts 👉 https://codeberg.org/guix/guix/milestone/66679

<civodul>janneke: 👆

<yelninei>how would that work with things like help2man?

<untrusem>civodul: how would exceptions work, for instance there is a package called oils-for-unix , which uses a tarball which is transpiled to c++ from python source by the projects itself, enge suggested us to use the source itself but the author of the program was against it

<untrusem>aah read the milestone again, i don't think it will effect all the packages

<postroutine>Hello. I have installed Guix on a "foreign distro" and I use Guix Home. I have seen that Shepherd is running on my user account. Does it run automatically after the installation of Guix or after the reconfiguration of Guix Home ?

<untrusem>are you using any home services?

<postroutine>Yes: home-bash-service-type, home-dotfiles-service-type and %base-home-services.

<postroutine>In the documentation, I didn't found the list of services in %base-home-services.

<ieure>postroutine, That's not usually in the documentation, you can look at it in the source code.

<emacsomancer>Ah, I see guix weather forecasts ungoogled-chromium substitutes available again.

<emacsomancer>is there any hope of Guix's ungoogled-chromium tracking more closely upstream ungoogled-chromium?

<emacsomancer>it's a bit worrying being 5 major versions behind

<yelninei>janneke: the check tests still seem to hang. should we skip them for now?

<yelninei>(a hang in this case is a full system freeze)

<yelninei>this again seems to work on debian, maybe it is something from libc?

<jab>emacsomancer: contribution welcome. :)

<jab>I feel like writing bash scripts is ugly.

<jab>I imagine that writing assembly is worse...

<Kabouik>We have emacs-claude-code and emacs-claude-code-ide in the channel, but does anyone remember how to install the actual Claude binary that is a dependency for it? I have been using it for the past week successfully actually but I can't for the life of me remember how I installed the binary initially; the bash install.sh procedure on their website does not work (seg fault, I tried more than 20 versions), and now that I broke my install trying to update

<Kabouik>using that method, I can't remember how I did it the first time and restore a working state.

<postroutine>I have launched a reconfiguration of my Guix Home config some time ago. But Guix is stuck in "check" phase of the build of "guix-1.5.0.drv". I have no error message or more info. Where can I search infos to unstuck it ?

<ieure>postroutine, It's not stuck, just slow.

<ieure>I am still not sure why reconfigures are compiling Guix.

<postroutine>The previous reconfigure where more quick and I didn't change anything in it.

<ieure>postroutine, Yes, but other things did change.

<postroutine>And my home config is small. A few package, home-bash-service-type, home-dotfiles-service-type and %base-home-services.

<postroutine>ieure, did you mean on Guix side ?

<ieure>postroutine, Yes.

<redacted>emacsomancer: the issue with ungoogled-chromium in particular is, I think, the long build times.

<redacted>That makes iteration on the package quite tedious.

<ieure>Is the build substantially longer than Firefoxen? It takes ~60-120 minutes to build LibreWolf, depending on the machine.

<ieure>Add ~2 hours if it needs a nss bump.

<redacted>Not sure. Maybe I'm projecting a bit. When I took a stab at updating it, the build times were too long for me to want to keep at it.

<redacted>I may try either offloading the build to another machine on my network or a VPS with more sauce.

<ieure>redacted, That's pretty much what I do. I have a decent speed box mostly dedicated to cuirass and it'll build the librewolf-updates branch of my Guix fork. I typically start a build and let it run a couple minutes, just to make sure it's not going to fail immediately, then push to the librewolf-updates branch and let the build box chew through it.

<ieure>redacted, We can certainly set up an ungoogled-chromium-updates branch on CI, if that'd help.

<ieure>I need to do that for LW as well, since untrusem has been contributing and would benefit from offloaded builds.

<redacted>I'm invested in doing the work of updating Ungoogled Chromium if I have the hardware available.

<redacted>So a branch on CI would help, yeah

<ieure>redacted, I can facilitate that in a couple hours.

<redacted>Thanks!

<ieure>I personally have generally negative feelings about all Chromium-based browsers, but I understand they're sometimes needed/preferred, and we shouldn't put Guix users at risk by shipping such an old version.

<ieure>I needed it last week for a WebUSB thing, which Firefoxen inexplicably still does not support.

<ieure>No delay adding AI junk I don't want, though.

<redacted>Yeah, I use Librewolf for almost all internet browsing, but I *also* needed Chromium for WebUSB. And sometimes for sites that don't work right with LW

<ieure>I'm cranky enough these days that if a site doesn't work in LW, I don't use it. But WebUSB, yeah, I do like to use that.

<ieure>Don't use the site, that is. No patience for this revived IE6-era trend of only working in one browser.

<janneke>yelninei: yes, let's skip them

<janneke>i think it's better/easier to skip them than to start upgrading glibc on hurd only again, and have two glibcs, wdyt?

<yelninei>we already have 2 glibcs, i split them last time to get the replacement for the intr-msg-clobber patch

<janneke>yelninei: ah, and oh -- sorry for not being up to date :)

<podiki>is this glibc stuff on core-team branch? we have that graft right now on master, i was wondering what branch we can ungraft on

<yelninei>i think skip it for now and have it magically fixed when glibc gets upgraded and we can reenable it later

<podiki>but if core-team has a while to go, i want to prioritize ungrafting earlier since the user experience suffers with so much grafting

<janneke>yelninei: very good, this is probably the easier path and save us (you!) time now

<janneke>the glibc update will come some time anyway

<yelninei>also changing check in the future is easier than libc

<janneke>yelninei: ok

<andreas-e>lilyp: I have stood on the shoulder of giants and fixed gp-saml-gui :) Also more of a removal candidate, to be honest, but well...

<andreas-e>(gnome-team branch)

<andreas-e>hydrus-network is probably due to the build agent it has been built on, I would disregard it.

<mange>Any committers around who want to commit a fairly trivial fix? https://codeberg.org/guix/guix/pulls/6896

<gabber>mange: let me have a look

<gabber>i could push that but i am not sure what this commit does :(

<gabber>can you tell me how i can see the issue with the current `vis' package and how i can confirm the fix with your patch?

<andreas-e>lilyp: tuba/tootle: Repaired by an update.

<gabber>andreas-e is on fire tonight, huh!?

<gabber>AWESOME!

<mange>gabber: The related issue is https://codeberg.org/guix/guix/issues/6843

<andreas-e>Hehe, not really; rather tired actually.

<andreas-e>But I would like to see the gnome-team branch get merged, there is not much missing!

<mange>I don't use vis, I'm just interested because I committed the breaking change. :P Essentially Lua's search paths need ; as a separator, not :, and I broke it with a recent change to how Lua search paths are handled in Guix.

<gabber>mange: i've seen and read the description of your(?) PR but since i am not a vis user and have never(?) touched any lua packages i'd need some guidance to reproduce the bug/failure and the proposed fix

<gabber>ahhh, i see

<mange>Sorry, to be precise: I didn't commit the breaking change, I wrote it. I don't have permission to commit to Guix, so I just beg whenever I need something merged. :)

<gabber>nah, i guess you committed but someone else pushed it ;)

<andreas-e>lilyp: texmaker is also spurious.

<gabber>mange: so you are carlo? or phong? doesn't really matter, i'm just curious

<andreas-e>lilyp: As is svt-vp9.

<mange>Carlo. :) I've been considering changing my nick to czan to make this more obvious, but I haven't got around to actually doing so...

<gabber>not sure that is necessary

<gabber>people here use all kinds of nicknames

<gabber>mange: weird. syntax highlighting seems to work

<gabber>or does it?

<gabber>(define foo 'bar) is printed in bold and yellow

<mange>Yeah, give me a sec to see what's going on. I think it might be related to whether GUIX_LUA_PATH is set in the outer environment, because the issue is with concatenating the environment path with the wrapper path.

<gabber>i've tested both in a normal (tainted) shell and a --container'ized one

<mange>Yeah, so "guix shell lua lua-filesystem vis -- vis" won't have syntax highlighting.

<gabber>both with my current checkout (from today i think) and the commit that is referred to in 6843

<gabber>ah

<gabber>ok

<gabber>thanks for the clarification!

<gabber>you are right! now onto checking the fix

<mange>Conversely, sorry for not having clearer reproduction instructions. :)

<gabber>not ideal but no problem

<gabber>(:

<gabber>the fix works!

<mange>Huzzah!