<jonsger>raghav-gururajan: would welcome the guix edit patch. As I usually dont have emacs installed... <mbakke>apteryx: I'm a bit confused about the parallel development of linphone on core-updates and master, do you think you can take care of merging master into core-updates? <Parra>where can I find the code related to cross-compile with mingw? <raghav-gururajan>mbakke Only linphoneqt is missing in master. All other packages related to linphone, including liblinphone are available in master. :-) <raghav-gururajan>jonsger Are there any instances in guix, other than `guix edit`, where a default editor value to be set? <dissoc3>is the guix emacs package broken when trying to install packages? <dissoc3>i want to make a package for apache cassandra. im pretty new to writing packages. is there anything anyone can think of that would be similar that I can refer to ? <plstohelp>guix pull: error: error parsing derivation `/gnu/store/6m9iiwb8xj9qgjvw4zb235srbj0m8nqd-compute-guix-derivation.drv': expected string `Derive([' <plstohelp>can't delete it, it says "read only filesystem <lfam>plstohelp: Are you using `guix system vm`? <ryanprior>In a package build, how do you get the path of installed files? ***jonsger1 is now known as jonsger
<ryanprior>(symlink (assoc-ref %build-outputs "io.elementary.sideload") (assoc-ref %build-outputs "sideload")) <ryanprior>No idea if assoc-ref is what I want or not I've just been looking at other packages for clues and saw that. <sneek>I last saw ryanprior in #guix 17 seconds ago, saying: hi dsmith-work =D. <sneek>RhodiumToad was in (here?) 1 hour and 13 minutes ago, saying: "flexible" presumably in the sense of not conforming to some fixed predefined schema.. <sneek>RhodiumToad was in #guile 54 seconds ago, saying: moo. ***wxie1 is now known as wxie
<dsmith-work>It no longer mistakenly responds to "seen". Only the bot's actual nick. <dsmith-work>The last seen info now includes the channel and a text time, like "5 years ago". <lfam>ryanprior: Likely, you need to build a path with something like (string-append (assoc-ref %build-outputs "out") "/bin/sideload") <lfam>Once in a while a package will define multiple outputs, and each output creates its own store item. <ryanprior>Oh I don't know that %build-outputs is even a thing. I see a package that uses (assoc-ref %build-inputs "libmediainfo") and I was like hmm that seems kinda like what I want. <ryanprior>How do you use the "out" output normally? In terms of getting paths to files installed there? <lfam>ryanprior: Just do the default thing, like `make install` <lfam>Check the fastqc package definition for an example of symlinks <lfam>Especially the 'install' phase <lfam>It starts by defining some variables like out, bin, share, and then it uses them <lfam>You're right, there is no %build-outputs. It's just %outputs <anadon>Would someone be able to guide me through dumb stuff I'm getting wrong in packaging tonight? <ryanprior>Holy buckets I got it right on the first try after reading the fastqc package. Perfect recommendation thank you. <ryanprior>I don't actually know what lambda* does in guile but I am going to look that up next. <ryanprior>no lmao my guile is actually terrible. I know a decent amount of racket, clojure, and elisp, so my default when writing guile is "write racket and go to the docs when something doesn't work" X.X <ryanprior>That link indeed looks helpful and I'm bookmarking it <lfam>raghavgururajan: The command line tools are in 'guix/scripts' <Gooberpatrol66>is there some way to get line numbers for errors in guix pull log files? <ryanprior>I'm working on packaging an app that requires libflatpak. guix already packages flatpak- is there a way to require it as a library dependency, or do I have to package it that way specifically? <Blackbeard>ryanprior: do you try as an input? What application is it? <raghavgururajan>bandali Was wondering i you received my email regarding thunderbird? <ryanprior>during phase configure: "Dependency flatpak found: NO (tried cmake)" <bandali>raghavgururajan, i did, i just haven't had a chance to send a reply yet :-) <lfam>pkill9: Can you reproduce it locally? There may be more detailed logs in the build directory <bandali>raghavgururajan, cheers :-) i'll get to it by the end of this week <raghavgururajan>bandali If you are available, would you be able to review and push #40708? <pkill9>lfam: it fails but i haven't checked it's for the same reason, but it probably is <bandali>raghavgururajan, looks like lfam just sent a reply on it :-) <ryanprior>Blackbeard another interesting thing about this (or maybe not interesting & just surprising to me) is that sideload uses the meson build system, but it's cmake complaining about not seeing the dependency. <ryanprior>It originally complained about not having cmake, so I added that as a native-input, and then cmake complains about not being able to see flatpak. <ryanprior>I'm trying to build an application (Sideload) that uses libflatpak, but despite using the meson build system, it's failing with a message about cmake can't find the flatpak dependency. I added cmake and flatpak to the native-inputs but something is misconfigured & I'm at a dead end right now looking for clues. <raghavgururajan>ryanprior (arguments `(#:configure-flags (list (string-append "OPTION_FOR_FLATPAK=" (string-append (assoc-ref %buid-inputs "libflatpak") "/include"))))) <raghavgururajan>ryanprior What does error message say? Could you share that as well? <ryanprior>Maybe I did your flag idea wrong, let me know what you think <raghavgururajan>ryanprior (arguments `(#:configure-flags (list (string-append "-Dflatpak=" (assoc-ref %buid-inputs "libflatpak"))))) <ryanprior>No apparent improvement, I updated the gist again with the new def & corresponding log <raghavgururajan>ryanprior "ERROR: Could not generate cargs for flatpak:" is different. <ryanprior>And it still says "Dependency flatpak found: NO (tried cmake)" <raghavgururajan>ryanprior In the package definition of flatpak, some dependency might not be propagated. And you package is looking for flatpak with it's deps. Since they are not propagated, it mistakenly tells you it could not find flatpak. But "could generate cargs", I have dealt with gnome packages. <raghavgururajan>ryanprior As a start, copy all the inputs of flatpak to the definition of your sideload package. Check wether error stops. If so, then try to remove copied inputs one by one and rebuild. Once you hit the error again, then that's the input you need. Keep that and remove other copied inputs. <ryanprior>Aaaand a little troubleshooting later, we've got a sideload package. Thanks guix!! & especially raghavgururajan, Blackbeard <ryanprior>Unclear whether it /actually works/, but it builds and runs. <ryanprior>I started by adding all of them,I can remove a bunch and test. Is there a way to automatically discover unneeded dependencies or do I have to do it by hand/hunch? <ryanprior>Now that I've got the package built, it runs, but it hits a snag when I try to actually install a flatpak by its ref. <raghavgururajan>ryanprior Probably trial and error. But my best is libostree is what is required. <ryanprior>I do have that as a propagated input already. <raghavgururajan>ryanprior Oh wait. You might actually need to add "#:glib-or-gtk? #t" as argument. <elais>does the clojure package install the cli tools as well? <elais>any thoughts on how to get clojure-tools-deps packaged? Are there some obvious problems installing clojure with the cli? <ryanprior>Woo I'm excited, this was my first attempt at a slightly hairier package, and it laid the groundwork for some other ones I'm interested in <ryanprior>elais I remember there being some talk about the difficulty of building a bootstrappable CLI, because it requires the CLI to build the CLI, or something like that <xelxebar>ryanprior: Nice work bootstrapping your packaging skillz. <xelxebar>Masochistically, I chose a tool which needs an associated service as my first attempt at packaging. <ryanprior>xelxebar it's been daunting at times! The last time I tried building a few packages, I bounced off pretty hard after working on it almost every day for a couple weeks. I read a ton of docs, did a ton of troubleshooting, and didn't get things building. <ryanprior>I picked it up again this last weekend and have had much more success. Looking back at my old package files I can see I was really close. <xelxebar>Holy moly. You must be working on highly non-trivial ones. What exactly did you get working this time? <ryanprior>I started with packages I hoped would be really trivial. I looked for things with very few dependencies, all of which were already in guix, so I would have a straight shot. <ryanprior>One thing that gave me much better results this week than before: instead of reading docs and Guix source code, I'm now reading other Guix packages, including grepping in the `guix/gnu/packages` folder for bits of text I think might be helpful. <xelxebar>Oh, yeah. Pilfering ideas from other working examples is pure gold. <ryanprior>Trying to piece together guix from the documentation just didn't work for me, I put about 30 hours into that and learned a lot of interesting things but did not build any packages. <xelxebar>Do you have packaging experience for any other distros? <ryanprior>Yeah I've done ruby gems, debian packaging, and docker. <ryanprior>I've also done a lot of Ansible and Chef work which is similar to packaging. <xelxebar>Hrm. I recall a similar Struggle TM when packaging something for the very first time. It happened to be on Void Linux, but I found the conceptual framework to translate relatively nicely. <xelxebar>Sounds like you've got a pretty broad range of experience. <rekado>ryanprior: Guix has a bunch of importers that generate code from upstream repositories <elais> Yeah I've done terraform which sort of prepared me to package stuff <ryanprior>I remember struggling for about a week when learning Docker, and I felt like the curve was steeper for guix than for docker. There's still some things I'm not clear how to do, but I feel like I'm making progress every day now. <rekado>for R packages, for example, that are on CRAN or Bioconductor you can do āguix import cran -r the-nameā and itāll spit out code that is almost ready to be pasted into gnu/packages/cran.scm <elais>though I'e only packaged midori and cmt <ryanprior>Ooh Midori is a cool package, thanks for doing that elais. <elais>I didn't package the one in the repo <ryanprior>rekado I have seen the docs for that but haven't tried it, I will definitely give it a shot soon. <elais>I packaged one a long time ago and let it rot in my repo because I couldn't get video working <elais>turns out that wasn't my fault, I think for a while webkitgtk was on the strugglebus <elais>I really want to get clojure working <elais>if I can bootstrap that I can use it as a bargaining chip for work <ryanprior>elais I would be thrilled if you do, and would be happy to contribute to the effort :) <elais>well, I wonder how hard it will be to bootstrap boot? <elais>Someone has already packaged lein <elais>I don't think lein is nonfree so its ok to talk about it <ryanprior>Oh is it in nonguix because it isn't bootstrappable? <ryanprior>lein is Eclipse license, same as most of Clojure <rekado>would be nice if the nonguix people separated free and nonfree stuff. <rekado>ideally the good stuff would make it to Guix proper <elais>I have an aborted attempt at libinput-gestures <ryanprior>bad could be for free malware samples and rats and stuff, useful to researchers but not stuff you'd want to install outside of your research vm <ryanprior>I'm in security so I install a decent amount of "bad" software X.X <elais>I had designs on turning it into a service but then I rage uninstalled Guix System. I'm still working on guixy things but I can't with system right now <rekado>mbakke: yesterday I ran a very long āguix gcā job to free 2T of data. This did free up 2T of data but in the meantime the used space had grown by 1T. Now weāre back at 3.8T free space, 1T less than we had just yesterday. <rekado>I see that there are quite a few mcron processes, all running at the same time. *raghavgururajan is working on packaging GNU Gnash, a good old swf player. It would be useful to play legacy flash files. <ryanprior>raghavgururajan fyi Gnash is years out of date and never supported modern versions of SWF. <ryanprior>Ruffle is a free software player that's funded by Newgrounds and others. It also cannot run modern versions of SWF, though. We'll see if they have more endurance than previous efforts. https://github.com/ruffle-rs/ruffle <raghavgururajan>ryanprior I am aware. There is no flash development anymore. Some good documentries, learning materials etc are only found in swf formats. This app would be good. <ryanprior>Oh good, yeah I agree there's a ton of web history in swf. <raghavgururajan>Yeah, primarily educational stuffs like animations for physics, chemistry and biology. <ryanprior>Ken Perlin's website has a ton of classic experiments in design and animation, but they all depend on Java Web Start which you can't even use with modern browsers anymore. <raghavgururajan>I learnt DNA replication soley by animation. Text book images was not helpful. <raghavgururajan>I am glad that atleast CRISPR-Cas9 (tool behind designer-babies) are now in webm format. <rekado>ryanprior: commit c47f48d972 adds visidata <raghavgururajan>One of my gene-therapy research goals involve extensive use of CRISPR-Cas9. <rekado>I made only a handful of changes: gnu/local.mk should use tabs, fixed the indentation of visidata.scm, ran guix lint over it, and adjusted the patch to gnu/local.mk which no longer applied <rekado>raghavgururajan: Iāve been working with Cas13 *rekado cleans up big stuff by explicitly deleting files mentioned at /gnu/big-stuff <ryanprior>rekado thank you for looking at visidata, I'm happy to be an official contributor now! I will have a few more packages for consideration within the next week. :) <rekado>raghavgururajan: we synthesized a bunch of Cas13 variants. The project is super secret as yet unpublished research :) <rekado>ryanprior: congrats on your first patch to Guix! <ryanprior>I used to work in solid state protein synthesis, fun stuff. <raghavgururajan>rekado That's cool! I didn't know you were working in wet lab stuffs. <rekado>I donāt, but Iām involved in a very small number of wet lab projects. <raghavgururajan>Ah I see. Do you perform the lab work or you are taking care bioinformatic stuff? <rekado>in this case neither; I designed and built some equipment thatās central for the experiments <rekado>hmm, issues.guix.gnu.org displays some emails twice. <rekado>I guess the debbugs log file must be listing them twice for obscure reasons <raghavgururajan>Ah that's great. I would like to discuss more later. I like this stuff, bio+comp. <raghavgururajan>My lab days are over. As I am changing my career to IT. But will integrate IT with BioInfo, to work on gene therapy stuffs. <raghavgururajan>sneek, later tell civodul: When you are available, let me know if everything okay with #40785. Thanks! <rekado>this has been reported before; not sure whatās going on there <rekado>would be good if someone could investigate where things go wrong <rekado>well, thatās the question: are the links wrong or has the manual not been generated fully? <rekado>I suppose the answer will be in maintenance.git <raghavgururajan>I just tried tinkering with URL.. Doesn't changing anything. So it leads html file not being generated. <rekado>mbakke: something must be keeping disk-image and qemu-image items alive; they cannot be removed <rekado>I see that, for example, /gnu/store/kcrvglnaqfdfk46v3fx42yykirsp6x1m-run-vm.sh has two links <rekado>but I canāt seem to find what else links to it. <rekado>hah, I think itās all those roots for ā*-testā *rekado reconfigures ci.guix.gnu.org and restarts mcron <xelxebar>rekado: What do you do that lets you contribute so much to guix? <xelxebar>"do" meaning in terms how you make a living ***plstohelp_ is now known as plstohelp
<plstohelp>and now pulling is broken, with the error message i posted earlier today <rekado>xelxebar: I have convinced the right people that Guix is work. <rekado>yes, but the tasks are unlikely to involve work on Guix <rekado>we do offer internships, but they are unpaid AFAIK ***langdon_ is now known as langdon
<xelxebar>Hrm. I need a full position. Just exited a Ph.D. since the future prospects were looking bleak. Trying to get back into industry. <nouqfon>Anyone know if it would be possible to (with some work) run Guix on the PinePhone? ***kelsoo1 is now known as spitball
<elais>maybe when we get it working on the pinebook? <xelxebar>nouqfon: Probably would require a bit of nongnu stuff, at least for the modem blobs <nouqfon>I see. I would be okay with that for personal use, but I understand that cannot be included in the repos. I'm wondering more about the amount of work involved to get it up and running. I've got a PinePhone on the way and a lot of free time during the summer, so I thought it could be an interesting little project. <liltechdude>Hello! While I am try to create package from small python package without dependeces I catch bag, that without explicitly add this library to PYTHONPATH <liltechdude>Hello! While I am try to create package from small python package without dependeces I catch bag, that without explicitly add this library to PYTHONPATH I cannot use this library from python https://paste.debian.net/1142364/ this paste contain package definition and shell output. Sorry for my poor english. <xelxebar>nouqfon: I can't speak from direct experience, but my guess would be that the work boils mostly down to packaging up the necessary userland utils. <xelxebar>Not sure if linux-libre would run on it. Apparently it typically runs the mainline kernel. <nouqfon>I would be okay with compiling the kernel from scratch. Mostly worried about device drivers and such tbh, and also the linux mobile applications e.g. Phosh. How involved do you think such things would be? <xelxebar>I interpret that to mean that drivers likely wouldn't be an issue. <nouqfon>Indeed. In fact I think I saw that Guix supports the SoC already (A64 was mentioned when reading the running Guix on ARM blogs) <nouqfon>But it seems every ARM device has a unique bootloader or something of the sort <nouqfon>My experience with ARM devices is limited to Raspberry Pis. But would love to be able to run Guix on my phone. <nouqfon>I also saw that NixOS is mentioned as supported. From what I understand it shares many similarities with Guix? <xelxebar>I've never *packaged* a python thing before... <xelxebar>nouqfon: Yeah, that's super encouraging, especially since guix has a nix importer. <nouqfon>Oh that's neat. I guess I will try this when I receive my device. Exciting times! <bricewge>nouqfon: It will likely work on Guix. The wiki page state that PureOS is supported, which is also a FSDG distribution so there shouldn't been any blob <nouqfon>Excellent. Then it will mostly be a matter of packaging the phone shell etc.? <rekado>nouqfon: I have seen recent work on building a kernel for the pinebook <rekado>vagrantc and efraim probably know more about this <xelxebar>bricewge: Yeah, the PureOS thing confused me. I mean, you need a modem blob unless you don't care about using the phone part of the phone. <nouqfon>I read on some posts at the Pine64 subreddit (I think it was) that the PinePhone needs drivers not included in the kernel (yet?), so it would need a custom kernel. But like I said earlier I would be fine building the kernel from scratch. <liltechdude>xelxebar: i dont think that it help, because package works fine. Only I need is add symblink to default PYTHONPATH *rekado looks into adding indexed search to the log viewer <xelxebar>liltechdude: Grepping for PYTHONPATH in gnu/packages/ turns up lots of results. Not sure if any of those are helpful? <xelxebar>Is there a convention on where to place new package definitions inside existing gnu/package/foo.scm files? <xelxebar>They look non-alphebetically organized, and a cursory glance at git blame seems to indicate that new packages are tacked on the end... <bricewge>So no wifi, bt or autofocus on Guix, the rest should run fine with free software only. <xelxebar>bricewge: "... the Quectel EG25-G LTE modem runs its own closed-source OS" <xelxebar>Is that on a rom chip or something which can be floss-ily initialized? <bricewge>No you are reading it right, since it's firmware don't need to be loaded ab boot and it interface trough USB there is no need to add blob here in Guix <nouqfon`>Thanks for this source. Sorry if this is taboo to ask here, but would it be possible to load the firmware for Wifi/BT for personal use? <xelxebar>Okay. Thanks for spelling it out for me. <rekado>nouqfon`: linux-libre does not support loading firmware blobs for deblobbed drivers. Itās a missing feature. <rekado>while we donāt include non-free software and wonāt encourage its use linux-libre should allow users to manually load blobs if they wish. <rekado>but this is a long-standing missing feature in linux-libre <rekado>I donāt know if upstream is working to address this. I donāt think they are. <lprndn>bricewge: Don't know if you've seen it, I sent patches for the new version of lightdm's service :D <cbaines>nouqfon`, I've never used OpenRC, but that usage of start-stop-daemon doesn't look great <lprndn>by the way, if anyone is daring enough to review or test it, help yourself! ;) (bait: it starts wayland sessions too) <nouqfon`>I agree. I am not really experienced with writing scripts for OpenRC, so I copied the one from the Guix overlay. <cbaines>nouqfon`, have you got an example of a better written OpenRC init file? <nouqfon`>Hmm, which application would you prefer? <bricewge>lprndn: I struggle a bit with debbugs, I tought that responding to a bug I would get all the next update from it but it doesn't looks like it's the case. <nouqfon`>I can paste any usual system application <cbaines>nouqfon`, I guess something similar, a simple daemon would be good <lprndn>bricewge: Might be my bad, I fiddled a little with my mail's header... <bricewge>I don't think so, i had the same issue with #40480 <nouqfon`>Thanks, I will have a look. Perhaps I will ask the Gentoo IRC for further help with OpenRC. I had kind of assumed it was something unique to the Guix daemon, but I guess I just have a poorly written config file. <cbaines>nouqfon`, I'd hope that if you can get start-stop-daemon to use a pidfile, then it'll work fine <nouqfon`>Could you give me an ELI5 of how pidfiles work? So we can find the pid with the script later on? <cbaines>nouqfon`, Sure, start-stop-daemon starts the guix-daemon, then writes the PID down in a file. When it comes time to stop the guix-daemon, start-stop-daemon reads the PID from the file, and tells that process to stop. <nouqfon`>Ahh I see. Thanks. I will read the instructions you provided and give it a try! <rekado>cbaines: I canāt get a shell on bayfront (ssh authentication works fine). Can you still connect? <cbaines>rekado, I'll try, it hasn't connected yet. <cbaines>I'll stop cuirass... hopefully that'll reduce the load <cbaines>Stopping cuirass took ages, but I think it's stopped now, and the load is starting to come down <cbaines>rekado, you should be able to connect without any trouble now <nouqfon`>cbaines: A question about the pidfile for the Guix daemon. In which path should such a file be put? <efraim>nouqfon`: do you have a /var/run? on Guix System I see a couple of PIDs there <cbaines>nouqfon`, I don't know where they'd usually be in Gentoo, but maybe /var/run/ <nouqfon`>Ah actually it is /run and /var/run is a symlink to this location. <nouqfon`>Can I run the Guix daemon as a non-root user? <efraim>maybe with some system modifications, it needs to be able to create chroots and use namespaces, but in general you want to run the daemon as root. <nouqfon`>Ah I see. I will continue to do so then. Just curious. <nouqfon`>cbaines: Just to let you know, stopping the Guix daemon works perfectly while using a pidfile. Thanks a lot! <nouqfon`>When updating Guix on a foreign distro, is it sufficient to run 'sudo -i guix pull', or do I need to run 'sudo -i guix package -u' as well? <cbaines>nouqfon`, assuming you're using root's guix for the daemon, that'll be updated by sudo -i guix pull <cbaines>you'll need to restart the daemon to actually update it though <nouqfon`>Thanks. I was confused as the output of guix pull seemed to indicate I also needed to update packages after the fact. <cbaines>Well, if you've installed any packages as root, then that would be true <nouqfon`>I only have the glibc locales installed for the root user. <cbaines>nouqfon`, oh, and going back to OpenRC, the Guix Git repository includes Upstart, Systemd and sysvinit scripts, but you could send the OpenRC config as a patch if you'd like <cbaines>also, the installer script doesn't have OpenRC support yet <nouqfon`>I'm afraid I don't know how to write shell scripts <xelxebar_>Why are the po/**.po files generated by make not ignored? <nouqfon`>Thanks. Actually I managed to get a working OpenRC init script for guix-daemon, but I wouldn't know how to patch the Guix installer to include support for it. <asmodehn>Hi everyone, I was wondering if there is a guide somewhere to build a guix image (vm, docker, etc.), but **only from guix packages** that can be installed on any OS/platform... <nouqfon`>Hmm. I see. Seems pretty simple. I will see if I get the time in a few weeks. Currently in the middle of exams and all of that good stuff... <asmodehn>to be more precise, I have guix (packages) on an ubuntu system. I want to build my guix config (to build qemu vms), but guix system vm-image fails, and I have no idea why... <nouqfon`>thvk-ivgf if you are in need of the OpenRC init script I can share it with you? <asmodehn>thvk-ivgf: nope... I m investigating and I think it has to do with the fact that I dont know how to use qemu... the log says : qemu-system-x86_64: failed to initialize KVM: Permission denied <thvk-ivgf>asmodehn: "To use guix system vm and related commands, you may need to add the _build users_ to the kvm group so they can access /dev/kvm," <sneek>Welcome back civodul, you have 1 message! <sneek>civodul, raghavgururajan says: When you are available, let me know if everything okay with #40785. Thanks! <bricewge>Is it okay to go over the line lenght limit whem dealing with long strings (54 character path here) to keep the code readable <bricewge>As is gnu/services/sddm.scm for example. <civodul>if it's not too long it's ok, but sometimes it's IMO clearer to split the string <civodul>well, dunno, it's very subjective :-) <civodul>"too long" as in "way beyond 80 chars" i guess :-) <bricewge>Okay, I'll go with the >90 character from report-long-line then <TZander>does guix do bounties? I'd support anyone fixing the slow downloads. Looks like it closes the connection after each sustitute if tinished downloading. <TZander>I'd pledge in Bitcoin Cash, if that helps. <nagamalli>Hi,When i applied build command for updation of package , It was stopped at this point ...... waiting for locks or build slots... <nagamalli>I thought it was internet issues, when i try to apply again it's always stucking on there only. <nagamalli>substitute: /gnu/store/q19l04vd2za80mk1845pz7r8cz29qk43-bash-minimal-4.4.23/bin/bash: warning: setlocale: LC_ALL: cannot change locale (en_US.utf8) <nagamalli> /gnu/store/3p24vc7v6cnd50a722kh910mnn3zlqsw-git-minimal-2.26.2 <rekado>nagamalli: this means that some builds are already ongoing, including some that are preconditions for your current build <bdju>deluge on nixos updated to the new major version, maybe that'll make it easier for someone to update the guix package <rekado>bdju: is deluge particularly difficult to upgrade? <bdju>it was out of date on both distros for a really long time, not sure if it's especially difficult <bdju>the new daemon was incompatible with the old client, though, so I've been stuck using deluge-console for ages, even started switching over some things to transmission <rekado>TZander: why did you delete the part that disables widevine? <anadon>I think the go-1.4 dependancy gcc:lib needs to actually be gcc-toolchain, and to exist as a propagated input. I'm running `go build...` on a project, and the failure related to glibc missing, it using gcc-7.2, and some other weirdness. Thoughts? <bricewge>bdju: After that I'll try making a service <rekado>TZander: you also removed a few lines that patch qtbase, but you didnāt remove the patches themselves. Please also mention changes like that in the commit message. <rekado>anadon: sounds like the correct thing to do to fix the error. Could you post to guix-devel to discuss the proposed change? <rekado>TZander: since a lot of packages would be rebuilt when changing Qt, please make sure to rebase your patches on top of the core-updates branch. <TZander>rekado: re widevine, I don't recall doing that, I'll have to check. May have been a mistake. <jonsger>crazy. YouTube is delivering some videos already as AV1 even on my powerpc :P <anadon>AV1 is a pretty good standard from what I understand. <jonsger>some weeks ago they still used only VP9 for my browsers <jonsger>now only all the other video deliveries apart from YouTube and Netflix need to switch from H26x to AV1... <buffet>hey guix, is there an easy way to check what module a package is in? atm i do `guix edit <PKG>` and then go to the top of the file, and i feel like that could be improved <cbaines>buffet, guix show will tell you the location <buffet>i just realized that guix search at least shows the file its defined in (which seems to match the module name) <thvk-ivgf>buffet: there is `recsel` from recutils (`guix show xyz| recsel -p location`) <buffet>thvk-ivgf, thats more like it, thanks a lot <civodul>anyone happens to know if we have marginnote.sty (texlive) in a package? <cbaines>I see /gnu/store/iqiy2p84yxd1sjchdqd7p5kdr0xkwdhg-texlive-texmf-20180414/share/texmf-dist/tex/latex/marginnote/marginnote.sty <civodul>ah, but that's the big texlive package <civodul>i was hoping to use the modular texlive <cbaines>that was the only occurance on my machine, but I don't know what modular texlive stuff I have installed <buffet>oh another question, i need lua for a package, `guix show lua` gives me 3 versions, and just lua seems to default to 5.3 (ie the newest). how can i select a different version? <rekado>our tex tooling really needs an overhaul <buffet>rekado, doesnt work for the inputs field inside a package definition <buffet>sorry thats missing in my question <pkill9>buffet: use the variable for that version instead, typically 'lua-5.1' <pkill9>typically 'package-version', in this case lua-5.1* <rekado>civodul: texlive-marginnote now exists <NieDzejkob>so, why is there a nix/libutil/.gitignore that ignores all *.cc and *.hh files? They are included in the repo... <anadon>Maybe they're autogenerated? That does sound strainge NieDzejkob. <NieDzejkob>anadon: I don't think so, they're there in a fresh checkout of the repo <efraim>hour 15 minutes for the test suite for python3 on powerpc <g_bor[m]>We had no time to relax our slot reqirement for guix in GSoC, but as of previous discussion we agreed that one should be essential. <g_bor[m]>So we finally got one slot assigned, which is in alignment with what we wanted. <jonsger>hm guile-fibers seems to be failing on master and core-updates for quite a while and its a dependency of cuirass <civodul>jonsger: there are test failures when run on a machine with "too many" cores <civodul>it can be worked around but it's annoying <jonsger>building it now on my server with 6 threads. Let se <jonsger>civodul: should I report a bug upstream or is this something wrong at Guix? <civodul>jonsger: i think rekado (?) already reported it upstream but you could check <civodul>upstream might even be reading this :-) <jonsger>I dont think he filed a bug at upstream github repo <mfg>i'm trying to use the texlive-latex-koma-script package, but it seems that kpathsea doesn't find it. When i read the apckage definitino corrrectly it also doesn't run anything special just copies the files into the store. How does kpathsea find latex files on guix? <rekado>mfg: I donāt know if kpathsea will ever find it. <rekado>what we do is run a profile hook that generates a TeX Live union and a configuration file + env vars <rekado>I donāt know if kpathsea ever does the right thing with just that info <rekado>but the latex, pdflatex, lualatex executables do find it. <mfg>Hm okay, i will ignore it then <mfg>thing is lualatex doesn't that's why i tried kpathsea <rekado>note that the texlive-{latex,generic}-* packages are all old style and may actually be incomplete <rekado>(fixing this is a lot of work and I got tired of it because the feedback one gets when working on texlive is generally discouraging) <mfg>texlive-tiny, that should depend on texlive-bin <rekado>texlive-tiny is really meant as an input to packages <mfg>Oky, i thought it was more like a texlive-minimal :D <rekado>the comment says ā;; For use in package definitions onlyā <rekado>the comment above texlive-base says ā;; For user profilesā <rekado>if that makes a difference we should do something about that and perhaps hide texlive-tiny from the UI <rekado>(obviously users shouldnāt need to look up comments in the source code to figure out what to do) <mfg>i didn't read near texlive-tiny :| <mfg>i will try texlive-base <rekado>itās our collective failure to address this problem, because it works well enough to build packages, and well enough for the use cases of those who care <mfg>i see that might also be the reason why it doesn't get updates that often :D <rekado>would be great if we could offer a paid internship to someone willing to work on this <rekado>the reason for a lack of updates is also because texlive packages are inputs to so many other packages <rekado>so updates generally would need to happen on core-updates <rekado>I need to go now; hope texlive-base fixes things for you. If it doesnāt please send a message to guix-devel or help-guix to kick off a discussion on how to improve this. <mfg>using texlive-base works btw <pkill9>i would like a claw-mail plugin that integrates the guix issues tracker <pkill9>or in general, any email archive service <janneke>hmm, i ran the guile webserver test on the Hurd "just for fun" and it passes...yay... <janneke>so now i'm wondering...i'm pretty sure it fails when doing a native guile build <janneke>and i'm also starting to suspect that there's a high probability it will pass when run in an aborted build <janneke>but that's all pretty time-expensive...so i'm wondering if i could think of a some hack that would give me more than one shot at running that test <civodul>janneke: perhaps it's timing-sensitive or something? <rekado>bah, the link deletion stage is much too slow. <rekado>it takes half a day already, and in the meantime more garbage has piled up than has been deleted in the hours before. <civodul>rekado: half a day? i thought you measured 67mn or so? <rekado>I started this after lunch; itās 7pm now. <rekado>I started a targeted deletion then <rekado>yes, with just a few dozen disk images as the arguments <civodul>and the list of images was already computed, right? <rekado>itās a tiny part of /gnu/big-stuff, the file that is generated by the cuirass roots cleanup cron job. <emiller88>Anyone know where the libguile header is packaged? <emiller88>guix environment --ad-hoc --container gcc-toolchain guile gnuplot make is what I'm using to make my env. <rekado>a couple of hours ago it told me it had collected 33970 MiB. And since then Iām waiting for ādeleting unused linksā <civodul>emiller88: you need to use 'pkg-config guile-3.0 --cflags' <rekado>when I started it we still had 4.x TB free space; now itās 3.4T <efraim>it also deletes other unused links which were previously marked for deletion <efraim>so in theory you could stop it now and run 'guix gc -C 1' later and continue on after it calculates freeing 1 byte <rekado>I aborted it and restarted it with the next chunk <civodul>reminds me of terrible old stories, rekado: on hydra, Mark had a C program that would get the entries of /gnu/store/.links, which he would pipe to sort, and then "xargs rm" or sth <rekado>I donāt want to become to ci.guix.gnu.org what Mark was to hydra. <emiller88>civodul: so guix environment gcc-toolchain ... pkg-config guile-3.0 --cflags ? <civodul>emiller88: the tutorial is outdated, but basically you'd need to use pkg-config in your makefile <civodul>perhaps #guile is a better place for this discussion actually :-) <civodul>one is to estimate what it would cost us to disable deduplication on berlin <civodul>we'd consume space more quickly, but GC's would be faster <civodul>in addition to that, we can try to better estimate what's taking time, like the link-traversal.c thing that you ran <civodul>see if there are ways to optimize this <civodul>we could also think of a different implementation using a separate database <nagamalli>when I tried to buld gnome-todo , build failed bacause of some builds already ongoing , including some as rekado said. <nagamalli>Now My question is I need to update the version of glib I think , Where i need to change , inorder to fix that issue <nagamalli>I thought I need to change in gnome.scm file of gnome-todo package,but i didn't understand where i need to change inorder to fix it <nagamalli>If i write mulitple package descriptions. the fastest way to make patches is adding ONE description, building, apply lint, commit patch and go for next one . Am i correct? Please suggest <nagamalli>I have a doubt. If i write mulitple package descriptions. the fastest way to make patches is adding ONE description, building, apply lint, commit patch and go for next one . Am i correct? Please suggest <TZander>What is fastests depends on how good you are with git and with not getting lost with lots of things happening at the same time ;) <nagamalli>i have etc/indent-code.el.in but not etc/indent-code.el <nagamalli>but running it is not working either. Please suggest about this? <mroh>try `make etc/indent-code.el` <nagamalli>I tried this emacs --script ./indent-code.el.in ../gnu/packages/cran.scm , It worked for me ***spitball is now known as kelsoo1
<rekado>nagamalli: you should run bootstrap and configure first <nly>which package includes the tools alsa_in and alsa_out? <ryanprior>What license do you use in packages for code that's Apache-2.0? <TZander>guix is a gnu program, it would be nice if it used gnu-like argument parsing. The tool is annoying... <TZander>Why are these not legal? `guix package -l --profile=a` and `guix package --profile a -l` The only legal one is to put the profile first and to use an equals between it and the argument <civodul>TZander: it's not about legality :-) <civodul>the command-line parser differs from libc's argp <civodul>but not all GNU tools use the same parser <civodul>i agree it'd be nice to allow for "--profile a" in the example above <nagamalli>when I tried to buld gnome-todo , build failed bacause of some builds already ongoing , including some as rekado said. <nagamalli>Now My question is I need to update the version of glib I think , Where i need to change , inorder to fix that issue <nagamalli>I thought I need to change in gnome.scm file of gnome-todo package,but i didn't understand where i need to change inorder to fix it <nagamalli>I have tried all methods but not getting how to solve .please suggest <raghavgururajan>How do I rename "configure.in" to "configure.ac", inside package definition? <TZander>civodul: order is also very confusing. You can call `guix -d 10 -S 9' and no `guix -S 9 -d 10` (or vice versa) because the tool rejects deleting the current. Even though I told it to change in the same command. <nagamalli>gnome-todo 3.28.1 fails to build with new glib2.0 >= 2.60.0 <raghavgururajan>I get "warning: autoconf input should be named 'configure.ac', not 'configure.in'" and ending up with "./autogen.sh: ./configure: /bin/sh: bad interpreter: No such file or directory". <TZander>civodul: what I mean is: command-line-parser should not set the order in which that arguments are processed. Gives a bad UX. It should ALWAYS first switch, in this instance. <civodul>TZander: could you frame it in terms of either a bug report or a proposal? <TZander>I propose that someone that can actually code scheme starts caring about the user interface of the cli tool <civodul>please, that's not the tone we'd like to see here <civodul>that's why i was asking for a concrete proposal <nagamalli>raghavgururajan:Actually As rekado said,not to update gnome todo..before i tried to update that and build was failed.. <nagamalli>When i applied build command for updation of package , It was stopped at this point ...... waiting for locks or build slots... <nagamalli>I thought it was internet issues, when i try to apply again it's always stucking on there only. <nagamalli>substitute: /gnu/store/q19l04vd2za80mk1845pz7r8cz29qk43-bash-minimal-4.4.23/bin/bash: warning: setlocale: LC_ALL: cannot change locale (en_US.utf8) <civodul>i think processing options in order has value in some cases <nagamalli> /gnu/store/3p24vc7v6cnd50a722kh910mnn3zlqsw-git-minimal-2.26.2 <civodul>the one one you gave is such an example: the user wants to know whether switching generations happens before or after deleting <civodul>so i'm not sure what the proposal is <civodul>nagamalli: please don't paste many lines at once :-) <civodul>nagamalli: the "waiting for locks" messages means there's another "guix" command running and building the same thing <nagamalli>civodul ,Yeah I understood from rekado's reply earlier , <TZander>nagamalli: try restarting your guixbuilder service. <nagamalli>where i need to make changes was my question! <raghavgururajan>civodul Do you know how to rename "configure.in" to "configure.ac" inside package definition? <civodul>raghavgururajan: you could add a phase that does (rename-file "configure.in" "configure.ac") <nagamalli>TZander, Is it the problem? I don't think it is ...THe problem is about to fix the glib version? <TZander>nagamalli: how does your glib version have anything to do with your "waiting for locks or build slots.." error? <nagamalli>and i tried multiple ways to solve , but not getting where i need to do? <nagamalli>waiting for locks or build slots is for gnome-todo updation and as per the rekado's directions i stopped to update the gnome-todo <nagamalli>mroh,Yes it was but as rekado said we can't update that <nagamalli>I am having doubt regarding this where i need to start to make the changes? <guix-ci>Problem: Too many processes on hydra-guix-105 Problem started at 21:41:44 on 2020.04.23 <TZander>civodul: I complained a bit more constructive in the bugreport. This command line interpreter baffles me. <lispmacs[work]>is there some package that get's you all the stuff you normally get from gnu-build-system in a package definition *kmicu has assumed evolution-data-services is a dependency for gnome-todo. <raghavgururajan>What is the correct syntax for `(("$srcdir/configure --enable-maintainer-mode "$@" || exit 1") ""))`? <nagamalli>kmicu,Yes I was aware of adding it to gnome.scm but the issue is about glib version i think? *kmicu did not see much folks complaining about guix args parsing but thatās not the case for GNU tar and GNU find šŗ <TZander>kmicu: those tools can be understood. There is a logic to the cl-parsing. <guix-ci>Resolved: Too many processes on hydra-guix-105 Problem has been resolved at 21:46:44 on 2020.04.23 <noobly>hi #guix, I need a super lightweight pdf tool to concat two files, ideally something command line.. any recommendations? I searched for both pdfsam and poppler with no success <raghavgururajan>I get syntax error (substitute* "autogen.sh" (("$srcdir/configure --enable-maintainer-mode "$@" || exit 1") "")). What is the correct way here? <jonsger>noobly: we have poppler which should provide pdfunite <nagamalli>kmicu, evolution-data-services ?? I Found evolution data server there in gnome.scm , is that are u telling about? <noobly>jonsger: oops i spelt poppler wrong, thanks! <ryanprior>me: im gonna try and package hugo, how many dependencies could it possibly have, <ryanprior>the package: 78 dependencies. the number is seventy eight. <kmicu>nagamalli: yes, of course, you are correct. <ryanprior>and that's direct dependencies, go doesn't have a formal package system that can actually tell me how many transitive deps there are. so we're gonna call that a mid-term goal <ryanprior>TZander that's a good research project for somebody who feels like spending 78 minutes to find out =D <nagamalli>kmicu, So I need to update the evolution-data-server inorder to fix glib version problem <ryanprior>TZander that list is correct if you vendor all 78 dependencies <TZander>it depends on nothing but glibc... (and a go compiler). *civodul hadn't read help-guix since end of Feb, oops! <ryanprior>and git, which fetches 78 go libraries which are required to build :) <kmicu>nagamalli: thatās my understanding based on a rapid look šŗ <nagamalli_>kmicu: ok.Thank you for your suggestion.I will try that once <kmicu>nagamalli_: wish you luck ( ^_^)ļ¼ <hugo>ryanprior: I'm not sure how I feel about being put into a package... <ryanprior>hugo okay I take it back, you can package yourself when you're good and ready šļø <mfg>rekado: i'm back at the texlive thing. it seems to only work reliably when i install the whole distribution :\. So i will have to stick with that (at least for now). <ryanprior>we badly need a `texlive:just-the-parts-i-need-for-the-love-of-god` package <nagamalli_>kmicu:Thank you , How can i know the issue is fixed about glib after the updation of evolution-data-server? <andrew>Are there any alternatives to using the official substitute server to get better download speeds? <kmicu>nagamalli_: Maybe you could compile gnome-todo locally by using your local guix repository with the patch and test whether gnome-todo works for you. <nagamalli_>I have tried to build the evolution-data server pkg,but getting erros of unbound variables of some packages like curl,static bash <nagamalli_>after updating with new version of 3.36.1 of evolution-data server <kmicu>nagamalli_: using a paste service and sharing those error messages here is a perfectly ok way to find out whatās wrong. <mfg>ryanprior: that seems nice, but it also seems like lot of work :D <andrew>20 KB/s seems somewhat slow when downloading lots of substitutes <kmicu>nagamalli_: š¤· I donāt know whether itās a locale issue or something else. Could you also build āhelloā package and share a log? <civodul>andrew: is this the throughput you're getting from ci.guix.gnu.org? <civodul>and you're consistently getting that throughput? <andrew>civodul: although it seems to vary, about 25% of the downloads get around 15MB/s *kmicu bets on the orbit, ISS. <civodul>the machine seems to have a reasonable load right now <rekado>mfg: do you have a test document that we could experiment with? <Emilia>Hi everybody! I'm trying to install guix the manual way. In my system configuration I have (keyboard-layout (keyboard-layout "uk" "altgr-intl")) under (operating system). I get the following error on guix system init: "build of /gnu/store/....-console-keymap.uk.drv failed". It doesn't fail if I change "uk" to "us" or to "fi". <bricewge>andrew: Does it occur with big and small substitutes? <Emilia>The backtrace gives: /gnu/store/...-console-setup-1.194/bin/ckbcomp: Can not find "symbols/uk" in any known directory <bricewge>Several people including connected trough the ISP Free are having similar speed issues since a week or so. <mfg>rekado: i'm exporting org files to pdf with lualatex. I'm using the polyglossia package, so i have texlive-latex-polyglossia in my manifest, but lualatex doesn't find it. Ideally this document should compile: https://dpaste.org/yhRd. (Some packages cannot be found and i don't know in which package they are, if they are packaged e.g. amssymb, mathtools) <mfg>So nothing too special i hope :D <bricewge>andrew: Could you try āguix environment --ad-hoc mtr -- sudo mtr ci.guix.gnu.orgā to see if you have some packets loss? <andrew>bricewge: Yes, it looks like the 25% that reported 15 MB/s were < 100 KB in size. Everything larger reported in the 20 - 40 KB/s range. <mfg>rekado: also i'm not a texlive pro :p <andrew>bricewge: So the smaller downloads probably didn't have enough data to report an accurate number. <mfg>i'm out for today o/ <andrew>bricewge: Looks like I'm getting arround 30% packet loss from a jump from LA to Hamburg <jonsger>would be interested if using Deutsche Telekom as ISP causing also troubles, because their peering with dfn (that network before ci) is bad <andrew>jonsger: Yeah also seeing around 10% loss on a couple dfn hops <katco>ideally, is it better to author a package based on an importer, or one which pulls from source directly? authoring one from an importer seems like it would remain more maintainable. does anyone have opinions? <civodul>katco: normally importers produce something that pulls from the source too <civodul>or did i misunderstand your question? <bricewge>andrew: So it really seems like the same issue <civodul>it would be great if we could get to the bottom of it <katco>civodul: well, e.g.: i was just considering updating `emacs-slack`, and for fun i ran `guix import elpa -a melpa slack`. this fetches from melpa.org unlike the current version which fetches from github.com <bricewge>I really don't know how to do about it tough... <andrew>civodul: I was using mtr as bricewge suggested <katco>i should say "this produces a package which fetches" <civodul>katco: oh right, it's better not to take from melpa.org because it's unstable IIRC (tarballs updated in place, etc.) <civodul>so yeah, better to use github directly <civodul>which also works with "guix refresh" anyway :-) <rekado>hmm, I donāt know why but file access seems to be really slow on ci.guix.gnu.org <rekado>Iām running āguix install gcc-toolchain@7ā, a package that has been built <rekado>in another screen I hit tab after pasting /gnu/store/qr5w1s1bg6c7pill74q0mhpwqbrr11aj-gcc-toolchain-7.4.0 and nothing happens <civodul>rekado: ah yes, some time ago i was going to install iotop to investigate but that hasn't happened yet <civodul>so maybe there really is something worth investigating :-) <rekado>it just continued to tell me that the profile derivation will be built <rekado>cuirass and the mumi worker spike *rekado restarted the mumi worker to make sure weāre running the latest version <rekado>mbakke: what are you building on ci.guix.gnu.org? <civodul>rekado: guix processes |recsel -e 'ClientCommand ~ "guix build"' -p ClientCommand :-) <plstohelp>ever since my guix vm crashed mid pull all i get is this <rekado>mbakke: are you building *all* packages for armhf on ci.guix.gnu.org? <plstohelp>guix pull: error: error parsing derivation `/gnu/store/6m9iiwb8xj9qgjvw4zb235srbj0m8nqd-compute-guix-derivation.drv': expected string `Derive([' <rekado>plstohelp: this is disk corruption <civodul>that's a lot of stuff but i think everything is offloaded <plstohelp>i'd really rather not reinstall, does anyone know hoe to fix this? <rekado>plstohelp: run fsck on your disk <mbakke>rekado: indeed, I'm helping Cuirass along a bit <rekado>mbakke: I think it shouldnāt be a problem anyway, just checking :) <civodul>rekado, mbakke: ok to restart cuirass? <rekado>plstohelp: the error says that the file is corrupt <andrew>plstohelp: could you boot the vm with a livecd and them fsck <plstohelp>where should i report this issue? seems like it runs counter to the strengths of transaction package managment <plstohelp>guix pull: error: You found a bug: the program '/gnu/store/0pdpq2nv4493nw651rwz28v337gzgrb1-compute-guix-derivation' <plstohelp>failed to compute the derivation for Guix (version: "0f25feb15917c48cd37ee9b5a2cfc40014b0c520"; system: "x86_64-linux"; <plstohelp>host version: "6e8f58fe31884e48ff771d0093fa6bbc77c40ed7"; pull-version: 1). <plstohelp>Please report it by email to <bug-guix@gnu.org>. <plstohelp>i tried to gc and pull again, ill see if it works this time