IRC channel logs
2024-01-26.log
back to list of logs
<bumble>$(guix build pipewire) # /gnu/store/8572wxb5138hanhvn1lbfdm1kicxsd2k-pipewire-1.0.0 <sektor>What are the session arguments you folks plug into org mode when doing a literate design? <sektor>Trying to have Geiser connect to the guix remote REPL. <apteryx>has someone dabbled with the cuirass web interface? <podiki>I think ludo had added a thing at the top for when last check of branch was <user363627>repeating question from earlier: when installing guix on a foreign distro where are the manpages installed? <sham1>I'd wager that they're installed into the /gnu/store like anything else <bilke>I have a locally defined package which depends also on other repos than guix main repo. I create a channels file specifiny all required repos. Now I want to build several variants of my package by using time machine to get the correct repos. So the following works: <bilke>guix time-machine -C channels.scm -- build my_package <bilke>I would like to use guix `shell` to run multiple `build` commands but if I run <bilke>guix time-machine -C channels.scm -- shell <bilke>I only get the guix repo from my default profile and not the repos defined in channels.scm. <bilke>guix time-machine -C channels.scm -- describe <bilke>it prints out the repos defined in the channels file. <jpoiret>bilke: because guix time-machine doesn't put the time-machine'd guix in your path, it directly resolves the command <bilke>jpoiret: Thanks! What workflow would you suggest to get a specific (but disposable) environment defined in a channels file for multiple commands? <jpoiret>another thing you could do instead is `guix pull -p ~/my-time-machine-profile --allow-downgrades -C channels.scm`, then load that profile using `. ~/my-time-machine-profile/etc/profile` <jpoiret>or rather, `GUIX_PROFILE=~/my-time-machine-profile; . $GUIX_PROFILE/etc/profile` <jpoiret>guix pull can operate on arbitrary profiles, not just ~/.config/guix/current <bilke>jpoiret: The guix pull approach works fine! Thanks! <sneek>Welcome back civodul, you have 1 message! <sneek>civodul, rlb says: not sure what it means yet, but it looks like the test-signal-fork parent is hanging forever in scm_i_signals_post_for() when it locks up. <snape>Hi, can "guix download" be used with other things than plain https files? <snape>(like a git repo on a particular tag) <snape>well reading the cookbook it seems it cant <snape>one has to manually download it and use 'guix hash' <janneke>is there an easy way to diff two drvs? <jpoiret>heh, I have a script to do that somewhere that I hacked on during last year's FOSDEM <apteryx>janneke: easy no, but you can open them in emacs with emacs-guix installed, and M-x ediff-buffers <apteryx>if you do this often you could write an elisp script to do so <jpoiret>doesn't seem like it's in a usable state :( <apteryx>civodul: any tip about modifying the CSS for cuirass? I don't get how it's laid out <apteryx>I was trying to understand why the compact (burger-shaped) <apteryx>menu, when using a narrow display, doesn't open on clicks <apteryx>another thing I'm seeing in Icecat: Source map error: Error: request failed with status 404; for the .js files <apteryx>the button that doesn't work has the class "navbar-toggler" <janneke>apteryx: yeah, that's sort of what i'm doing now <janneke>iwbn to have something that sorts the depedencies by name (not by hash) <apteryx>it was proposed to move the hash after the name in the past to have store items sorted natively <apteryx>that's also make things like ls /gnu/store/bash* trivial and cheap <janneke>apteryx: yeah, i guess i was one of the people suggesting that... <janneke>but apparently we'd need a transitions scheme that isn't obvious, iirc <janneke>(apart from reaching consensus on the change itself, of course) <civodul>apteryx: i noticed that bug too (on narrow screens) but never took the time to investigate <apteryx>I'm trying a potential fix; I believe data-bs-target should use "#navbarDropbown", with the hash, instead of just "navbarDropdown" <apteryx>rekado: the only imperfect thing I can see with my mumi patches is when navigating by keyboard, the new coppy-message-id button outline is off-centered <apteryx>but that seems to be caused by interactions with Pico I don't know how to fix (it resolves if the 'padding: revert;' CSS directive is removed (but then the layout is borked) <apteryx>I don't think it's a deal breaker as it's only visible when tabbing the elements with the keyboard <rekado>apteryx: why do you use role=button when you’re overriding the styles associated with a button? <rekado>it’s not really a button then, is it? <rekado>you don’t need the reverts if you remove role=button afaics <apteryx>rekado: because it *is* semantically a button, and I want the browser to turn the pointer into a little hand, perhaps other stock behaviors <apteryx>but the Pico CSS styling gets in the way here <apteryx>but re-looking at it, it seems the pointer doesn't get turned into a "hand" anyway because of the tooltip <rekado>you can get the hand with “cursor: pointer”, I think <apteryx>I can get rid of role=button and the Pico CSS reverts, and I get the same result <apteryx>the icons are somehow off-centered compared to their flex container (div.header-buttons) <apteryx>my previous assumpting appears to be wrong (that Pico was at cause) <rekado>you give the buttons a size (1x1 rem) but the contents are actually larger. <apteryx>should I hardcode everything as pixels? I wanted to avoid that <rekado>one reason for the misalignment is “vertical-align: middle” for the svg element <apteryx>one thing I haven't researched well is the 'viewBox' property of the SVG <apteryx>it seems we can't set it to a relative unit without getting warnings in icecat, so I left it at its stock 0 0 16 16 (that's in pixels <rekado>message-button is 18px (effective) high, but that includes the border <rekado>so the actual contents are *less* than 18px (~ 1rem) <rekado>perhaps you can remove the height rule on message-button <weary-traveler>good morning, before i reach out to the mailing list, anyone here know how to install the manpages for guix on a foreign distribution? the binary installation seems to install the info pages, but i can't locate the relevant manuals <apteryx>if the guix installer doesn't do that already <apteryx>looks like it does so, in a file /etc/profile.d/zzz-guix.sh <apteryx>which should be sourced upon relogin in <weary-traveler>apteryx: that's for the info pages, correct? yes, i see those. i'm talking about the manpages accessed via "man guix-gc" etc <apteryx>rekado: ah! I could improve things by setting the icons size in css directly, and leaving the container size alone as you suggested <apteryx>seems we should also set the SameSite attribute for cookies <apteryx>as well as "The byte stream was erroneous according to the character encoding that was declared. The character encoding declaration may be incorrect. 33299" <rekado>…why do we set any cookies in mumi? <apteryx>I guess (current-output-port) is not opened in utf-8 by default in sxml.scm? <rekado>the theme switcher uses the localstorage API, not cookies <apteryx>I have a "session" named cookie for mumi <apteryx>maybe created automatically by one of the libraries we use? <rekado>I once added this as one of the prerequisites of enabling the comment form <rekado>to block the simplest vector for spam I would set an encrypted cookie that would be read when validating the comment form <rekado>the comment form had been disabled a long time ago, though, so the cookie is not used for anything <apteryx>OK; we can keep it but try setting the SameSite attribute to get rid of the warning <apteryx>is does 'run-server' from (web server) call the body proc with a UTF-8 encoded port? <apteryx>ACTION got lost trying to follow the (web server) code <old>does Guix has a mechanism to detect ABI breakage? <apteryx>about run-server using which encoding, I think the procedure dealing with that is response-body-port <guestinator>hi guix, does anyone have experience with using a cross-gcc-toolchain interactively via a manifest? <guestinator>cross-gcc-toolchain was recently added to easily make cross toolchains, which is nice <guestinator>but I’m not really sure how to add libraries to its sysroot <guestinator>I think because they aren’t native search paths, variables like CROSS_C_INCLUDE_PATH aren’t exported in the guix shell <villageidiot>I am trying to use `guix shell guile-goblins` to try out goblins but when I enter a guile REPL and run (use-modules (goblins)) I get back "no code for module (goblins)", where did I go wrong? <villageidiot>Might be a question for #guile but it's a bit in-between. <apteryx>rekado: I've redeployed mumi with v4, I think we're good <dariqq> villageidiot: I think you need to also add guile to guix shell to get the guile environment variables updated <jackhill>villageidiot: actually have you tried `guix shell guile guile-goblins` you made need guile there to trigger the search path logic <villageidiot>Haha let me give it a shot! If so I've learned something new... <apteryx>rekado: I tried SameSite "Strict" in the cookie-header alist, but that didn't help <apteryx>not sure where that 'set-session' procedure comes into life too <redacted>I'm writing some NodeJS software and I want to vendor my dependencies to make packaging for Guix possible. Anybody tried this? <apteryx>to test cuirass from source (pre-inst-env), is a default (service postgresql-service-type) service OK to satisfy its database needs? <apteryx>does set-xorg-configuration not work with GDM? <ieure>apteryx, I have the same config and use GDM and my keyboard layout does take effect. <ieure>Maybe you're launching a Wayland session? Not sure of the status of Wayland in Guix, because I dislike and don't use Wayland. <ieure>Or it might be GNOME messing things up, you can make keyboard layout changes in its settings, which would take precedence over any server-level configs. <apteryx>I'll check in gnome settings next time I login into gnome <kara_>Hi, im trying to figure out how i can get Python with Tkinter Support to work? I've installed the packages for guix shell with python python-pip and tk but python seems not to be configured for tkinter. I know that other distros have packages like python-tk. Maybe i have to compile python with tk support. Thanks for any advice <apteryx>'tk' is one of the two outputs of the Guix python package, as you can see with 'guix show python' <apteryx>don't we have a postgresql-service-type anymore? I don't see it documented in our current manual <Guest14>Since Linux Libre 6.7 is now supported, does that mean I can use bcachefs? <janneke>crap, guix copy is broken in the core-updates hurd-vm <PotentialUser-37>Alright, I'm tired of these <expletive> slow aarch64 sustitute servers. I'm willing to budget money this year to try and contribute to the build farm. Can I just donate on the website with a note to use it for arm sustitutes, or do I need to do research and buy specific harware and send that? <PotentialUser-37>90% of my computer usage is Guix on PureOS on Librem5, so seems like a good investment <apteryx>PotentialUser-37: the most readily usable form of donation would be some ready to use server (hosted) solution that we can control. <PotentialUser-37>brb 1 hour for luch, do want to discuss this further. Gessing best aproach is buy x86 harware and run arm build in qemu? Or are there big arm servers I haven't heard of... <apteryx>anyway to filter by QA OK'd patches directly on mumi? <apteryx>PotentialUser-37: QEMU emulation is a no go for anything serious; it's just too slow <apteryx>nice to hear that other people are also enjoying Guix :-) <ulfvonbelow>in your experience, is it usually worth passing --max-jobs=... in order to get more things building in parallel? It seems to speed thing sup somewhat, especially when lots of small builds are happening, but the tradeoff is you get way more nondeterministic build failures due to timeout issues <apteryx>it shouldn't fail more, unless your resources limts are reached <weary-traveler>what's the standard way to refactor and reuse phase modifications (performed via modify-phases) across package definitions? i know one way of doing it would be by defining a build system, but is that the recommended way when the goal is to modify the phases to allow a test-framework to work as intended? <apteryx>isn't the Cuirass README 'guix shell' command to setup an environment forgetting to share the postgresql database? <weary-traveler>case in point, in order to use "eldev" for running tests on an emacs package, three things need to be done: 1. tests enabled, 2. a specific test-command given, 3. a phase added after 'unpack to install an Eldev-local file. how can these three things be refactored so it can be used for different packages as needed. <apteryx>weary-traveler: you could use a functional approach and define a function that takes arguments and outputs a package <apteryx>allowing to reuse the same boilerplate <apteryx>or you could have some base package, and inherit from it <weary-traveler>apteryx: i think the vernacular for the first one would be "package transformation" correct? do you have a reference (info page or otherwise) of package inheritance? <apteryx>when I run './pre-inst-env cuirass register', I get: connection to server on socket \"/var/run/postgresql/.s.PGSQL.5432\" failed: No such file or directory\n\tIs the server running locally and accepting connections on that socket? <apteryx>weary-traveler: I see package transformation more as a way to rewrite the package graphs, e.g. you'd like to use a variant of X package to be used across a collection of package <apteryx>it could be used, but I don't think it's the easiest nor the most convenient option for what you seem to be after <weary-traveler>apteryx: okay then i'm not sure i follow. do you have a reference (or two, one for each proposal of yours)? <apteryx>your best help will probably be the code base; inheritance is often used (package (inherit x) ...) <weary-traveler>okay thanks for the inheritance reference. and for the first proposal, it's not clear what the argument to said function would be <apteryx>for package generators (procedures), there are some examples, such as make-u-boot-package <apteryx>sometimes implemented as syntax such as make-openjdk <ieure>attila_lendvai, Congrats! Moved most of my stuff there a few years ago and have been happy with it. <lispmacs[work]>Hi, I'm trying (for the first time) to build and install a local scheme package (my own) using guix. I (also for the first time) used hall to create scaffolding, and used "hall build" on it. <lispmacs[work]>I wasn't quite sure what to do next. I tried running "autoreconf" and that creates the configure file <attila_lendvai>ieure, i'm abusing it a bit, and the code is holding up. it's behaving as i expected it, which is surprising... :) (created a second account to hold my patches, then transferred those repos under an organization in the first account -- without any hickups) <lispmacs[work]>./configure: line 2540: syntax error near unexpected token `3.0' <ieure>attila_lendvai, It's been good for me. I have Gitea running locally, not exposed anywhere, and it push-mirrors out to Codeberg for public consumption. Excited for the federation stuff that Gitea's working on. <apteryx>is there no cuirass command I can use to have it populate its database? <apteryx>ah, I need to use a postgresql-role in my postgresql configuration, I think <apteryx>I'm trying to launch a local cuirass from its pre-inst-env without relying on the cuirass-service-type machinery <apteryx>I think I need to add a postgresql role for my own user <apteryx>otherwise I get the SQL database related error I posted above <apteryx>does running postgresql-role-service-type automatically starts (require) postgresql-service-type? or are they independent? <apteryx>hm, the (service-extension postgresql-role-service-type ...) example from (info "(guix) Databases Services") gives me an error: In procedure service-kind: Wrong type argument: #<<service-extension> target: #<service-type postgresql-role 7f7be585d7c0> <civodul>apteryx: the README shows how to run a test instance, with pg_tmp instead of a full-blown postges <apteryx>ah, I was reading linearly and stuck on attempting to run the first command: ./pre-inst-env cuirass register <civodul>right, maybe we should reorder things <apteryx>I think it could be fine to test using a full-blown postgresql server, but we should provide the necessary Guix System snippet <apteryx>if the necessary info was provided we'd have two valid options to test with <rekado>civodul: huh, so in December 2023 Guix suddenly had -8 comments. <PotentialUser-37>apteryx: Do you have any recommendations on arm hardware for such a cluster? First result I found is from Nvidia, and I'd really like to avoid giving them money... <PotentialUser-37>And I assume guix in a container would be fine (via guix on a foreign distro), as I beleive the booting options for arm are slim? <apteryx>Last I read ARM was making some standard for making booting via EFI easy. If you get one of these it should be able to boot Guix System, which is the prefered option. <rekado>ACTION updates the repository URLs there <civodul>rekado: what happens when it’s given several repos? does it aggregates stats?