IRC channel logs
2024-07-29.log
back to list of logs
<ryan77627>Hey everyone, question about packaging something that requires EGL on wayland. I'm currently trying to package hyprlock, have it building, but when i execute it I get the error EGL_EXT_platform_base not supported. Anyone have experience with EGL on guix? <ryan77627>I'm suspecting I may not be wrapping the app correctly? But, figured I'd ask here if it was common enough that someone already knew a direction for me to go in <lilyp>hey, does anyone have any pointers on using cgroups etc. inside the build container? <adanska>Hi Guix! Quick question: if an upstream package does not have any releases, what should I put for the version number? "0.0.0" or "1.0.0"? <adanska>i've finally got lem working perfectly!! going to start the process of moving it out of the package.scm and into the guix repo proper to contribute upstream <lilyp>try to find the closest thing to a version number you can get, they may be embedded in source or build files – then construct a git-version from that <Francima1>hi, is there any future hope to see zotero in the guix repos? <attila_lendvai>if i can fix a build by either 1) add a backported .patch file, or 2) disable a test, then which one is preferred? backporting a patch sounds better at first thought, but it adds extra cruft... disabling a test is a line in the .scm file with a comment to the issue's URL. <attila_lendvai>Francima1, packaging js projects is usually quite some headache because of the way node.js and related cruft is (not) organized. it's not impossible, but it's quite some work to package js projects according to the guix standards <Francima1>damn zotero, why is it so helpful and so js heavy :( <attila_lendvai>Francima1, another option is to get it packaged in a less clean way, and add it to a channel that is more permissive like the guix'r'us channel <attila_lendvai>Francima1, actually, zotero is already packaged in... that other channel... <attila_lendvai>IMO it's a general deficiency of the project that there's no central place to list the available channels... <attila_lendvai>i guess it would count as endorsement, and many of the channels are sub-standard in some aspects that guix sets out as values. <attila_lendvai>yeah, but you already need to find the toys website... a chicken-egg issue... <h4>2024-07-27#224523 dariqq: My problem is that since a Guix update ` (commit channel-commit (default #f))` don't seem to work anymore. It used to, but now it says "error: $hash: unbound variable hint: Did you forget a `use-modules' form?" <h4>I think maybe it used to work on a8353e9d6b34fd8d42d2e8f14ce844849fe9c293, but not sure, at least it worked for sure on 09b19cf8dabdcfeb7a79b0d69c46febf68a53639 <attila_lendvai>h4, that's very little context, i have no idea what your problem is. if it requires long explanation, then maybe you should ask on the guix-help mailing list <attila_lendvai>xerty, i can confirm, and that's unfortunate. maybe it's some automatic abuse prevention service that either blacklists all tor exit nodes, or it detected abuse from the ones we used and they are temporarily blacklisted. <h4>attila_lendvai: Each time I talk here, midway I tell myself that I should switch to mail. So much more convenient, but way less replies. Anyway my problem was that it worked without quotes before and now it works with, so it's resolved <attila_lendvai>h4, but what is "it"??? if you don't get answers, then one possibility is that you didn't actually ask a question. here you just pasted a line from some code from somewhere... at least use a paste service for a larger context, e.g.: https://paste.debian.net/ <dariqq>regarding my lightdm issue from yesterday: I reinstalled completely and now setting the default entry works. Not sure what was causign this <fnat>When defining a new service and its configuration, is the (general) recommendation to use 'define-configuration' over an ordinary record? <xerty>attila_lendvai: may be it is sth else, because when I use some techniques like tcb desync to modify tcp packets globally to avoid isp DPIs, GNU Icecat will also show SSL_ERROR_BAD_MAC_READ, though it is different from the out of time error on tor. <xerty>an it also occurs on ci and issues but not bordeaux <xerty>meaning I am connecting to those domains directly not through proxy. (but I've used nat (I might need to further investigate into it. <xerty>i think guix works on fedora, I think you might need to modify selinux policy? <xerty>at least it worked for my asahi fedora remix <nikolar>you should probably disable selinux if you want the simplest solutions <nikolar>the usual security warnings and all that apply <worni>I have a weird issue where occasionally websites take forever to load and `guix pull` stalls on downloads and updating substitutes. I suspect the uni network since I don't have this issue at home. The `guix pull` stalling is particularly annoying as it doesn't fail and needs to be interrupted. Any pointers to what I could do? <xerty>worni: is it dns related? if no, you might want to consider downloading it over a proxy? <worni>I dont know, it's very random if it works and if it doesn't. I just keep reloading web pages and eventually they load. <worni>I also don't know how to download over a proxy <xerty>for guix-daemon it is ( sudo herd set-http-proxy guix-daemon http://ip(or address):poert <xerty>*for guix-daemon it is ( sudo herd set-http-proxy guix-daemon http://ip(or address):port <xerty>but I forgot whether if the guix pull will be proxied through <xerty>you can use a dns over tls or dns over https or doq to figure out if the campus networked stalled for you because of dns issues. <xerty>this one above is provided by cloudflare, but maybe you can use the one from wikimedia foundation <worni>sorry I don't see any message referencing a cloudflare dns <xerty>oh because you disconnected from that time range, don't worry I ll repost that <xerty>on librewolf (also packaged in guix) it is the same I think <xerty>this way you can also use the mirror of a git repo directly other wise you might need to compute the new config when declare it in scheme. <worni>Thanks I'll try this tomorrow or so need to abort it for today I think <worni>actually I couldn't because it wouldn't give me rest lol, but I think the issue is that the AP tells me to restrict txpower and I keep losing connection that way <nikolar>i am looking at harfbuzz and it's apparently been scheduled for building for like 51 days <attila_lendvai>nikolar, it's not excellent, AFAIK. but there also was some bug in the build scheduling, and some performance related regression fixed, so things may get better in the near future <nikolar>Well ffmpeg seems to have been stuck for 2 months without getting built <nikolar>Which is definitely something that should be looked into <attila_lendvai>ACTION will soon port guix to his new aarch64 BPI-R4 board, and will become a contributor <nikolar>i ask, because i wanted to make custom ffmpeg and mpv packages that could use hardware decoding on rk3399 (pinebook pro) <nikolar>but i am not building any rust code on that thing lol <attila_lendvai>i wonder... there are ways to customize packages with e.g. ./configure switches... is it possible to make sure that substitutes get built for such customizations? <attila_lendvai>ideally, it would be nice if someone could define a more refined arch target than the umbrella aarch64, and get substitutes for that <futurile>attila_lendvai: there's a cpu option for transform - probably where it could fit in (seems unlikely anyone's done anything on that yet) <dthompson>has there been any talk of switching to pipewire as the default thing rather than pulseaudio? the transition seems awkward since it's moving from a system service to a home service but I have been using pipewire for months and it's pretty solid. <fnat>Hi, I've got this containerised service (in a least-authority wrapper) that's exposed to the network. It works fine when connecting to it from the LAN but not via WireGuard. <fnat>The same daemon works fine (via the VPN) when containerised with 'guix shell --container --network ...'. <fnat>Anyone knows what the difference might be here between the two forms of containerisation? <fnat>Hm, they seem to be implemented in more or less the same way, so the problem is likely to be somewhere else... back to debugging. <jaft``>dthompson: only tangentially related to your actual question but do you know, any, why a home service was opted for with PipeWire while PulseAudio got a system service? <kreved>jaft``: probably because there were no home services when pulse service was written <nikolar>is there an advantage to switching to pipewire <dlowe>advantage: you no longer have to worry if you should switch to pipewire <dthompson>jaft: I've wondered the same thing! I don't know why. it certainly is inconvenient in that way. <jackhill>nikolar: qpwgraph is awesome :) But I guess also having Jack and Pulse connections going through the same server. And I already need to run Pipewire for screensharing, so may as well use it for audio too. <nikolar>i really don't like how wayland shells out to other things for core functionality <jackhill>nikolar: I am, but the first two points still stand. Presumably it brings some benefit not on wayland too, but only for fancier configurations that I don't use. <nikolar>well i am on pulse (not guix system though) and i haven't really felt a need to switch or try anything else <nikolar>but also i don't do anything fancy with audio <jackhill>Anyway in my mind I think pulse is the "build one to throw away" system, and pipewire is the one we aren't throwing away. :shrug: <nikolar>lol not sure what's that supposed to mean <dthompson>the new screensharing on wayland is definitely improved over X. the only thing that isn't good right now is the integration in guix. <dthompson>I had to write wrappers around some things to get pipewire to work <dthompson>pipewire unifying the pulse/jack rift has been a game changer for me <nikolar>i am not talking about whether it's better or not <nikolar>i am saying that i don't like that it's shelling out to pipewire <dthompson>but I'm no expert in how all this stuff is put together <nikolar>why does a protocol for handling all things video need to shell out to a third party tool, de facto making it a part of the wayland standard <gabber>Does answering "yes" to the (binary) installer's question "Customize users Bash shell prompt for Guix?" make guix alter all .bashrc (or .bash_profile) files found in home directories in /etc/passwd or the ones in /home/*/? <dlowe>Most distros have a shared profile or bashrc in /etc too <gabber>so it only alters the system-wide profile? <jfred>dthompson: I will say I'm glad there's a home service available for Pipewire since I'm using it on a foreign distro <jfred>which is not to say there shouldn't be a system service too <gabber>huh, it indeed alters each and every .bashrc found in /home/* and /root/ <dthompson>jfred: yeah I like it as a home service, too. it just complicates a transition from pulse, especially for those that don't use guix home. <jfred>(the split between home and system services always feels a bit strange to me) <weary-traveler>jfred: what's the experience like using guix home on a foreign distribution? any caveats that you're aware of? <jfred>weary-traveler: It's been surprisingly good all things considered, but be wary of conflicts between distro-installed services and Guix services. Using Pipewire from Guix for example required that I disable the existing Pulse/Pipewire services, because the distro really wanted to autostart those <jfred>...it did take a bit to get that all working right <nikolar>also, guix paths take precedence over your systems paths, so if you want your systems programs, you might need to edit that in some way <nikolar>also library search paths could be an issue, can't remember if i've ran across that particular problem <weary-traveler>right. i should probably start by re-considering what i want to have installed/managed via guix vs the host OS's package manager <jfred>It's also been made slightly more annoying by the fact that I'm running into what I think is a Shepherd bug, so it's been sometimes difficult to deal with in ways the distro native services aren't: https://issues.guix.gnu.org/72166 <jfred>...but that's not specific to using it on a foreign distro <jfred>I think at least some of the modify/test iterations while I was getting it set up for the first time were a result of that though <gabber>is git.savannah.gnu.org currently unreachable for anyone else? <abbe__>are there any guix git mirrors ? <abbe__>yeah, it's inactive for more than half-a-year <gabber>this is probably still far more recent than a fresh installation (1.4.0) <abbe__>well, i was looking for something to replace in channels.scm to unstall things while FSF infrastructure comes back online. I guess I'll just use my own repository copy :) <graywolf>However I cannot recommend having your own tree strongl enough