<mbakke>civodul: nice fix & test on the memoization problem :-) <vagrantc>wow, was able to start sway on the pinebook! ***drakonis1 is now known as drakonis
*ecbrown follows hurd blog post instructions with big bag of popcorn <noobly>ummmm so i installed mupdf, confirmed by output of "guix package --list-installed | grep mupdf", but command $ mupdf returns command not found.. any ideas? <Blackbeard>noobly: can you please share your .bash_profile or .bashrc <noobly>Blackbeard: sorry had to handle something, here is which: which: no mupdf in (/run/setuid-programs:/home/daws/.config/guix/current/bin:/home/daws/.guix-profile/bin:/run/current-system/profile/bin:/run/current-system/profile/sbin) <noobly>bash_profile: if [ -f ~/.bashrc ]; then . ~/.bashrc; fi (I imagine that's not too substantial so I'll post bashrc in one sec <Blackbeard>noobly: would you please logout and login again? or reboot whichever you prefer <noobly>Blackbeard: interesting, yeah I'll try that <noobly>Blackbeard: so i exited i3 and logged in again through the graphical log in screen, is that sufficient? mupdf is still not found <Blackbeard>noobly: it is strange that your path is giving you this trouble, can you please reboot? it should not be necessary but let's see what happens <noobly>ok it didn't feel sufficient so i rebooted, no luck <noobly>Blackbeard: nothing is returned.. <Blackbeard>noobly: try again $ guix package --list-installed <noobly>Blackbeard: yes, mupdf is still there <Blackbeard>hahaha what's going on with your path noobly did you modify something? <noobly>/run/setuid-programs:/home/daws/.config/guix/current/bin:/home/daws/.guix-profile/bin:/run/current-system/profile/bin:/run/current-system/profile/sbin <noobly>Blackbeard: lol, not intentionally <Blackbeard>noobly: this should not change a thing but can you add this to your .bash_profile and then reboot <Blackbeard>noobly: no actually just copy paste that on a terminal <noobly>i tried doing this earlier too when it complained about a different thing, it says either 'nil: command not found" or Define alias for "nil": <Blackbeard>export PATH="$HOME/.guix-profile/bin:/$HOME/.guix-profile/sbin${PATH:+:}$PATH" <jackhill> noobly, Blackbeard I think it's just that the executable is not called mupdf: <jackhill>/gnu/store/iqw8ig54dx5dba095jm77b1lw5j8c5k9-mupdf-1.16.1$ ls bin/ <jackhill>mupdf-gl mupdf-x11 mupdf-x11-curl muraster mutool <Blackbeard>jackhill: thanks! I should have started with trying mupdf <TAB> <noobly>i ran the export command, still no mupdf. but mupdf-x11 works! <Blackbeard>but I have not used pdf on guix and I used to ran it as mupdf <noobly>thanks guy i owe both of you some beers <jackhill>Blackbeard: you're welcome. I'm glad it was something easy instead of something mysterious :) <Blackbeard>noobly: add this to your .bash_profile: alias mupdf="mupdf-x11" <Blackbeard>it was just that noobly had trouble with the path and latex earlier so I was expecting to be something over complicated <xelxebar>Man, guix environment is so nice. I can install whatever I need with impunity! No need to worry about cluttering up my system. <damo22>i was thinking of using guix to generate binary cross-toolchains so i can keep them backed up <damo22>deploy them as tarballs anytime i want <damo22>otherwise i would need a full distro just to get gcc ***wxie1 is now known as wxie
<lisbeths>I wanna try switching my laptop to guix and I think I have the skill level to do it. I vaguely remember putting debian on this machine and not having drivers for wifi or for the touchpad or something. Is there another distro I can try a live install of that has the same drivers as the kernel in guix so that I can try the drivers without formatting? <damo22>sounds like you need a live guix usb <damo22>i thought guix was a package manager not a distro <lisbeths>there is also a distro named guix that uses the guix package manager <lisbeths>part of the philosophy of doing this is this prove that guix can act as a standalone system <lisbeths>but the core reason guix exists is two fold: <lisbeths>1. Make more packages intercompatible with each other <lisbeths>2. Make the package manager portable to nearly any other distro <damo22>As of version 1.0.1, the standalone Guix System can be installed on an i686, x86_64, ARMv7, or AArch64 machine. It uses the Linux-Libre kernel and the GNU Shepherd init system <damo22>i would assume any peripheral supported by linux-libre would work on guix therefore <damo22>lisbeths: therefore you could try installing linux-libre kernel as an additional kernel and reboot into that <lisbeths>I don't know too much about kenrels but I am fairly sure the drivers are in the kernel, yeah? <lisbeths>so I just need a distro with a live installer that's got the same kernel or newer, yes? <damo22>but the standard linux kernel has non-free firmware, and linux-libre does not <TZander>damo22: practicaly all kernels separate the non-free firmware in an optional package. So saying its part of the standard linux kernel sounds wrong. <lisbeths>I'm on xubuntu which has a nonfree kernel <damo22>TZander: not really, linux-libre disables the loading feature of non-free firmware, so you need a standard kernel if you want to load non-free firmware <damo22>ie you cant get non-free firmware addon for libre kernel <lisbeths>theoretically I can change my kernel out <TZander>thats a very different thing than saying that the kernel itself is non-free <raghav-gururajan>sneek, later ask apteryx: Were you able to look at the mirror and formatting patches for linphone? Thanks! <damo22>lisbeths: yes, you can just swap your existing kernel that loads non-free firmware with a libre kernel <damo22>to test if everything still works <damo22>lisbeths: in that case you could just find a live iso of some fully free distro that already has a libre kernel in it <lisbeths>is there no easy wawy to look up my hardware and see if it supports linux-libre? <lisbeths>yeah but I am not sure which distro I can use <damo22>i am pretty sure the intel wifi does not work with linux libre <lisbeths>If I'm gonna install guix I have to manually switch out the kernel to a nonfree kernel <damo22>lisbeths: you can swap the intel wifi chip with another one that works with free software <rekado>damo22: installing just linux-libre isn’t going to give you Guix System. <rekado>Guix System is more than just linux-libre and other packages. <rekado>you’ll need to do “guix system init” to actually replace your system with Guix System. <damo22>rekado: i know, i am suggesting lisbeths can test her hardware for linux-libre support before wiping her system and replacing with guix <rekado>please note that we don’t want the official Guix project channels (including this IRC channel) to host discussions about installing non-free software. <lisbeths>If I am not free to find out how to make my computer run guix I am not free to use it and I shall leave. <rekado>I would suggest booting a live USB image. <lisbeths>It is the user's right to install a nonfree kernel. <damo22>rekado: you just lost a potential user by not reading the backlog correctly <damo22>no one was suggesting to her to install nonfree softwrae <rekado>what I wrote above is a disclaimer <rekado>my suggestion was to boot a live USB image <rekado>and I was about to point them to the blog post about customizing the kernel <rekado>guess I’m not typing fast enough <TZander>has anyone else ran into the "this could be nicer" with reverting one generation? doing `guix package -S 17 -d 18` does not work. It doesn't switch before checking delete and then just errors out. I would like it to first switch and then delete. <rekado>TZander: this sounds like a good idea. Could you please send email to bug-guix@gnu.org? <chiefgoat>Taking my first steps with Guix System on a Dell R720 using the ESXi hypervisor. Fingers crossed this works! <krusnus>Hi all! My guix seems to have borked. None of the user installed programms work and most guix commands return "error: unsupportet manifest format". I'm new to guix so I'm 90% sure I'm wrong about this, but I think I've somehow messed with an env file after accidentally selecting it when choosing a wm for lxqt. I think that it's only my user that's <krusnus>been messed up so maybe a full "reset" of my user would help but I'm not sure how to do that. any help would be appreciated! :) *kmicu has nothing against ‘losing users’ with such attitude. [Jokin’] Later pals! <kmicu>(Friendly reminder that ‘user’s right to eat meat (install blobs) ≠ user’s right to request meat in official guix vegan (libre) restaurant’. Being angry at such restaurant for not serving blobs is at least rude (to put it gently).) <krusnus>I tried creatia new user and specified (shell zsh) but when log in i t gives me the error "/gnu/store/[hash]-zsh-5.8: /gnu/store/[hash]-zsh-5.8: is a directory" and logs out again :/ <xelxebar>make authenticate on master is failing for me: could not authenticate commit 77704cb13e5bebf412297dab764a00849a3cfdc0: key A0C5E3522EF8EF5C64CDB7F0FD73CAC719D32566 is missing <xelxebar>I pulled master earlier today, but just for clarity git rev-parse HEAD returns 406130550912ee1254198b6349d7d7892e59217b <NieDzejkob>my guess is that it's somewhere that's not ~/.config/guix/current/bin/guix <krusnus>NieDzejkob I would but i tried creating a new user but that didn't work (it didnt get it's ow home directory even after that option was specifically set to true) and now I've somehow messed up even more and my originaluser has somehow lost permissions for its own home direcory. is there any way for me to, from within guix, just reset the system acc <krusnus>ording to my config? like guix init again? <krusnus>neither rolling back nor reconfiguring with a config that i know has worked before has given any results <guix-vits>nly: you know, i'm start to think that tor-service is rather broken :) i've found myself restarting it time-to-time. also nft-service can fail on boot, if there is name instead of ip (httpOK = {logs.guix.gnu.org, ...}). <ATuin>anyone using stumpwm having problems after upgrade? ***verne.freenode.net sets mode: +o ChanServ
<nly>guix-vits: see (define-record-type* <tor-configuration> in gnu/services/networking.scm <nly>maybe you need to provide a configuration file in your guix.scm? <sneek>nckx, raghav-gururajan says: If you try the gajim and gajim-omemo, let me know what the issue might be. May be I can work on it, Thanks! ***verne.freenode.net sets mode: +o ChanServ
<nckx>kmicu: Good thing popcorn's vegan! Some people… <guix-vits>nly: config that provided is appended to the default one. and it 'somehow works', but if connectivity lost -- restart. Is Your's tor-service working today? <guix-vits>nckx: strange: #40495. Word 'shader' tells me nothing, and tests for the one test-dependency failing since 2018 or so. <nly>guix-vits: ah i see, will (service tor-service-type) work? <nly>it's not working but i'll give it a go <nckx>guix-vits: A ‘shader’ is a little programme that runs on your GPU. Where's it mentioned? I can't find it in that thread or my mailbox. <nckx>I'm building taisei. Let's see. <guix-vits>nckx: taisei working, but tests #f for spirv-cross, and spirv-tools (spirv-cross needs spirv-tools for the tests). <kmicu>nckx: ‘some people…’ as my analogy is not helpful or is that about the op? <nckx>kmicu: Our inflammable friend. Not you. <kmicu>Roger. I tend to use vegan vs meat or smoking vs non‑smoking space to show how out‑of‑touch some folks are. <nckx>‘Failed to check if MSL 2.2 is not supported. It probably is not.’ <NieDzejkob>krusnus: Just run `chown -R youruser /home/youruser', tbh <NieDzejkob>that's the only part here that's not being managed by guix ***ssouth is now known as simonsouth
<ecbrown>i'm on wip-hurd-vm with a fresh guix+buster. following blog, and did: ./pre-inst-env guix build -f gnu/system/hurd.scm and get (among warnings about #:declarative? #f) "error: failed to load 'gnu/system/install.scm': No such file or directory <nckx>guix-vits: I'm trying to update spirv-tools and spirv-headers because this whole thing looks like a coupled mess, but let me know if you already have. <nckx>If that doesn't work I'm tempted to just downgrade spirv-cross if it does. <guix-vits>nckx: only thing i'd tried is to update headers (maked compile fail for -tools if -Werror not OFF) <guix-vits>funnier part is, that -cross "release" versions want non-release (just commit) versions of -headers. <sys2>hi Guix! does anyone here use guixsd for rust programming? If so, do you use a patched rustup? Or what do you use to get the tools/source/other rust versions you need? <sys2>no problem. I think I can get rustup working.. that seems to be the easiest way to get some tools + source <brendyyn>sys2: you can look at cargo-build-system <brendyyn>It seems to pull meta data out of cargo files to make packages <brendyyn>in the git repo thats guix/build/cargo-build-system.scm, and guix/build-system/cargo.scm <nckx>guix-vits: Thanks for that hint. So 1.5.1.corrected won't work? Genious. <nckx>I don't know the first thing about Vulkan, I'm just banging black boxes together until they stop crying. <guix-vits>nckx: i didn't tried to update both -tools and -headers, though. <sys2>brendyyn: to package rustup? I think it requires nightly rust... which would be a blocker <nly>guix-vits: tor won't start for me, '... cannot read "var/lib/tor/keys" directory...' <nckx>Welp, they're still crying. -tools 2020.2 + -headers 1.5.1.corr = error: ‘SpvOpTypeRayQueryProvisionalKHR’ was not declared in this scope <brendyyn>sys2: You would define your own packages for each and/or some importer to help. <sys2>gotcha. yeah I think I could get away w/o rustup if I had all the tools (I think clippy is available, but I need to look into if I should be using rls or racer or both, and I'm not sure rustfmt is available). The other thing is source. I see the source in the store... but I'm not sure I can install /just/ the source; can I? <nly>i wonder if a chmod counts as a fix, but it's working now <nckx>sys2: efraim is probably our resident Rust expert, but IIRC doesn't actually write Rust. <nly>"herd start tor" that is <nckx>One of those Is stands for ‘ironic’. <nckx>nly: The service should do that, please report a bug if you haven't previously messed with those permissions. <sys2>good to know! thanks. Hi efraim :) <brendyyn>sys2: You sound like you know a lot. You'd likely be a pioneer of rust development. In Guix there is a whole bootstrap path for rust starting from mrustc <sys2>brendyyn: I'm sort of perpherially aware of the bootstrap path (in that I've watched icecat build a couple of times and read that one guix blog post) <brendyyn>sys2: i think in generall developing in guix means using `guix environment' to get everything you need, defining new packages as guix packages. then compiling from source locally requires that the build system's method of finding dependencies doesn't conflict with guix's design. for example if it hard codes some path like /bin/foobar, that will fail <GuixGuy>Greetings, I have a question about licensing for a package. I have packaged python-evdev and the current copyright info in the file is from the person who created the package. Python-evdev itself has it's own copyright from the creator. Do I add both into the file? <nckx>GuixGuy: Which is the first ‘the file’, and is the same as ‘the file’ at the end? <GuixGuy>I don't understand what you are asking nckx <brendyyn>GuixGuy: the mailing list was broken at the time i sent that patch. if you have made corrections and its better than my patch you should email yours in <nckx>Then we are both in the same spot 😃 <GuixGuy>Your package is more complete than mine brendyyn. <nckx>GuixGuy: I don't understand your question at all, but maybe it's not relevant with brendyyn's news. <brendyyn>I think i used the texinfo format lazily <nckx>I can only parse it as ‘do I add Copyright © 2020 The Person Who Wrote Python-evdev’ to, say, python-xyz.scm and that answer is no. Only your name. <GuixGuy>The first copyright is the creator of python-evdev and the 2nd is the person who put the package together that I found on github/gitlab <brendyyn>Oh. in the (license...) field you only put the license of the software its self <nckx>brendyyn: I'd drop the markup entirely for the package name (just write ‘Python-evdev’), use @file{/dev/input/} and @code{evdev} and @code{uinput} unless those are actual binaries you can type at a shell prompt. <nckx>GuixGuy: Since you based your work on someone else's you would indeed add ‘Copyright © 2018 Nils Gillmann’ to the .scm file's header. <nckx>But since they were a regular contributor to Guix chances are they're already in there. <nckx>(As ng0, which is probably the name they prefer now.) <brendyyn>GuixGuy:did you have the same home-page? <sys2>brendyyn: would you recommend reading the manual to get set up w/a testing environment for that? Or suggested blog posts? I'm wondering if guix vm is a better fit for testing out new package definitions (to not be completely deleted after testing), or if guix environment is enough <brendyyn>sys2: why no just install guix on your distro and start testing it <brendyyn>things are only deleted if you run `guix gc' and they are not in a profile. <sys2>brendyyn: way ahead of you :P. I'm on guix sd right now <sys2>ahh, so guix environment would only leave behind data, if anything? <brendyyn>you can think of /gnu/store as like a cache for everything downloaded and build. <brendyyn>to "install" something means to have it all magically symlinked into your profiles directory <GuixGuy>How do you reply to someone and it shows as (name): in irc? <sys2>I guess what I'm getting at is guix environment just creates a new pointer to stuff in the store... then it's removed (leaving stuff in the store ok to be garbage collected), and that's about it, right? <sys2>I guess I don't get much more isolation from guix vm.. except for data (e.g. if a package I'm testing pollutes /etc/ or something) <sys2>but maybe that's worthwhile? <brendyyn>then even you wont have the ls command unless you add coreutils <sys2>but stuff in that environment could still modify other parts of the system, no? <nly>nckx: i had messed with the permissions, so it's probably my bad <brendyyn>as your user. not with root permissions. Oh also there is --container <sys2>hmm... --container looks like what I'm looking for... just access to the PWD? looks like /etc only has "group" and "passwd" when I do that. should be enough isolation. thanks! <brendyyn>i think container uses some linux container feature, maybe chroot. may be more convienient that an entire vm. also vm can be made both with a shared store or as a complete image. its fast to generate new ones with a shared store <sys2>or cgroups! no idea though <sys2>definitely still getting my bearings w/guix. my main profile is a mess. need to split it out and figure out a good upgrade strategy. but that's for later :) <nckx>GuixGuy: I don't know, but I don't see any (nick)s above. If you do, that's probably something your client is doing on your end. <nckx>guix-vits: So I've got -tools 2020.2 to build with the latest -headers commit (you mentioned that earlier so perhaps I retracing your steps; oh well) and now get… different -cross test failures! <nckx>So many ‘home pages’ to choose from, indeed, but that's the most homey one. <nckx>Ah. You mentioned that it needed unreleased headers earlier, no? Anyway, I'm about to punt on this. <nckx>'#extension' : extension not supported: GL_EXT_demote_to_helper_invocation <nckx>I'm not going to learn what the hell SPIRV actually is just to bump a package. <brendyyn>GuixGuy: do you need the python-evdev package? <nckx>guix-vits: I'm fine with disabling -cross tests if the game actually runs. <guix-vits>nckx: i'll try tomorrow to build it "as is", outside of build env. <nckx>guix-vits: Is that a sarcastic ‘runs’ or a real one? 😛 <sys2>is there a way to get the source of a package installed? I see /gnu/store/*package-I-want-src.tar.gz <sys2>and /gnu/store/*package-I-want-src.tar.gz.drv <nckx>‘guix build --source PACKAGE’ will give you the source for the current definition of PACKAGE. <sys2>nckx: thanks! Can I do that in a manifest? I don't see the source listed in the outputs <nckx>Packages are just records, so you can always (package-source PACKAGE), this will return an origin record but I don't know if they can go straight into manifests. <sys2>"lacks output :src" and :source <nckx>Because it's an input, not an output 🙂 <sys2>guix build --source rust just outputs the path to the source tar lol <sys2>which I guess is a start... but I'd like to avoid unpacking and managing that myself. might go the rustup route then <nckx>…? What's lol about that? <sys2>I guess that's the "built" version. it makes sense... I just want the unxz'd source somewhere <sys2>right... built version of the source <sys2>more or less... the source of the version that's installed <sys2>I wonder if I can give racer a path to that xz and it can reference the compressed version... <nckx>sys2: Gotcha. I'm afraid that's not generically possible: ‘unpack’ is run as part of the build, not the source, derivation. There's no way I know of to ask for (unpacked-tree-for (package-source any-random-package)). <sys2>ok thanks for your help! the options I see are: get rustup to run on rust stable (not sure how much work), download the source via git, check version installed, checkout tag (doable but a little gross + maint), package rust nightly (talked about on the ml a bit, seems untenable), request source be available (unsure if this is desired by anyone else), patch the rustup binary to work (it's missing a link to glibc or something). I think I'll <nckx>I don't know Rust(up) so 🤷 We have a discussion list at guix-devel at gnu point org for more in-depth discussion. <nckx>When you find yourself posting compound lists to IRC it's worth considering putting your thoughts to mail. <sys2>going to fix (2) locally to get it going. not sure if there's something that can be fixed in the project or guix to make that easier. not a huge fan of running untrusted binaries... but this seems like the easiest way forward for now <nckx>sys2: Definitely check out patchelf (in Guix). <nckx>It allows you to change the path to the interpreter (ld.so) and any libraries. It's a hack, and a minority of binaries don't like to be patched, but it often works. It's just a pain to maintain all the patchs by hand. <sys2>do you know how to find the relevant libgcc_s.so.1 in the store? that's the one that needs to be linked <nckx>sys2: I wouldn't use store file names directly, but install packages to your ~/.guix-profile/ (or a separate profile just for rustup). libgcc_s.so is a pain because it's provided by ‘gcc’ which is hidden, and I'm honestly not sure how you're supposed to install it. <GuixGuy>sneek: later tell brendynn I need python-evdev for packaging Lutris <nckx>Oh: guix-vits: taisei works here too (ThinkPad X230T, i915 GPU)! Also updating -tools & -headers allows me to enable the -tools tests. So that's progress. Perhaps. Laterz o/ <sys2>nckx: ok thanks. there's a bit of guile to install it (asked about it before). I have it because of another package... but I think I /will/ use store paths because this is a oneoff.. just going to make sure I'm using the installed version and not an older version <nckx>sys2: Sure! But what did you point them to? If it was /gnu/store/<hash>-…, bear in mind that Guix has no idea the store item is ‘in use’ and may delete it when you run ‘guix gc’. If you're aware of that, no prob. <sys2>it was the store... but I think this is fine because rustup is a bit strange... it's a `curl ... | sh` script that downloads a prebuilt binary... then /that/ installs the rustup binary to ~/.rustup/. The new binary doesn't appear to have this issue and can self update from here, so I don't expect issues in the future <emsyr>Hi! After I run 'guix package -u' and 'guix gc' (after a reboot to be sure all updates are activated), enlightenment applications menu keeps pointing to previous versions of /gnu/store/<hash>-[app_name] that no longer exists. Any idea how to deal with it? <civodul>emsyr: that sounds like a bug, probably Enlightenment has a cache under ~/.cache or so <civodul>if you could find it and email bug-guix, that'd be great <emsyr>civodul I don't see a relevant enlightenment directory or file under ~/.cache <hulten>I just upgraded packages, including chromium@80.0.3987.163. It crashes. <hulten>Are there some obvious things to check before I report it as a bug? <hulten>Is anyone successfully running Chromium 80.0.3987.163? For me, chromium@80.0.3987.149 works fine (same system but different user, so not comparable). <emsyr>civodul as a matter of fact there is a directory ~/.e/e/ that includes several info about enlightenment. I have already searched it but I didn't find something useful about my problem. <civodul>emsyr: does "grep -r /gnu/store ~/.e" give hints? <emsyr>hulten I'm running Chromium 80.0.3987.163 without problem. <hulten>Thanks emsyr. I'll dig a bit, because I guess there's no obvious bug. <nckx>hulten: Have you tried moving all caches out of the way before starting it? Does running it from a terminal print anything interesting? <vincelevi>Hello, I'm trying to add support for runit-based foreign distros to guix-binary tarball && guix-install.sh script, but I'm missing something, the binary tarball I generate in my checkout does not contain the new service scripts (I did the same as in 73fbe04107d38f4561636c74d28d7a4935cbb1ef)... <alextee[m]>if anyone's free there are a few patches I sent for new packages that are pending review/merge :-) (40537, 40510, 40471, 40468) <hulten>I'll get back to you, nckx, thanks so far! :) <civodul>emsyr: this may be the culprit; could you send all this info to bug-guix@gnu.org? <nckx>hulten: Keep your thanks, I won't be able to help you further as I don't use Chromium 😉 <sys2>nckx: I spoke too soon >.<. I get "no such file or directory" from the installed rustc now. The interpreter and ldd output look good... so I'm not sure what's wrong here. it appears to be linked to 64 bit libs too... so that's fine. Odd <emsyr>civodul I'll delete some of these file to see if I'll force enlightenment to rebuild applications' menu. <emsyr>civodul I'll delete some of these files to see if I'll force enlightenment to rebuild applications' menu. <civodul>i had an idea that looked great but the damn thing resists <nckx>Don't you hate it when the bug ignores your fix. <nckx>civodul: So why does GDM show a Gnome session? *nckx puts on rubber ducky hat. <nckx>Your mail hadn't arrived yet. <civodul>nckx: so there's that gnome-session.desktop that's picked up when it shouldn't <civodul>for some reason, the solution i came up with doesn't work <civodul>like, how do you attach gdm to the right process? <nckx>Uhuh. My own limited experience with GDM debugging (the cache mess way back when) was not positive. <nckx>It is not designed to be debugged. <bandali>another idea: use gdm only for gnome, otherwise use lightdm or sddm for other DEs and/or WMs <efraim>emsyr: I'm on my phone right now, delete ~/.cache/efreetd <efraim>I talked with upstream, it has to do with the store having a ctime of 0. I try to delete mine on shutdown <nckx>bandali: That kind of defeats one of the main purposes of a display manager: to choose your session type. <efraim>So delete that directory and then kill efreetd. When it starts up it'll recreate the cache with the versions in your PATH <bandali>nckx, do we have anyone that frequently chooses/switches between gnome desktop and something else? <bandali>if not, that issue only concerns gnome <bandali>afaik lightdm and probably sddm too both support choosing from sessions <nckx>bandali: No, it locks all the *users* of the same machine to using the machine's desktop environment. <nckx>If it were per-user that would be bad but not as bad. <bandali>nckx, hm. does gnome only work with gdm? <bandali>man, why does gnome has to be so awful -_- <nckx>AFAIK (limited testing) certain shell features are missing if you don't use GDM. <nckx>Plus one nice thing about GDM that I don't know if other DMs do is multiple parallel sessions. <bandali>and the former too, but not sure if it works with gnome <nckx>I'm mostly familiar with SLiM (which is great for minimal set-ups). <nckx>Last time I tried setting up a spare ‘home’ laptop to be shared between me & my partner (so it has to boot nicely to GNOME and i3), GDM was the only one that worked without tiny little bugs propping up. But that seems so long ago now. <nckx>Problem is that I don't use GNOME & don't know what's missing when it's missing. *nckx .oO remember when I said I wasn't going to get nerd-sniped into debugging SPIR-V test failures? I'm debugging SPIR-V test failures. <nckx>But it's mainly just boring ‘upstream versioning is a mess and stuff is broken at release time, despite all coming from the same place’, nothing… y'know, fun. :-/ <sys2>nckx: terrible workaround to patch all rust toolchain binaries: https://paste.debian.net/1139767/. Asked in ##rust and they said what I said earlier (patching the rustup binary meant I didn't need to patch the toolchain) was incorrect <civodul>janneke, rekado, mbakke: i've pushed to core-updates the beginning of wip-hurd-vm <civodul>that is, everything before (gnu system hurd) <bandali>hey folks, has there been any discussion in guix forums on freedom concerns around rust? <lfam>The trademark thing, bandali? <bandali>lfam, that, and iirc there's also some money-related thing too <lfam>What's the money-related tihng? <bandali>discussing with folks in #gnu, we think they render rust nonfree <bandali><pehjota> [01:54:39] What actually makes Mozilla software like Rust and Cargo non-free is the prohibition on distributing *unmodified* copies for a fee. *nckx has only so much quarantine popcorn, people. <lfam>As always, citations please <bandali>and will either post here or on guix-devel <nckx>bandali: It is, if (say) FSF's legal counsel were to raise these concerns. <nckx>No offence to random IRC randoes. <bandali>pehjota is far from an `IRC rando', but i see your point <drakonis>that seems like a fine way to kneecap guix's repositories <drakonis>so, everyone else is wrong and only gnu is right in serving rust and cargo? <jlicht>drakonis: not really relevant in any issue except a popularity contest :-) <lfam>Please stay calm. The Rust code bears a dual-license, MIT (expat) or ASL 2.0. I think we can find a successful resolution to these questions <nckx>bandali: I didn't mean to denigrate them, I mean that I literally have no idea who's behind ‘pehjota on #gnu’, which was my point. <bandali>and not a contest indeed; serious concerns <drakonis>it just makes people less likely to want to use guix because someone arbitrarily decided that rust isn't okay to provide biaries <lfam>This question isn't about binaries <bandali>if rust/mozilla's trademark policies impose freedom-restricting conditions then it renders their software nonfree as-is <nckx>drakonis: Guix has a good track record of not listening to noisy but unfounded FUD (see ungoogled-chromium). It's not something I'd worry about. <bandali>in the event if rust is deemed nonfree, one possible solution i think will be simply rebranding it (as done with firefox) and shipping that instead <bandali>anyway, i'll report back with concrete citations/references about this <lfam>Well the difference between this question and ungoogled-chromium is that we discussed the ungoogled-chromium subject exhaustively. Whereas citations are still being gathered about the rust stuff <nckx>lfam: And we will weigh them as calmly as we did for u-g. <drakonis>i'd like to point out one issue with rebranding rust, keeping it up to date <str1ngs>civodul: does adding (service gnome-desktop-service-type) help with (service gnome-desktop-service-type) <hulten>I have again font issues with Chromium ("Could not find the default font"). <hulten>I have a bunch of fonts installed. <hulten>The first three lines of Chromium crashing on my system may be important: <hulten>[30947:30947:0411/194736.529718:ERROR:vaapi_wrapper.cc(481)] vaInitialize failed: unknown libva error <hulten>[30947:30947:0411/194736.577035:ERROR:sandbox_linux.cc(374)] InitializeSandbox() called with multiple threads in process gpu-process. <hulten>[30921:30921:0411/194736.390509:FATAL:platform_font_skia.cc(97)] Check failed: InitDefaultFont(). Could not find the default font <lfam>hulten: Wild guess, but did you regenerate the font cache recently? <hulten>No, could you remind me how to do that? <hulten>Also, should and/or could that a task of 'guix package'? <lfam>`fc-cache -f`, it's from fontconfig <lfam>I'm not sure! I'm not really using Guix on the desktop in this way <lfam>There has been discussion on the mailing lists <lfam>And, maybe it won't help you here, dunno <hulten>Okay, in any case, generating the font cache helped: chromium works, thanks! <nckx>It's not caching. There's no way it's caching. It was caching. <lfam>There must be a way to prevent it from crashing the program <xavierm02>- 'set-environment-up' phasebuilder for `/gnu/store/d45izzcrryzx5rg0k3gf5ryacgyj5377-xen-4.11.1.drv' failed with exit code 1 <xavierm02>How do I find the last version that worked and install that instead? <xavierm02>(I also tried with the ./pre-inst-env after pulling from git and it does not work either <guix-vits>xavierm02: are you did ./pre-inst-env from `guix environment --pure guix`? <lfam>xavierm02: I confirm it fails on our build farm <xavierm02>@lfam Is there some way to find the last commit at which it worked? <lfam>xavierm02: I'm not sure right now. It was possible a few years ago but I don't know how to do it with our new build farm <lfam>You could just try a commit from like a month ago <lfam>Or try fixing the bug :) <lfam>Sorry the answer isn't better <xavierm02>(I'm already trying to package something, if I manage to make it work, I'll look into fixing xen) <nckx>bandali: Do you use lightdm? It fails to build here due to several segv's during the test suite ☹ <bandali>nckx, not on guix; but i'd used on arch, debian, etc <nckx>I don't use it either, just encountered the failure and remembered our earlier discussion. <bandali>cheers, and sorry for the non-useful reply <nckx>What? I've played M-x tetris during Guix installations before… 😊 <Blackbeard>nckx: and that's a pretty good idea actually. I always use SSH though so I never felt the need <nckx>I use my own installer. Unfortunately, I don't think the default one comes with ‘real’ emacs. <nckx>I wonder if it would really make a significant difference. One gets quite a bit in return <nckx>eww, tetris, actual you know emacs, … <nckx>But yeah, as a games platform is pretty much dead. <nckx>So do I, but the lack of any recent triple-A titles is a bad sign. *nckx still waiting for emacs-doom-eternal. <sneek>Welcome back apteryx, you have 1 message! <sneek>apteryx, raghav-gururajan says: Were you able to look at the mirror and formatting patches for linphone? Thanks! <civodul>with current core-updates you can reproduce the famous animated GIF :-) <civodul>now we need the even crazier things janneke has been working on <apteryx>apteryx: I'm afraid not! I've caught caught up in something that draged on rather late in the night. I'll ping you when I do! <str1ngs>civodul: understandable, I'm just trying to think which what generate the sessions since it seems none are listed <str1ngs>I guess there is a derivation that should link /gnu/store/64pjwsrf6nr56g9k18sj1v6ylqvkqm78-emacs-exwm-0.23/share/xsessions/exwm.desktop ? <str1ngs>or I read the issue wrong and exwm is still listed? <civodul>str1ngs: correct, sessions are taken from the .desktop files <civodul>actually i sent another message to that issue but i think issues. isn't getting updates <str1ngs>gnome.desktop is found in gnome-session. is there a way to check if it's being installed ? ***ssouth is now known as simonsouth
<nckx>Blackbeard: I'm home. Apart from shortness of breath (no sport == Toby go coconuts) I'm doing pretty well. <nckx>🙂 You? You sure chose the right year to go to e-school. <Blackbeard>nckx: yes. I found a link to Ph.D.'s on privacy and most of the are cancelled :( <Blackbeard>But I gotta finish my master first so at least I can be distracted for now <Blackbeard>I was thinking on maybe doing the Ph.D. online too. <Blackbeard>I found this wonderful organization that helps with money to Latin American organizations that promote Digital Rights <nckx>Sounds cool. Best of luck. <efraim>'Requires' in a pkg-config file doesn't mean propagated-inputs, right? <pkill9>i made a bash one-liner for opening the homepage of a given package: xdg-open $(guile -c "(begin (use-modules (gnu packages) (guix packages))(display (package-home-page (specification->package \"$1\"))))") <pkill9>hmm it's not working with tennix <efraim>I'm only familiar in that I know on guix you should use 'flatpak --user ' <marmulak[m]>I was able to add repo and install a package using sudo, but then when I run the package (telegram) as my user, the window says telegram is running as superuser <noobly>I can't seem to find my gcc or g++ compilers, which gcc returns: which: no gcc in (/run/setuid-programs:/home/daws/.config/guix/current/bin:/home/daws/.guix-profile/bin:/run/current-system/profile/bin:/run/current-system/profile/sbin) <marmulak[m]>I expect flatpak to work systemwide because of the heavy dependencies it pulls (300mb for kde env, for example) but I suspect it hasn't been configured to work that way on guix system <marmulak[m]>and we're supposed to install everything within our own user environment <nckx>You can just look at $favourite_task_manager to see who's running telegram. <pinoaffe1>noobly: gcc is probably not available in your profile <pinoaffe1>you should probably install / open an environment with gcc-toolchain <nckx>noobly: Which package(s) did you install to get them? <nckx>What pinoaffe1 said in the meantime. <noobly>pinoaffel is probably right, but i installed libgccjit <marmulak[m]>well task manager shows the flatpak running as myself <nckx>noobly: But libgccjit doesn't contain gcc or g++. <nckx>marmulak[m]: Phew at least. <noobly>nckx: weird the synopsis and description imply it does. anywya, it was all that came up when i did guix package -s gcc, i have not been able to figure out how to 'Run `guix search ... | less` to view all the results.` <nckx>noobly: Good catch, thanks. That's a bug. <marmulak[m]>I guess using sudo to install or update flatpaks systemwide is reasonable <nckx>noobly: I don't understand the second part. What needs to be figured out? <marmulak[m]>now I'm having a problem with a random opam package that insists on having a "cc" executable <noobly>nckx: i had a feeling, but im sure i have some gcc presently installed, but which still can't find it <nckx>noobly: ‘guix package -I gcc’ will tell you which packages with ‘gcc’ in the name are installed, if you're curious. Although I guess we can be fairly certain now that none of them actually provide a gcc/g++ executable. <noobly>nckx: im starting to think the second problem has more to do with emacs eshell, so nvm <vagrantc>is core-updates going to be merged for 1.1? <nckx>noobly: Yeah… there something Special about eshell and pagers, I know that, but I don't use it myself. <nckx>Your guess is almost certainly correct. <noobly>nckx: how would I not have a gcc compiler? <nckx>By not installing one to your profile. <noobly>OO, ok, lol. I was thinkin system wide <noobly>ok I'll try and figure out the profile stuff <vagrantc>not much of anything system-wide on guix <nckx>In Guix, ‘having something on your machine’ is completely separate from ‘can run it at my prompt’. Takes some getting used to. <nckx>‘guix install gcc-toolchain’ will give you that. <nckx>And yes, it will probably just re-use a GCC you already had in your /gnu/store already. <nckx>With a wrapper to make it actually do what people expect, which is compile programmes for some reason. <nckx>1.1 is core-updates free. <noobly>so when I run `GUIX_PROFILE="$HOME/.guix-profile" ; \ source "$HOME/.guix-profile/etc/profile"`, it asks for an 'alias for nil", any idea what that means or implies, or what I should I say? <noobly>and thanks for the guix install gcc-toolchain idea <nckx>noobly: Eh, that sounds like eshell getting in the way again. I've certainly never seen that before. <nckx>We don't really have package maintainers, roptat's just interested & active. Not on-line now. <nckx>My tab key has betrayed me? <nckx>vagrantc: Why? It's not ideal, but it means it will actually be released this year. <vagrantc>nckx: fair enough ... i'm really itching to get the mesa fix in <vagrantc>but i guess that will happen when it happens <noobly>ok, guess I'll shelf eshell until I know my way around guix <noobly>thanks though, i never would've assumed that <nckx>vagrantc: Well, if it's something very important, can be grafted, and you can convince Ludo' to cherry-pick it to 1.1-rc2… <nckx>noobly: Eshell is one part of emacs that's never been able to tempt me. <nckx>bash takes care of my muscle memory and termite actually does the things that I expect from a terminal. Not… whatever eshell thinks that is. <vagrantc>nckx: it means that graphics acceleration works on pinebook-pro (and possibly other rockchip platforms) ... which is pretty niche <noobly>nckx: i would run M-x shell but it's too 'dumb' and won't open info pages and such <vagrantc>oh, it also *only* changes armhf/aarch64 ... <vagrantc>if that makes any difference in the cherry-pickable calculus <vagrantc>ee401ed9249fbe284ef1b9b437d39207ca88131b fwiw <nckx>vagrantc: So it's a core-updates-worthy rebuild but only on those platforms? I don't think that'll cut it. All substitutes would have to be rebuilt on those (slow) machines before we can release. <noobly>nckx: so after this guix install gcc-toolchain finishes, will it already be in my profile, or do i have to add it? <noobly>didn't mean to send that last msg <nckx>Judging by the sound my Overdrives have been making the past few weeks that's a lot of work. Might just be the weather. <nckx>noobly: It should be ready to go! <vagrantc>if i get a couple more dc-dc power supplies, i could probably spin up some aarch64 build machines when the sun is shining <noobly>awesome :) is that true in general when using 'guix install'? <nckx>noobly: Once your Guix is set up, yes. That's kind of its main job 🙂 <nckx>There are edge cases when dealing with environments (well, not edge cases, but people sometimes forget that you can't change environment variables in a parent process like your shell from a child process like ‘guix’) but in general, ‘guix install foo’ means you can run foo straight away. <roptat>marmulak[m], is there anything you need from me :) *nckx doesn't use a big fancy desktop so can't speak for menus. <nckx>roptat: IIRC opam works but tries to run ‘cc’, and setting CC doesn't help. <noobly>great, i've just been blindly hammering guix package -i $package, is that an alias for the same thing ? <nckx>‘guix install’ came later. <roptat>oh yeah, it happens sometimes when I use opam too, but that's because the package I want to build uses cc, and doesn't read CC (because most of the Makefiles in OCaml world are hand-written, so they don't care about that) <roptat>that's an issue with the way opam works, it uses whatever tool is available on the system, so if you have no cc, the package cannot build <roptat>marmulak, on m'a donné un vague résumé, mais je ne sais pas si j'ai bien compris <marmulak>roptat: I was testing the latest update to opam and it was working good. I got one project installed which pulled a lot of dependencies. I later ran into a snag with another project that needs the tls package. One of its dependencies in its build runs a command that I guess blindly invokes `cc`. opam init also complains about wanting cc, but it seems most of the software finds gcc on its own <roptat>(kinda assuming you speak French because of that "oui", sorry if that's your only French word ^^') <marmulak>I know more French words but my French is not good in any way at all. It's mainly because I know Esperanto and English that I can decipher some French texts <marmulak>let me find the command ouptut for more detail <roptat>so I think the package invokes "cc", but we don't provide it in our gcc-toolchain package, so it can't work <roptat>not sure about that at all, but what if you try "alias cc=gcc"? <marmulak>which works in my user shell, but I think opam is using its own environment <roptat>oh but the build occurs in a container, so you have to make the symlink be in a directory included in that container <roptat>iirc, that's /usr, $HOME and /gnu <roptat>I can't remember exactly, but the issue is that your cc is not included in the environment <roptat>hopefully the alias is passed to the environment <roptat>another possibility is to use "guix import opam" to import your package and make it a proper guix package <roptat>and you can of course report the error to the upstream of that failing package, saying that some systems don't have cc, so they should at least provide a way to specify CC <roptat>(and bonus point if you provide them a patch ;))