***Xenguy_ is now known as Xenguy
<gabber>i am unable to see fonts in the SuperCollider Help Browser (which is a panel on the top right if you start `scide`). can others reproduce that issue? (running `guix shell -- scide` should suffice) <jab>anyone here running nextcloud with guix system? I am trying to find the easiest, stable, will be easy to upgrade type way to self-host nextcloud.... ***califax_ is now known as califax
<clever>cnx: at least with nix, the nix-channel tool downloads a tarball of the latest stable for a channel, and that tarball is what winds up in the store <breathe>I'm here to say thank you to whomever packaged Gajim 1.4.x! <cnx>thanks, clever, i was wondering if it's possible to pull from the system channel, but it seems it's not the case <cnx>what are the dependencies for running ./bootstrap? <cnx>nvm it's the same as for building guix <NaturalNumber>I hate to be the " well, ACHTUALLY" guy, but Sudoku isn't a Japanese game, or at least there's no reason to believe it is. There's a theory that it's difficult to make word puzzles in Japanese, so they gravitated towards number puzzles. <NaturalNumber>Sudoku is a Japanese logic game that exploded in popularity in 2005. GNOME <NaturalNumber>Sudoku is meant to have an interface as simple and unobstrusive as possible <NaturalNumber>while still providing features that make playing difficult Sudoku more fun. <cnx>how do i specify the output with specification->package? e.g. i want git:send-email <cnx>thanks jackhill, it's another issue but interesting nevertheless ***taiju` is now known as taiju
***darosior6 is now known as darosior
***kir0ul7 is now known as kir0ul
<cnx>lol are avatars on mumi always people's initial? <PotentialUser-53>so, if I use the qemu image I can set up Guix the way I want and generate an installer from the config.scm for the real machine ? <dale`>How can I specify that a certain file appears in a user's home directory in a system configuration file? <dale`>I'm making a boot disk, and need encryption keys to be available to the user. <unmatched-paren>although it shouldn't be too hard to hook up `guix home` to a field in the user configuration <dale`>Could you elaborate, I'm interested to do that? <unmatched-paren>What I mean is that this is a feature that would be entirely possible to implement, maybe even easy, but it hasn't been implemented yet. <unmatched-paren>and then being able to specify (user-account ... (configuration (local-file "my-home-config.scm"))) <dale`>That *would* be great! Is there a request for the feature to go in, or should I perhaps submit one? <unmatched-paren>but I'm not sure whether you should send it to help-guix or bug-guix <dale`>Hmmm... probably I'll send to bug-guix. <unmatched-paren>dale`: Could you add that it should be possible to use either (home-environment) or a file-like object for the (configuration ...) field? <teythoon> CXX nix/nix-daemon/guix_daemon-nix-daemon.o <teythoon>/gnu/store/ndmfigzifg96qfppq2kp4kpkidi99rv1-gcc-12.1.0/include/c++/bits/new_allocator.h: In member function '_Tp* std::__new_allocator<_Tp>::allocate(size_type, const void*)': <teythoon>/gnu/store/ndmfigzifg96qfppq2kp4kpkidi99rv1-gcc-12.1.0/include/c++/bits/new_allocator.h:125:20: error: '__throw_bad_array_new_length' is not a member of 'std'; did you mean 'bad_array_new_length'? <antipode>Does someone have Emacs set up with Guix Home + Guix System? <dale``>teythoon: I don't know if this will help, but guix shell is preferred over guix environment nowadays. Perhaps you could try using 'guix shell -D guix'? <unmatched-paren>teythoon: It was decided that `guix environment`'s command-line interface was a bit weird, so it was deprecated in favour of `guix shell` <teythoon>the documentation doesn't seem to reflect that <unmatched-paren>`guix environment foo` == `guix shell -D foo`, `guix environment --ad-hoc bar` == `guix shell bar` <teythoon>if that is bad advice, maybe don't serve that <teythoon>sorry, it's just my patience is running thin atm... <unmatched-paren>I know, it's generally agreed that non-devel manual/en should be moved to manual/1.3.0/en and manual/devel/en should be moved to manual/en <teythoon>i stopped leaking my environment into the guix shell -D guix, and that at least fixed the build daemon issue :) <unmatched-paren>Maybe it's a bad interaction between the Guix gcc and the foreign distro's libcxx/libc++/whatever the standard C++ library is called. <jeandudey>Hi! Can someone take a look at #56252? I'ts a small fix for polkit-mozjs and polkit-duktape. <NaturalNumber>Hello Guix,. I can't seem to get XMonad-contrib working. I've been trying to build everything from source. using xinitrc-xsession and ghcup helped a lot. Here are my error logs in case anyone can help. I'm headed to bed soon but I will read your messages.ghc-make.log https://paste.debian.net/1247797 <RudraveerMandal[>In the installation process, after the partitioning it shows an error /dev/sdb is still in use even though I didn't assign anything in /dev/sdb <antipode>antioxidant-build-system now supports building and running (some of the) tests <efraim>looks like I need to bump the version of mono to fix building on aarch64 <NaturalNumber>i'm running into a pretty confusing error after removing a package "λ guix remove fontconfig <NaturalNumber>guix remove: error: while setting up the build environment: creating directory `/gnu/store/p9pmngmxjaazc128h674bnw1mvxlxvxq-ca-certificate-bundle.drv.chroot/gnu/store/yzq4jfa6qa35kvqhj76y0c5vg7xq41im-python-pytest-bootstrap-6.2.5': No space left on device" <NaturalNumber>somehow i ran out of disk space, so i deleted generations and guix gc <efraim>are you out of space or out of inodes? check 'df -h' and 'df -i' for the partition that has /gnu <NaturalNumber>it's so confusing because i also get a message that says something like "you have 80GB free space" when i guix gc -F 30GB <NaturalNumber>efraim, yeah, it said I was out of space, but the weird thing is i'd delete like 50GB then get the same messages again <antipode> NaturalNumber: Are you using Nix somewhere? <antipode>Maybe one of the NIX_... environment variables are set, leading Guix to talk to the nix daemon instead of the guix daemon <antipode>sneek: later tell NaturalNumber: NaturalNumber: Are you using Nix somewhere? Maybe one of the NIX_... environment variables are set, leading Guix to talk to the nix daemon instead of the guix daemon. <civodul>in a nutshell :-), it's easier to use & faster <mbakke>anyone know how to deal with "error: cannot convert from y to UTF-8" from 'git am'? <efraim>maybe try git apply? it seems to be less picky <mbakke>that works, but then I need to manually adjust author, date, and message <mbakke>it will probably be faster for the 10-patch series than figuring out how to convert the messages to utf8, but less fun! :P <mbakke>the 'file' command says the file is "RFC 822 mail, Unicode text, UTF-8 text, with very long lines (788)" <apteryx>or perhaps git am -- fails and leave the commit message pending, then fix it manually with 'patch' or 'git apply', then commit, reusing the message still in place <apteryx>yay, got jami to build without qtwebengine, reducing its transitive closure by 450 MiB. <podiki[m]>apteryx: did you happen to get a chance to try out the fhs container? I have a more polished and complete version i will submit soon to guix-patches <podiki[m]>it has been working well for me, even got those darn appimages to work <vivien>I’m trying to use the copy-build-system. My source is a tarball that extracts to a single directory, named "usr", containing sub-directories (bin/, include/, lib/, etc). My install-plan is ''(("usr/" "")), but guix complains with: In procedure stat: No such file or directory: "usr/". I pass -K to keep the failed build, and right under /tmp/guix-build-….drv-N, I have the "usr" directory. What do I do wrong? <podiki[m]>question: I have a hidden package but want it in a manifest (in the guix shell container code), naturally it complains...what to do? <civodul>podiki[m]: you can probably refer to its variable instead of using specification->package <taiju>I have a question about IRC channels: I created a channel called #guix-jp without knowing the concept of IRC namespaces, but I don't have the authority to manage this channel. <taiju>Should I not continue to use this channel? <taiju>We have created a community called Guix-jp. So we would be happy to use this #guix-jp channel. <apteryx>podiki[m]: sorry, I've yet to try it! <apteryx>will be happy to look at the more final version though <drakonis>taiju: it has to be registered by someone that manages the #guix namespace <podiki[m]>civodul: hmm...so I would place this in the functions that create the manifest for the profile...let me see <podiki[m]>apteryx: no worries! I've been using it for some work I do that works well in the container (some python stuff that likes to use binary web browsers) <taiju>drakonis: I see. I know that the Chinese community uses the channel name #guixcn, but the community name itself is guixcn, so it doesn't feel strange. But our community name is Guix-jp, so we want to use #guix-jp. <taiju>It is possible to name it #guixjp, but we wanted to check the feasibility of creating a channel named #guix-jp. <nckx>taiju: I'd answered you in PM just before you asked here. <taiju>nckx: I was not aware of it. Thanks! I will check it out. <nckx>Omitting the ‘-’ (e.g., #guixcn) doesn't magically exempt channels from namespacing; don't waste time with that. <nckx>taiju: Well, it was mainly me saying I'll discuss with the rest of the maintainers, so I'll await their opinions. <nckx>I'll continue the discussion here: <nckx>My personal opinion is that if your channel abides by the GNU FSDG & Guix CoC there is no issue at all, and the discussion is over, enjoy your new channel :) <nckx>If not… things get complicated. <nckx>Of course I'd urge you to follow the FSDG/CoC for general reasons, because they are good, but of course I am biased. <podiki[m]>civodul: hmm, not getting that to work, but maybe I was trying in the wrong spot <civodul>podiki[m]: you need to find the variable of that package, and then add the corresponding use-modules and variable reference <podiki[m]>yes, already have that; I think it does work and now getting an error due to some other refactoring <podiki[m]>in making the mappings and symlinks more declarative, I'm trying to create a link for something that doesn't exist now <podiki[m]>since I added to mappings in the container, without checking that e.g. the profile has a /lib directory <taiju>nckx: Frankly, it is difficult for me personally to answer this question, but I believe that it would be practically difficult to mandate compliance with the GNU FSDG in our community. <nckx>Interesting! Why would that be? <nckx>Note that *people* don't have to ‘comply’, they can *run* whatever they want, the FSDG here simply asks them not to *promote* proprietary software to others. <taiju>nckx: For example, when we get questions from community members about how to install Guix on their laptops, we may give the inevitable answer. <nckx>…which is of course ‘buy an old ThinkPad or spend 4x that amount on a Librem’ 😛 <nckx>No, OK, I understand. I've added that to the mail. <taiju>nckx: I see. We may be able to prohibit to promote. We have been very careful about that in the wording of our website, and we have made the source code public and provided it under CC BY-SA 4.0. <civodul>"prohibition" sounds strong (and scary!) <civodul>it's more about making it clear that the focus is and remains free software <podiki[m]>civodul: with (specification->package "glibc-for-fhs") no longer complains about unknown package, but doesn't end up in manifest/container. weird <civodul>podiki[m]: i think i'm missing context; could you share the manifest? <taiju>One concern I have is that we will be managing a channel in our community, and if software is distributed there in binary form (e.g. packages using copy-build-system, etc.), will this be a problem? <civodul>taiju: you can do whatever you want in your channel :-) <civodul>the only thing is that the Guix project won't advertise channels that provide non-free software <podiki[m]>civodul: so I'm trying to insert a hidden package into the container at the code level (previously just cons-ed to the opt list) <civodul>podiki[m]: looks like a typo at the bottom: '(add-hoc-package (specification->package "glibc-for-fhs")) <antipode>taiju: On IRC channels: in my understanding, #guix is only de-facto English, there is no rule against other languages <antipode>so merging #guix-jp into #guix seems to be a possibility <civodul>there's even a rule that all languages are accepted here, if i'm not mistaken <podiki[m]>civodul: ah right. alas, the container option list just wants package (string) names, so I'll have to add it elsewhere I think? <podiki[m]>antipode: for me too, along with main guix website <civodul>antipode: should be back now; looks like nginx went awry <antipode>On an unrelated matter, seems like crates.io sometimes has the same test problem as pypi <taiju>In conversations with community members, some expressed concern that if we declared compliance with the GNU FSDG <taiju>some expressed concern that they would not even be able to have conversations about nonfree software within the community. Do such conversations constitute promoting? <nckx>What you don't see there is the delay. <civodul>nckx: ah yes, so it looks like shepherd fails to stop the thing <civodul>and then it gets stuck in waitpid on it <nckx>I was testing the /devel/ → sane URL migration (although I still have to write redirects for old /devel URLs). <nckx>I would have restarted it but thanks for noticing. *civodul uses M-x view-hello-file as a somewhat limited phrase book <nckx>civodul: /en/manual/devel/de/ → /de/manual/de and /manual/ → /manual/version-1.3.0 <nckx>I will submit it but do want to make sure we're discussing a working patch. <nckx>What's annoying is that the manual now 404s despite not running any new code. <nckx>So state has entered the building. <roptat[m]>taiju promoting is saying or implying that proprietary software is good, and that's very context dependent <nckx>And will be purged either by mcron, or me returning from dins. <mbakke>oh, I think I may have misunderstood the </1> style tags, they are not resolved in the translation <mbakke>only some are missing though, checking <taiju>roptat[m]: Perhaps I am worrying too much, because I am sure that people who like Guix share the common value that free software is desirable. <podiki[m]>Rudraveer Mandal: have to run, but which install image are you using? the 1.3.0 definitely has had partition issues reported, try the "latest" image if you didn't already <RudraveerMandal[>I am using a custom built image from system Crafters that includes nonfree Linux kernel and drivers <RudraveerMandal[>The workflow rebuilds the iso every week and the one I downloaded is about 5 days old <podiki[m]>so it should be the same as the standard guix installer, but you'll probably have to ask elsewhere (since this is the official guix channel) <podiki[m]>but you could try with the regular guix installer to see if there is any difference <KarlJoad>What is the best way to call out to a specific Makefile target in a package build? An example package definition from the Guix repo is all I need. *podiki[m] has to run now, literally <KarlJoad>Or perhaps just an arbitrary shell command?... I'm not sure the best way to put this package together. <RudraveerMandal[><podiki[m]> "but you could try with the..." <- I can't get internet working with it so I can't use that installer <jpoiret>RudraveerMandal: are you on bare metal or in a vm? <KarlJoad>Specifically, I am trying to get gllvm to build. The go-build-system fails because there are no Go files in the import-path, because they are split into subdirectories, depending on the binary. <drakonis>is it time for the monthly dose of rust trademark trolling? <roptat>KarlJoad, there's import-path and unpack-path, maybe that can help somehow? <roptat>you might need to define multiple packages for the same source, too <acrow>vagrantc: thanks. I've got some rudimentary licensing code, it's tools for use from the repl right now. Is it too soon to share? <KarlJoad>roptat: I should not have to define multiple packages. All the binaries are in the same module, so I think I am in the clear about that. I have changed the import-path and things broke. But I managed to make it work in a development shell, compiling manually. Need to tinker some more. *vagrantc has managed to trick, er, encourage... acrow to write some stuff to help with license/copyright review of guix for debian packaging :) <roptat>KarlJoad, maybe have a look at go-github-com-operatorfoundation-monolith-go for instance, it needs to build from the subdirectory monolith, so it unpacks one level above the import-path, maybe that can help? <roptat>taiju[m], do you install ibus-anthy in the system profile, or in your user profile? <KarlJoad>roptat: That may be what I have to do. I am not entirely sure. I thought I tested that approach earlier and it failed. But, I am not entirely sure. I need to try again later. <RudraveerMandal[>Like could anyone post the default config by guix when you choose awesome or some other wm <drakonis>just add qtile to your system packages and it should show up on the display manager <roptat>do you know how to configure dconf properly? <roptat>"dconf: GDBus.Error:org.freedesktop.DBus.Error.ServiceUnknown: The name ca.desrt.dconf was not provided by any .service files" <nckx>Let's look at this manual business. <roptat>so I guess it's a dbus issue. I have a ca.desrt.dconf.service file in my profile <nckx>civodul: So I ‘guix system roll-back’ to a generation from 16 May (the previous generation, in use earlier today). It has /gnu/store/2wblf4hv7dgssdg0fdb0rxx8zngwfpcb-nginx.conf. Yet killing -9 nginx and running ‘herd (re)start nginx’ still uses /gnu/store/s64q287nimmr4kw95ijsc4yb1l1qdbm4-nginx.conf, from a later generation. <nckx>This sounds bogus as heck. <roptat>mh, I even have "(simple-service 'dconf dbus-root-service-type (list dconf))" in my os <nckx>And now nginx is running (berlin PID 78097) but clients can't connect to the server. <podiki[m]>roptat: does the dbus service file exist and in the right directory to be picked up? (forgetting exactly, but something like share/dbus-1) <podiki[m]>well in better news I figured out how to hack a hidden package in the container code; there's the "expression" option that will read/eval a string :) <roptat>podiki[m], the service file exists in the dconf package, but I'm not sure how to check which directory dbus will search in <podiki[m]>roptat: I think it uses the general xdg_(something) variables <roptat>the file is in one of $XDG_DATA_DIRS <nckx>* The way that the manual and home page have been brought back up is not reassuring. I ignored the Shepherd, and just ran sudo -i nginx …. <podiki[m]>i've had to try to debug dbus before and it was a real pain, though there are some tools to look at what dbus is receiving and looking for <roptat>dbus-monitor doesn't show anything to me <podiki[m]>as in it shows no output? or doesn't show receiving the call to open something with that service file? <podiki[m]>sneek: later tell civodul I realized there was the "expression" option that will read/eval a string to add a package, so that's how I hacked in the hidden package <roptat>it doesn't show receiving the call <roptat>I tried to strace the daemon, and it doesn't show anything either <podiki[m]>are you in a DE or just WM? for just WM I run mine with dbus-run-session which fixed some dbus issues for me <roptat>there are two dbus-daemon running <podiki[m]>well for me pgrep dbus-daemon shows 4 processes, so.... <roptat>there's one run by the shepherd, which doesn't receive the messages, and one by my user <roptat>and it's running with a config file from at-spi2-core <podiki[m]>I find dbus to be magic/frustrating; I guess the only maybe useful thing I have for you is that I run my WM (xmonad) with dbus-run-session to launch it <podiki[m]>for me I use xinitrc-xession, basically it executes my .xinitrc file to run the session; the last line is "dbus-run-session ~/.xmonad/xmonad-x86_64-linux" <podiki[m]>so you could also just change the Exec line in the xsession file? or explicitly run dbus another way, I'm not sure <vagrantc>nckx has bee so prompt with the diffoscope updates i haven't even gotten mine in for the month! <nckx>Truly I am renown for my promptness. <roptat>I couldn't start a session like that, but I don't know why <mekeor[m]>podiki: what does dbus-run-session do? (i also use xmonad via xsession) <podiki[m]>mekeor: it is supposed to run dbus as long as the process it launches runs, basically <podiki[m]>I was just having some weird dbus issues with my plain wm setup, and that was somthing I found online that worked for me <roptat>oh, do you do "exec dbus-run-session openbox" or "dbus-run-session openbox"? *mekeor[m] has "exec ~/.xmonad/xmonad-x86_64-linux" in ~/.xinitrc *mekeor[m] uses "xinit -- ~/.guix-profile/bin/Xorg :0 vt1 -keeptty -configdir ~/.guix-profile/share/X11/xorg.conf.d -modulepath ~/.guix-profile/lib/xorg/modules" to start x as user <podiki[m]>I've used sddm, gdm, still the same way I launch though, final line is "dbus-run-session /path/to/xmonad", as opposed to "exec /path/to/xmonad" that I used to use <podiki[m]>but I haven't looked or touched that in a long time <roptat>ok, I forgot to make .xsession executable <roptat>it works now, and ibus saves my settings properly <roptat>although, I still can't use anthy <florhizome[m]>couldn’t we have a home or system service that creates a session file for any wm and makes sure it’s displayed in the login manager? <florhizome[m]>in general, I wonder if we couldn’t move desktop environments to user config more or completely, at least wayland using ones … <roptat>so I can switch to anthy, but it still doesn't input Japanese <jackhill>florhizome[m]: I'd also be interested in such a thing. However, I think there may be some reason it can't be done, at least for wayland sessions. At least sddm needs to have globally-installed sessions: https://github.com/sddm/sddm/issues/916 <roptat>I just had to restart the session, it works now! やった! <florhizome[m]>jackhill i have a manjaro installation and there I use greetd/tuigreet to start into a wlroots compositor (wayfire) session <apteryx>why doesn't "\bNO_DEFAULT_PATH\b" match NO_DEFAULT_PATH in " NO_DEFAULT_PATH)" ? <KarlJoad>I am working on a project in a guix shell. But I cannot run the binary because it dynamically linked to LLVM libraries using a name, rather than a path. Is there a way to fix that? So the binary gets a full path to the .so? <apteryx>are you building using the gcc-toolchain? the ld-wrapper should bake runpath in the produced ELF, allowing it to be found, I think <KarlJoad>apteryx: No. This is a compiler built using LLVM. <apteryx>does it use 'ld' from PATH, like GCC would? <roptat>I think there's a clang-toolchain <KarlJoad>apteryx: It is using the ld from my path, yes. binutils is installed through guix home and made available that way. <apteryx>you should add one of the ld wrappers to your environment <apteryx>the best is to use gcc-toolchain or clang-toolchain <KarlJoad>Gotcha. I need clang-12, so clang-12-toolchain, right? <roptat>that would be clang-toolchain@12 from the CLI <KarlJoad>roptat: I am using packages->manifest, so that syntax does not work. <KarlJoad>Nevermind. I confused where the major version number goes. clang-toolchain-12 in this case. <jackhill>florhizome[m]: I use GDM, but remember the sddm issue. Maybe we could install a global "user's choice" session, which checks for a session in their default profile. <efraim>this help-guix thread about nim and nimble sent me down a rabbit hole. I don't know how to use it but I almost have the test suite passing. <KarlJoad>roptat, apteryx: Using clang-toolchain-12 fixed it. Thanks. <civodul>nckx: i think rollback doesn't "downgrade" system services (?) <sneek>Welcome back civodul, you have 1 message! <sneek>civodul, podiki[m] says: I realized there was the "expression" option that will read/eval a string to add a package, so that's how I hacked in the hidden package <florhizome[m]>jackhill: I would more think of a computed sessions directory because I like the idea of installing a profile for each „desktop environment“ <florhizome[m]>This kind of means we need some profile management code in guix home… <KarlJoad>I want to build a special version of stumpwm that includes contrib modules in the package, similar to how gcc-toolchain does special stuff to make gcc, binutils, and glibc understand each other. Would gcc-toolchain be a good idea to look at for inspiration?