IRC channel logs


back to list of logs

<rekado>it’s not unexpected then than projects that have been around for a long time and have had actual independent project management see the association as a loss.
<rekado>“then than” –> “, then, that”
<rekado>or whatever
<pkill9>rekado: what are the benefits of guix joining the GNU project?
<vagrantc>i thought guix was a gnu project ... almost since it's inception?
<vagrantc>since guix's inception, that is
<rekado>pkill9: GNU means different things to different people, and the perception of the umbrella project may change over time as the member project evolves.
<rekado>personally, I see Guix as an embodiment of what I consider to be the core of the original ideas of GNU
<rekado>to build a unified operating system that has is built around user empowerment (for the lack of a better term).
<rekado>it’s not just for the sake of building something neat.
<rekado>more concretely: Guix liberates users from oppressive sysadmins, it dissolves the hierarchy that is often baked into traditional GNU/Linux systems, it adds a safety net to make this practical, and it gives users unified tools to exercise their freedom and control over software distributed via Guix.
<rekado>it provides a unified way to manage the system at almost every level through Guile
<rekado>to me Guix delivers on the promise of the GNU idea, and it is a big part of what I wished GNU as an actual *project* was.
*rekado always waxes philosophically at this time of the day/night
*rekado —> zzzZ
<raghavgururajan>rekado: Wow! This is the first time I see you (in #guix) talking deep. ;-) In good way I mean.
<roptat>raghavgururajan, that's not very nice ^^'
<raghavgururajan>roptat: Hey I meant in a good way, nice way.
<roptat>I know :)
<raghavgururajan>Rarely, we all have moments where go into philosophical train of thoughts right. Thats what I meant.
<raghavgururajan>*Sometimes, we all
<leoprikler>We all know the true GNU project is debating how much influence RMS should have over the FSF (:
<raghavgururajan>oh. I was referring to his thoughts on how guix liberates users.
<raghavgururajan>Pearls of wisdom, i'd say. :)
<lfam>I thought his statements were awesome. I feel similarly about Guix :)
*raghavgururajan flies back to test snikket hosting
<leoprikler>I think most people working on Guix do so due to its liberating aspects. I just wanted to make a joke at the expense of those, who'd rather GNU be a personality cult.
<lfam>Who else is looking forward to a new release of Guix this weekend :)
<lfam>Last chance to fix bugs and change packages!
*raghavgururajan screams Me! Me! Me!
<lfam>Then, we can dive in to core-updates
<lfam>Or something like that... I haven't been paying attention to anything besides the release prep
<raghavgururajan>lfam: Will c-u be merged to master this weekend?
<raghavgururajan>Ah, okay.
<lfam>It won't be included in the release (it's not ready yet)
*raghavgururajan had to clarify
<lfam>I meant, we can dive into the process of working on core-updates
<lfam>I'm really optimistic. Our CI / build farm is better than it's ever been. It's going to be a pleasure to work on
<raghavgururajan>I have few more outstanding patches (#47770), for c-u.
<lfam>It won't freeze for a while, don't worry. My impression is that there is still much to be done
<lfam>This is the longest we've ever gone without finishing a core-updates cycle, so it will be kind of thorny
<raghavgururajan>I see.
<lfam>But, problems with the CI plus the pandemic really slowed everything down
<lfam>I think a lot of us lost focus for many months
*raghavgururajan wants to REPL the shit out of COVID-19 virus
*raghavgururajan wants to make it seg-fault inside our body
<lfam>Did you see that the RNA sequence of the mRNA vaccines was reverse engineered and published?
<cdegroot>Far out question: I can't help being sad about that C++ wart called nix/ between all that beautiful Scheme code. It's not that much code - are there plans to redo that work?
<lfam>Replacing it is a goal, cdegroot, and we have made progress on it over the years. It's not a first priority, since it does work, but we share your sadness
<lfam>It's true that it's not much code, but the structure of the code doesn't lend itself to piecewise replacement, if I understand correctly
<raghavgururajan>I didn't look though. But I defer to lean towards mRNA for vaccines.
<lfam>I'm very grateful, I was able to get shot 1/2 yesterday
<lfam>I showed up at a clinic one hour before close, and they had some leftovers
<cdegroot>lfam: thanks. I might want to take a look. There's things I want to do to the daemon that are currently extremely No Fun At All so I'm not doing them ;-)
<lfam>What do you have in mind?
<raghavgururajan>> lfam‎: What do you have in mind?
<raghavgururajan>I'd go for proteins which are more immunogenic than nucleic acids.
<lfam>I meant that question for cdegroot :) I don't know *anything* about medicine
<raghavgururajan>Oh. xD
<lfam>Gene therapy is a big deal here in Philly, though. I know that there are next-gen cancer treatments being used with great success, based on gene therapy
<lfam>It sounds like you have some knowledge, though
<lfam>This stuff:
<raghavgururajan>OMG! I love GT. I was planning on researching to use it to cure thalassemia.
<lfam>Is this the kind of work you do?
<lfam>It's not the same as thalassemia, which I had never heard of, but I noticed that people on working on use of mRNA to address sickle cell
<lfam>All this work is such a gift for humanity
<raghavgururajan>I use to though. But my interests now leaned more towards IT and fsw.
<lfam>Maybe you could do IT for medical research :)
<raghavgururajan>That;s my idea. Bioinformatics. :)
<lfam>Nice :)
<cdegroot>lfam: mostly mDNS additions for substitute servers. I have a bunch of machines and I only want to build stuff once.
<lfam>We recently added Avahi support, for fetching substitutes from the LAN
<lfam>The crucial thing is the substitute signing key. If you are building substitutes yourself, you'll need to pass your builder's key around
<cdegroot>Yeah, but the daemon doesn't support it AFAICT.
<cdegroot>(also: RNA is my favorite molecule; it's the thing that started this whole thing that is us ;-))
<cdegroot>And indeed, key management is the interesting bit :)
<lfam>Another "quick fix" is to run an Nginx substitutes mirror. I do this for my home network, which includes 4 Guix installations. This way, I only download things from the main build farm once
<lfam>This is documented here:
<lfam>But, it's not exactly what you asked for :)
<lfam>Signing key management is declarative on Guix System now, but that doesn't help on other distros
<cdegroot>Yeah, I know there's way around it, I was mostly looking for a neat project and then I found C++ code and had to run and get a bucket :P
<cdegroot>So now I'm angry lol.
<lfam>False advertising!
<lfam>I don't think anyone ever expected to still be using the C++ code for this long. But everyone got distracted working on a distro :)
<cdegroot>(I am on Ubuntu, looking to move to Guix system but there are still small hurdles - for example, I need asdf-vm for work and builds of asdf-vm on Guix tend to be unhappy often; currently struggling with Erlang)
<lfam>The attitude of the early days:
<lfam>"Yeah, Perl..."
<lfam>As if we wouldn't have included it otherwise ;)
<lfam>(A lot of Nix tooling was in Perl at that time, IIUC)
<lfam>Commited on 2012-11-04
<lfam>Old school!
<cdegroot>I briefly considered porting the C++ code to Erlang or Elixir. Then you could have clustered Guix daemons and tons of fun. But probably a hard sell.
<lfam>Yeah, you'd have to make your case persuasively
<cdegroot>(I last got paid to type Perl code in '05. After a ~ten year pause)
<lfam>We'd really like it to be Guile. Otherwise, it's hard to justify the cost of a change
<cdegroot>I even considered adapting LFE ("Lisp flavoured Erlang") to understand Scheme ;-)
<cdegroot>But then I looked at my schedule and decided that I would not finish all that in a lifetime lol
<lfam>That was part of the effort to rewrite the daemon in Guile
<cdegroot>thanks for the pointers.
<lfam>cdegroot: You mentioned that Avahi "local substitute discovery" isn't part of the daemon. But, I think it is
<lfam>At least, the daemon has code for discovering these substitutes
<lfam>Sorry if I misunderstood what you said
<cdegroot>Hmm... then I must have missed something.
<cdegroot>I'll dig again :)
<lfam>Or maybe it delegates functionality to Guile portions of Guix. But, the daemon has a --discover option for this :)
<lfam>I'm looking at commit 79f9dee3c4c0e6d21066f142116a537207ae7ba4
<cdegroot>Ah, yup, you're right. I looked in the C++ code whether things were wired up for Avahi discovery but came up empty-handed. Well, something to try :) (the main reason I want this dynamic is that a bunch of installs run on laptops and even on VMs on laptops, so they're not always there; currently, with fixed substitutes, stuff times out which is not nice)
<roptat>isn't the avahi stuff delegated to publish / substitute, which are written in guile?
<roptat>oh, unrelated, but I sorted the translation components on weblate by priority (guix > website > documentation > packages), and added screenshots to add context for the website translation (,position for instance)
<roptat>I found that sometimes the context is helpful to understand how to translate (more so in Guix than the website though)
<roptat>(the screenshot is in the column on the right)
<roptat>also, a reminder that you can easily help translating Guix to your native language for the release if it's not English :)
<lfam>That interface looks great
<roptat>it's just a bit slow compared to a local editor, but it has a lot of nice features
<roptat>come on, do you want your language to be behind French and Slovak? :D
***Server sets mode: +cntz
<mikegerwitz>Hey, everyone. I'm going to be submitting a patch series soon. The Guix manual says to send one message to guix-patches@ first, followed by the debbugs address. But looking at other patch series on the list, the timestamps in the headers suggest people aren't doing that. Is a References header (generated by git format-patch) sufficient for debbugs to handle this itself?
<nij>There's a missing "=" behind "--profile"
<nij>Hello! Currently, I put additional channels I want under /root/.config/guix/channels.scm , as suggested by the official manual. I wonder if I can add this in to config.scm, so that I don't have to manually copy channels.scm into that dir every time I start working on a new machine?
<brendyyn>hmm, doesnt look like %default-channels is configurable
<apteryx>mikegerwitz: Hi! If your patches are related (or dependent), they should be reviewed together, so should go to the same debbugs ticket
<brendyyn>apteryx: they're asking how to do that in one go, rather than two steps by first waithing for the debbugs response
<apteryx>if you just git send-email to bug-guix it'll create one ticket per patch, that's we recommend to use 'git format-patch --cover-letter ...', send the cover letter to bug-guix, then send the rest to the email associated to the opened issue (you'll get a message back from Debbugs after sending the first mail).
<apteryx>err, s/bug-guix/guix-patches/
<apteryx>I'm not sure I understand the question, but AFAIK that's not possible
<apteryx>(to have all the patches appear in the same issue with a single git send-email command)
***iyzsong-- is now known as iyzsong-w
<mikegerwitz>apteryx: Okay, thank you
<apteryx>does someone has experience with debugging a one shot Shepherd service?
<apteryx>I guess copy the use-modules block and lambda to a REPL, and evaluate it there
<apteryx>Also, I reckon the one shot services are run as root?
<apteryx>(for the system instance of Shepherd)
<logiz>where can I find the settings for the package lightdm? I assume there is a .scm somewhere but I can't locate it
<apteryx>not sure, but the package itself is in (gnu packages display-managers) (guix edit lightdm)
<logiz>I assume that gets added the the system/configuration.scm, but I guess my confusion is that my configuration.scm is the one that installed the system and I guess I have to dump the current one?
<brendyyn>logiz: there is currently no lightdm service in guix. there is one submitted to bug #35305 though
<logiz>oh I guess I was using gdm then :X, thought of the wrong install
<brendyyn>logiz: in gnu/services/xorg.scm its all defined. its added to the os through the default %desktop-services. you can modify it with modify-services
<logiz>I found an xorg.scm in /gnu/store/hash-xorg.scm can't find a services dir
<logiz>sorry super new to all this fs stuf
<brendyyn>I'm not sure what you're actually doing to get that
<logiz>I'm just trying to figure out how to set a default resolution on boot instead of running xrandr -s everytime
<brendyyn>i mean why are you looking in /gnu/store?
<logiz>because I don't know where a gnu/services/xorg.scm is and I'm just running `find`
<brendyyn>i was pointing to the git repo of guix
<brendyyn>so you want the resolution set just for gdm?
<logiz>I figured gdm was the way to login to a resolution
<logiz>either that or grub
<brendyyn>could even be something you just configure in your desktop for your user
<logiz>maybe I'll search more
<brendyyn>actually in xorg-configuration there is a resolutions opiton
<logiz>ah just saw it
<brendyyn>If you know how you would do the same thing on for example Arch linux, it would make it easier with understanding how to translate it to Guix's way of doing things
<logiz>I'm actually running guix in a vm, but that's because I'm trying to do all my dev work in a vm these days, and I ran into the same issue with fedora too, but I've just suffered typing xrandr every login, figured there's a better way
<brendyyn>when you run qemu you can set a resolution i think?
<logiz>yeah I'll try adding an xml change to virt-manager and see if that does it
<logiz>and then continue trying to learn guix :)
<logiz>thanks brendyyn
***pocketroid_ is now known as pocketroid
<brendyyn> is a sponsored project to add a Rust front end to GCC
<tissevert>hi guix
<Frosku>Has anyone had any luck installing gccemacs on guix? I tried the flat channel but it seems to be completely broken.
<Frosku>"Cannot open load file: No such file or directory, comp."
<brendyyn>Frosku: i have it.
<Frosku>brendyyn: Teach me your ways
<brendyyn>Frosku: Does your EMACSLOADPATH still have a refernce to the eamcs 27.1 path?
<rekado_>raghavgururajan: maybe you can come work with Altuna’s bioinfo group at the MDC. We write free software and hack away at cancer, herpes, and all that which can seriously spoil your day/life. And we use Guix.
<Frosku> echo $EMACSLOADPATH
<Frosku>/run/current-system/profile/share/emacs/site-lisp /run/current-system/profile/share/emacs/27.2/lisp
<Frosku>27.2 yes
<brendyyn>Frosku: simple way to fix that is to log out and log back in again. also probably you want to make sure you uninstalled the old emacs
<Frosku>Lemme try
<Frosku>What *should* loadpath be?
<brendyyn>Frosku: last week some patches by Leo Prikler were submitted to fix the need to do that, but they aren't in guix yet
<Frosku>Are you using that watsonflat repository?
<Frosku>brendyyn: Thanks, any idea how I set native comp output path? Just setq or is it an env var?
<brendyyn>Frosku: you're on your own there. I'm just using it with Doom. I ran doom upgrad, sync, and compile and just use it as normal emacs now, nothing much is different
<Frosku>I'm in spacemacs
<brendyyn>i switched when I heard Doom was cooler
<brendyyn>still a few annoying things with it. like in the text line in C-x C-f, I can't backspace at all, I have to use ctrl backspace
***sneek_ is now known as sneek
<Frosku>Cannot find suitable directory for output in ‘comp-native-load-path’. -- this is stumping me, I've set it in init.el (which I have no idea if it's even reading tbh)
<Frosku>OK, removing .emacs.d and re-pulling spacemacs works, so I'll just copy config and let it rebuild
<rekado_>cdegroot: reepca whittles away at the C++ daemon, implementing missing bits on the Guile side. There is no rush to complete it, but when it moves it’s moving in the right direction.
<mdevos>I failed to build a GNU/Hurd installation image.
<mdevos>util-linux fails to build on the Hurd
<mdevos> (<>)
<mdevos>Any idea why I can't see i586-gnu builds on <>
<rekado_>mdevos: I can’t see *any* builds by limiting to “system”; I think you need to also add a search term
<rekado_>but for i586-gnu I can’t get it to work either
<smartineng>Hello I've just noticed some weird behavior inside the "guix environment --pure". Whenever I want to correct some commands in the [env]$ shell typing backward on my keyboard the cursor moves forward.
<rekado_>smartineng: I think this behavior is controlled by the TERM variable
<smartineng>btw. on "guix environment --container --pure" it works fine
<smartineng>rekado_: do you know how to fix it, 'export TERM=linux' doesn't help
<rekado_>starfish: hello
<starfish>I installed guix on top of another distro, I have some questions
<rekado_>smartineng: I don’t know. I know that the behavior can *also* be controlled by the terminal emulator (e.g. if you’re using Gnome’s terminal you can configure the behavior of backspace in a profile)
<rekado_>starfish: ask ahead
<starfish>the packages I installed with guix (e.g., mpv), with 'man mpv' I can read the mpv manual. But when I run 'man mpv' from eshell it doesn't work. How do I fix that? Also similarly 'info guix' from CLI works but not from eshell
<mdevos>did you restart emacs?
<mdevos>Maybe you started emacs before "guix install mpv", and now emacs has an obsolete PATH
<smartineng>rekado_: you right it affects termite emulator in my case, thx
<mdevos>In any case, as a work-around, you could run "guix environment --ad-hoc mpv -- mpv" from eshell
<starfish>mdevos: Oh I see. I didn't restart emacs after I installed mpv. I'll try to do that
<rekado_>starfish: a lot of discovery mechanisms are controlled by environment variables
<rekado_>starfish: Guix generates an etc/profile file, which lets Bash set these variables (when the file is sourced)
<rekado_>I don’t know if eshell is at all affected by your .bash_profile — I assume it isn’t.
<starfish>I restarted emacs, it didn't work
<mdevos>... maybe restart the whole login session (sorry for the ‘turn it off and on again’ answer)? Or start emacs from the CLI where "mpv" worked
<rekado_>starfish: when starting emacs from an environment where these variables are set I’d expect eshell to inherit those variables.
<rekado_>for a running eshell session I’m afraid you’d have to set the variables manually
<starfish>So I try to logout and login again?
<mdevos>starfish: yes, that's my suggestion
<rekado_>starfish: before that I’d just start a separate instance of Emacs
<rekado_>…from the session where “man” works, i.e. where the required variables have been set already.
<rekado_>if that *doesn’t* work you don’t need to bother with logout/login
<PotentialUser-17>Hi! Does anyone know how to include exFAT filesystems in the configuration?
<PotentialUser-17>I get an error that the UUID is invalid when I try this (with a valid UUID):
<PotentialUser-17> (mount-point "/data")
<PotentialUser-17> (device (uuid "xxxx-yyyy" 'exfat))
<PotentialUser-17> (type "exfat"))
<starfish>So I tried logging out and logging back in, didn't work. Does it matter that Emacs I'm using not installed from guix?
<starfish>rekado: 'youtube-dl' when run from eshell wasn't working either. 'which youtube-dl' complained that youtube-dl is not found in PATH. I had to add .guix-profile/bin seperately in eshell using eshell's 'addpath' command. But I don't see something like that in case of info and man. :-(
<Frosku>brendyyn: It's working :D
<civodul>PotentialUser-17: hi! if you wrote "xxxx-yyyy" literally, it is indeed an invalid UUID (should be hexadecimal digits)
<civodul>PotentialUser-17: also 'exfat is not recognized in the 'uuid' form
<civodul>but maybe 'dce works for exfat?
<civodul>or just 'fat?
<PotentialUser-17>@civodul I did not. The UUID I used works on Arch Linux.
<PotentialUser-17>Thank, I'll try that!
<civodul>PotentialUser-17: though i think you'll have problems down the road because internally Guix doesn't know how to find exfat file systems by UUID (yet) :-/
<PotentialUser-17>'dce gave the same error, 'fat changed it to UUID not found.
<civodul>yeah :-/
<civodul>if you're curious, this is happening here:
<PotentialUser-17>Ok, thanks. I'll try to find another way to mount it then.
<civodul>there's fat32 and fat16, but maybe exfat is different
<starfish>solved! Emacs looks for info paths in "Info-directory-list" variable.
<rekado_>starfish: the etc/profile file would also contain variables for man (MANPATH) and info (INFOPATH), though this may depend on having an info reader and man itself installed in the same profile.
<nij>Hi! Is %default-channels configurable? I hope to configure it with a more declarative way, instead of managing an extra file `/root/.config/guix/channels.scm`.
<nij>Also, how do I check the value of %default-channels? I evaluate that in a guix repl, but it's unbound.
<rekado_>nij: the module that defines %default-channels must be in scope
<nij>How do I find out the name of the module?
<nij>Any tricks or tips? Or I just have to memorize them.
<rekado_>in the repl I just guessed: ,use (guix channels)
<rekado_>there are some modules that are loaded automatically, but not all of them are.
<rekado_>I wonder if perhaps this should be changed.
<rekado_>I think in some cases the number of re-exported modules has not been updated when it should have been.
<rekado_>so for now I’d resort to searching the code to find where something has been defined.
<BlackMug>hi there
<nij>i see @@..
<nij>lemme try to search it in the source, and see if I can find it independently.
<rekado_>@@ is to access unexported values
<rekado_>it’s better not to use it.
<nij>can I customize %default-channels declaratively in my config.scm?
<nij>oh no @@ = emoji:googly eyes
<rekado_>oh, sorry
<rekado_>I thought you meant (@@ (the module name) the-variable)
<brendyyn>It's necessary to use @@ a lot because so many variables aren't exported
<nij>lol im sorry
<rekado_>there’s also (@ (the module name) the-exported-name), which is for accessing values that are exported but without having to (use-modules (the module name) first.
<rekado_>brendyyn: if it’s “necessary” then those variables really ought to be exported.
<nij>"necessary" is a subjective word xD
<brendyyn>rekado_: should we export all the record type <> variables and the make-... symbols etc... ?
<rekado_>brendyyn: oftentimes, though, the reason for not exporting something is that the mechanism really isn’t meant to be relied upon by external code.
<rekado_>brendyyn: depends on the code in question
<brendyyn>for service record types
<nij>hmm.. so is %default-channels configurable in config.scm?
<rekado_>sometimes we chose to keep the make-* constructors private, so that values of that type can only be generated through other means
<rekado_>nij: %default-channels is merely a list of channels
<rekado_>nij: you don’t need to use it; you could simply provide a different list of channels
<nij>i don't want to use `/root/.config/guix/channels.scm``
<nij>i want to do it declaratively in config.scm
<rekado_>I don’t think you need to use a separate file at all.
<rekado_>what exactly are you trying to affect here?
<nij>that's what i learned from the official manual.. @_@
<rekado_>the behavior of “guix pull” for any user?
<nij>so i have another personal channel
<BlackMug>i have question: since guix save old versions of the software (through generations?) does that mean the old software itself is still exist in the distro or just a record? (i mean by a record to it that it just save software name,version...etc then it will reinstall it again from the repository if roll back requested)
<nij>oh uh for the root user
<rekado_>nij: if you want a different set of channels to be used for packages in the operating-system declaration you’d have to use inferiors (which allow you to specify a list of channels).
<rekado_>I also know that you can specify a system-wide default for a channels *file*, which you can generate from the os config.
<rekado_>but per-user configuration of channels is not something you can do with the os-config as far as I can imagine.
<nij>I will check inferiors out! Thanks :)
<nij>Ah! Actually, currently I fetch packages by using manifest
<nij>Oh no sorry that's not relevant... I thought I can redefine the value of %default-channels in the same manifest file.
<brendyyn>i dont think there is a way to change it without editing the source
<nij>ok.. i will stick with ~/.config/guix/channels.scm for the moment :/
<BlackMug>anybody know the answer to my question?
<brendyyn>BlackMug: yep you can still get the old software, although eventually the substitutes will be deleted, and sometimes upstream sourcecode is deleted or changed too
<brendyyn>but in theory you can go back to a version of guix from 1 year ago and then install all the software that existed at that time specified in guix
<rekado_>nij: you can use channels in the manifest as well.
<rekado_>nij: again with inferiors
<rekado_>nij: instead of just declaring the symbols that are then looked up in the variant of Guix used to read the manifest you would declare the package names and the channels to be used to resolve the packages.
<BlackMug>brendyyn so the software its not really there when its upgraded, guix will save a record for the previous as lets say just data as package name x , date x , version x ..etc but the exact software has been replaced when its upgraded right?
<rekado_>BlackMug: I don’t know what you mean by “record”
<rekado_>whatever is installed via Guix ends up in /gnu/store
<brendyyn>BlackMug: Guix is way more advanced than that.
<rekado_>that directory is a huge cache.
<rekado_>so even if you upgrade packages you never *replace* existing packages.
<nij>rekado_: "Technically, an “inferior” is essentially a separate Guix process connected to your main Guix process through a REPL (see Invoking guix repl)."
<nij>How to connect to *that* main Guix process?
<rekado_>you just add those news things to the big cache and create a new set of links that point to the *new* things instead of the old things.
<rekado_>nij: you are “connected” by default
<rekado_>nij: when you ask for a package in your manifest, by default it will be looked up in the “invoking” Guix.
<BlackMug>rekado_ x package version 1 , new version of x package, upgrading to the new x package version 2 the question what happen to x version 1? is it kept as entire software or removed and just the record of it kept?
<rekado_>BlackMug: x version 1 will still be in /gnu/store
<rekado_>BlackMug: until none of your profile generations refer to it and you run “guix gc” to collect garbage.
<rekado_>only *then* will x version 1 be deleted from /gnu/store and thus no longer be accessible
<nij>rekado_: I'm a bit confused. I use `guix package -m muh-manifest.scm` to get packages.
<nij>But in this process, I'm never prompted into a repl.
<rekado_>nij: correct
<nij>Afaik, it doesn't say how to here either
<rekado_>the channels of the Guix that you use to run “guix package” will be used to look up packages in muh-manifest.scm
***zap1 is now known as zzappie
<rekado_>when using inferiors Guix starts a new “guix repl” process and talks to it to look up packages
<BlackMug>i see, but rise 2 issues bloating the of size and security concerns, for size bloating thats clear but for the security concern doesnt increase the surface attack + vulnerable software will be kept in the distro despite using the latest upgrades?
<rekado_>without inferiors we can simplify things and look them up directly
<rekado_>BlackMug: that’s why “guix gc” exists: to remove stuff you (provably) no longer use.
<nij>OH! So I want to `guix repl`, change %default-channels in it, load manifest.scm in it, and `guix package` in it?
<nij>`guix gc` helps you remove bloated ones
<nij>you can even tell it how much space you want
<rekado_>BlackMug: re security: programs that you don’t use but that reside as files on your disk — that’s not an attack vector
<nij>you can config it so that it does it automatically over a specific period
<rekado_>nij: uhm… I don’t follow.
<brendyyn>* In theory
<nij>rekado_: like, without an inf repl, I'd just `guix package -m manifest.scm`
<nij>how do i do this in an inf repl?
<rekado_>nij: to use different channels you’d declare the channels to use in manifest.scm and then use “lookup-inferior-packages” instead of “specification->package”.
<rekado_>nij: you don’t explicitly interact with the “guix repl” process when using inferiors.
<rekado_>it’s handled for you through the API specified in (guix inferiors)
<nij>OH I see
<nij>thanks :)
<BlackMug>rekado_ can you give more details about security thing, i couldnt catch what do you mean by only resides as files.
<BlackMug>everything in gnu/linux is file no?
<rekado_>if you have a file on your disk that defines a program that would delete your disk when run, but you never run it — is it a problem to have that file?
<Frosku>Isn't that file called rm?
<rekado_>it is
<Frosku>(or dd)
<rekado_>it’s also called dd
<leoprikler>Especially if that file is setuid root.
<Frosku>or fdisk
<rekado_>any discussion of security must specify an attack scenario that you like to protect yourself against
<rekado_>there are few things in security that are static and globally true.
<rekado_>I’ll have lunch now and choose to not eat detergent that happens to be stored under the kitchen sink. Security and safety!
<BlackMug>one question before the scenario, does these old software has same accesses as the new upgraded ones? (internet, writing to disk..etc) or just freezed/contained without any privileges?
<prirai>I'm new to Linux and new to irc, unaware of the rules. If someone might guide me where to refer for things.
<rekado_>prirai: welcome!
<rekado_>prirai: there is no guide.
<nij>hi prirai :)
<prirai>What's the difference between a Linux and GNU Linux? For those who've used both.
<rekado_>prirai: the only rules are to be friendly and to paste big things outside of IRC (link to a paste site instead)
<nij>and also stay on-topic?
<rekado_>ah, sure
<rekado_>prirai: there’s no “GNU Linux”; there’s “GNU/Linux”, which means “the GNU system with the kernel Linux”.
<rekado_>prirai: there is also a fork of the kernel “Linux”, which is called “GNU Linux-libre”
<rekado_>prirai: it’s the same as Linux but with binary (i.e. non-source code) blobs removed.
<prirai>rekado_: Does it mean, a barebones system?
<rekado_>Guix System comes with linux-libre as the default kernel, but Guix is flexible enough to let you use a different variant of Linux.
<brendyyn>if nonfree firmware is added but the linux-libre kernel is used will it be loaded?
<rekado_>prirai: no, it’s not barebones; it just means that if you have hardware components that require the binaries that have been removed, you won’t be able to use these components very well.
<rekado_>brendyyn: no.
<nij>!! linux-libre blocks everything that's unfree, without exception?!
<rekado_>brendyyn: there is a long-standing missing feature in linux-libre that prevents it from loading arbitrary firmware.
<leoprikler>Well the exception is that you need the correct deblob script for the correct kernel and you need to assume it to be correct
<rekado_>nij: no, but vanilla Linux gives us little control over what is and isn’t loaded, and the implementation of linux-libre is somewhat primitive.
<brendyyn>nij: Don't go jumping to conclusions like that. Linux developers haven't made it necessarily trivial to just remove or add proprietary parts. For example I once read there are code blobs embedded in source code some times
<rekado_>it would be nice if there was a more convenient way to enable or disable firmware loading, but Linux doesn’t have it and it’s not even acknowledged as nice-to-have.
<rekado_>so … we’re currently stuck with the way linux-libre implements this feature.
<brendyyn>nij: Most Free Software developers want to create a system that is simple Free by default, not create a system deliberately rigged to make running proprietary software hard.
<rekado_>that’s also why Guix has no arbitrary limits on what you can build and install with it.
<rekado_>it’s just that the Guix project will not “bless” non-free software by distributing it through the project channels.
<nij>i see..
<rekado_>Guix can trivially be extended with third-party channels (maintained by yourself or other people you trust), but our focus for this project is to offer the best Free Software experience.
<rekado_>we don’t go out of our way to sabotage other uses
<rekado_>(but this goes both ways: we also won’t add features to Guix that would *only* benefit proprietary uses).
*rekado_ –> lunch
<mdevos>Example of a channel with free software:
<nij>rekado_: have a great lunch!
<abcdw>hey guix!
<nij>hi :)
<nij>hello! I'm testing my knowledge of pulling a package from a 3rd-party channel (using inferior). Here's my manifest.scm (, and here's the error ( What's going on?
<PotentialUser-17>@civodul thank you for the link with the source. I am trying to add exfat support myself. How would I include a modified 'file-systems.scm' in a build or include the modifications in my config to check if it works as intended?
<mdevos>PotentialUser-17: that is a file in the Guix source code
<leoprikler>nij looks like you're failing to build the Guix derivation for that commit
<mdevos>to use it, you need to modify a "git checkout" of guix
<leoprikler>(note: Guix, not guix-wigust)
<leoprikler>Perhaps you want to target a different one?
<mdevos>and from there run "make && ./pre-inst-env guix system ETCETERA"
<nij>leoprikler: i copied that from a page of the official manual. Let me take that commit away.
<leoprikler>Use "guix describe" to get the commit you're currently using
<mdevos>I would start with "16.1 Building from Git" in the manual
<leoprikler>That is probably likelier to give you a good result
<nij>if i don't specify a commit, it will pull the latest one, right?
<PotentialUser-17>@mdevos Thank you, I'll look at that and try!
<leoprikler>I wouldn't do that.
<leoprikler>Do use a commit, but one that you personally know to be good, so to speak.
<mdevos>don't forget passing "--localstatedir=/var" to ./configure
<nij>oh i see, otherwise it could be unstable, and might even be dangerous
<leoprikler>W.r.t. the manual, perhaps someone should check whether it is still up to date or someone broke the time machine?
<nij>It's taken here
<leoprikler>Exactly, there are dependencies between channels, that may or may not be obvious.
<nij>Btw I've been reading it intensively.. any way that i can report an error quickly and easily?
<leoprikler>Does the example from the manual work as-is?
<raghavgururajan>rekado / rekado_: I'd very much love to. IIUC, its in germany, correct? The situation is, I am currently in Canada as temporary resident, half-way way through to be eligible to apply for Permanent Residency (PR). Now I need to work in Canada for at-least a year, to be fully eligible to apply for PR. Does MDC have branch in Canada?
<leoprikler>If it does not, send a mail to bug-guix.
<rekado_>raghavgururajan: they don’t have a branch, unfortunately.
<nij>any rule? any format? or simply just mention that it doesn't work?
<raghavgururajan>Ah :(
<leoprikler>The more information the better, typically.
<nckx>Good morning Guix.
<nij>no place to open an issue like on github or gitlab?
<nij>gm nckx!
<leoprikler>Accurate information, that is.
<rekado_>raghavgururajan: also, hiring people from the americas is always a little tricky, and there are mechanisms in place that result in preference for people living in the EU.
<leoprikler>No, we're somewhat old-fashioned, we don't use JS forges.
<leoprikler>Mailing lists it is.
<raghavgururajan>nckx: Morning Toby!
<nckx>nij: bug-guix at is that place. is the partially read-only Web view, and oh, it's down, guess I know what to do.
<raghavgururajan>rekado_: I see.
<leoprikler>(Interestingly, the main gitlab instance also handles issues/merge requests via mail, which I find hilarious.)
<rekado_>raghavgururajan: however, work is fluid and we sometimes hire people for project work or collaborations. You can send me things you are interested in via email and I’d check if there are ongoing projects that could involve you.
<nij>nckx: that site is great
<rekado_>nckx: down? Works for me.
<raghavgururajan>rekado_: Sure thing. Thanks a lot. I appreciate it. :)
<nckx>nij: Praise rekado_ for all of it.
<nckx>rekado_: It works now, thank you for fixing 502s with your mind.
<rekado_>especially the many bugs that I implemented to encourage contribution
<nij>NO JS at all - ??!
<rekado_>no JS needed
<rekado_>nckx: you know, I got just a white page as a response just now
<nij>That's brilliant.
<rekado_>nckx: so perhaps it really does need kicking; my mind isn’t strong enough today
*raghavgururajan had glowing ring on this head when he saw 'no JS needed' 😇️
<brendyyn>civodul: The thing you added is pretty awesome.
<nij> has no JS but it still looks awesome aesthetically
<brendyyn>nij: there is sort-table.js there
<nij>oh.. still got a bit of js.. coolio
<nckx>nij: To expand on ‘partially’: you can reply to existing issues but not open new ones. The unauthenticated reply function, mind-bogglingly, has not led to a spam wave the size to wipe out all humankind. I wonder what black magic we have to thank for that.
<rekado_>brendyyn: that’s progressive enhancement; it will all work fine with JS disabled.
<rekado_>nckx: I think it’s just spam-protection by obscurity
<nckx>And 502s.
<rekado_>uhm, … we still only have one process serving all of this. Perhaps we should change that and have a bunch of application servers that nginx can proxy to.
<nij>Why is guiler a bit slow? It took a noticeable amount of time to calculate (+ 1)
<brendyyn>nij: in the repl?
<rekado_>nij: how did you run that?
<brendyyn>time guile -c '(+ 2)' runs in 0.007s for me
<rekado_>nij: Guile is pretty damned fast (on average faster than Python, for example), but performance depends on whether you interpret vs run compiled bytecode; and starting up Guile itself also takes a little bit of time.
<nij>`M-x guiler`
*rekado_ has never used “guiler”
<rekado_>I thought that was a typo
<nij>how do you use guile then?
<rekado_>with geiser
<rekado_>or by running “guile”
<rekado_>or by arranging for “guile” to be run by whatever launcher script the application has
<nij>`M-x geiser` doesn't seem to give me a repl
<nij>there's no prompt, and pressing <ret> doesn't evaluate
<nij>hmm.. lemme learn it properly
<rekado_>it does for me, but you may need more configuration
<rekado_>“M-x run-guile” is also part of geiser.
<rekado_>oh, “guiler” is part of “gud.el”, which is … probably not what you want
<nij>why not? (yeah at least it's very slow)
<nij>hmm the official manual appears to say that a repl should be loaded directly
<nij>maybe my config is fked up xD
<nij>error: " can not find guile implementation"
<mdevos>PotentialUser-17: Did it work?
<civodul>brendyyn: the is going to improve things, it was a longstanding issue!
<civodul>and still is, for non-ELF programs
<nij>`M-x geiser` cannot find guile implementation. But `$ guile` in terminal works.
<rekado_>nij: does “M-x run-guile” work for you/
<mdevos>is there a phase that checks whether (ELF) binaries/libraries in OUT/bin and OUT/lib are compiled for the correct architecture?
<mdevos>If there is, I wonder if it can be taught to verify whether guile .go were compiled for the correct architecture
<nij>rekado_: there's no run-guile on my emacs
<rekado_>BTW: Yesterday night I mentioned something about the GNU vision and what it means to me. The same ideas are expressed more eloquently and more grounded in reality at
<rekado_>nij: it’s part of geiser-guile.el — perhaps something is missing from your Geiser installation?
<pkill9>civodul: what's the thing?
<civodul>mdevos: there's no such check, but i'd say it's rarely useful?
<rekado_>pkill9: it’s a per-application cache of library locations
<mdevos>Actually, it would be very useful I think, when cross-compilin
<rekado_>pkill9: having it speeds up library look-up
<pkill9>ah nice
<mdevos>Search for "CC=gcc" in the guix source code -- most should use "CC=TARGET-gcc" when cross-compiling instead (e.g. using cc-for-target)
<civodul>pkill9: (merged in core-updates)
<mdevos>Also, there are guile libraries that forgot to pass "--target" to "guild compile"
<mdevos>A 'verify-elf-architecture' check would be useful for debugging cross-compilation issues -- by preventing a cross-build from succeeding when the architecture of the binary is incorrect
<mdevos>(I've posted a patch fixing some cross-compilation issues in guile libraries some time ago) (unrelated to "guilx compile --target")
<nij>rekado_: turns out that geiser.el isn't enough. I still needed geiser-guile.el. Now it works! Thanks :D
<mdevos>I'll add 'verify-elf-architecture' to my TODO list. Maybe I can reuse some bits from 'validate-runpath' ...
<nij>Hello! I use `nmtui` for network management on my archmachine. But it's currently not available on the official guix channel. Whatelse would you recommend before I successfully learn to package myself?
<pkill9>it is available nij, i think it's in the network-manager package
<nij>thank you ``pkill -9``! How do I find this out in the future (for similar issues)?
<nij>guix search didn't show me
<pkill9>currently the best you can do is search /gnu/store with fd, which is faster than the find command, and see which package it's in, and hopefully you can find it
<pkill9>but the best solution is to add file lists to packages in the guix data service
<pkill9>so you can look it up there
<raghavgururajan>rekado_: Excellent site! I completely endorse the ideas. But why another site/infra for though? If there are internal conflicts (RMS or anything else), I truly believe that maintainers of *all* gnu projects can come to resolution through conferencing. I think its better to have one unified home for GNU; to avoid fragmentation, conflicts, friction, confusion etc. :)
<raghavgururajan>*for GNU though?
<nij>pkill9: hmmm.. but before i know that the binary `nmtui` is situated in the package `network-manager` how would that help?
<pkill9>what do you mean?
<nij>I wouldn't be able to find out that nmtui belongs to network-manager independently..
<PotentialUser-17>@mdevos I haven't gotten to it yet
<nckx>No, it would (currently) require an external service, that's what pkill9 means by the Guix Data Service.
<tissevert>nij: yes you can search recursively in all the /gnu/store/* for the subdir/file bin/nmtui
<nij>tissevert: before i even get network-manager?
<nckx>tissevert: That's true, but it works only by the rare coincidence that NM is very likely in the store by default.
<nckx>nij: No.
<tissevert>oh, yes, I was assuming that, sorry ^^
<nckx>It's also feasible only on SSDs. 😉
<nij>no worries :)
<nij>oops. got the binary `nmtui`, but it says NetworkManager is not running.
<nij>Is that something for systemd? Or it works on guix system as well?
<tissevert>that's possible if you've disabled the service in shepherd
<tissevert>(so not systemd, we have a custom init system)
<nckx>nij: It's nothing with systemd. I use nmtui exclusively.
<nij>how to enable and start networkmanager?
<nckx>nij: Does ‘sudo herd status’ mention it?
<nckx>And if so under which heading?
<nij>it only shows "networking"
<tissevert>hmmm isn't it supposed to only mention networking
<nckx>Never mind, it's confusing, I remember. There isn't a ‘network-manager’ service.
<nij>under "Started"
<tissevert>yeah I had the same
<nckx>pgrep Networkmanager then? 😛
<nij>nothing shows up
<nij>maybe i should reboot? i just got network-manager for my user
<rekado_>raghavgururajan: we are not in disagreement, and there have been *many* attempts to having this conversation internally, just as there have been equally many attempts to avoid it or block it for the sake of some concept of “unity” that was never there to begin with.
<nckx>nij: <i just got network-manager for my user> Could you elaborate?
<nckx>I use (service network-manager-service-type ...).
<rekado_>raghavgururajan: the assembly provides the infrastructure that is required to have this conversation among equals.
<nij>I put "network-manager" in my manifest.scm, and ran `guix package -m manifest.scm`
<nckx>I also add (file-append network-manager "/bin/nmtui") to my setuid-programs so I don't have to sudo every time.
<nij>That's all I've done
<nckx>Is this Guix System?
<nij>Guix the operating system
<nckx>(That's ‘Guix System’ ☺)
<nij>Yep ☺☺☺
<nckx>nij: Remove it from your manifest and add (service network-manager-service-type) to your services.
<nckx>in your system.scm or whatever you call it.
<nckx>nij: I just meant it's the official name for ‘Guix the operating system’.
<nckx>After reconfiguring your system it should start without a reboot.
<nckx>And ‘[sudo] nmtui’ should connect to it. I don't know if it will properly manage your connections without a reboot, though.
<nij>cool lemme try
<nij>guix system: error: service 'networking' provided more than once
<nij>I don't see why it thinks I've provided it twice..
<raghavgururajan>rekado_: Ah I see. Sorry to hear that.
<nckx>Because something part of %base-services also does. Let me see.
<tissevert>because ti's already in the %deskop-services
<nij>you mean base-services?
<tissevert>hmmm it is called %desktop-services here
<apteryx>civodul: I think I'm almost done with a service for jami, that'd allow us to host live conferences, but I'm hitting some problem where the service hangs shepherd; any idea? The code is here:
<nij>I didn't use %desktop-services, but rather %base-services
<nckx>tissevert: %desktop-services is a superset of %base-services. nij is using %base, not %desktop.
<rekado_>raghavgururajan: I like to see things positively. Showing our faces gives us the opportunity to recognize one another, and to do together what would have been unthinkable alone.
<tissevert>oh, I didn't know about that other one, thank you !
<nckx>nij: Nothing in %base-services provides networking (static-networking doesn't count -- I combine it with NM too). Maybe dhcp-client-service conflicts. I don't have it, I think NM implies it if needed.
<apteryx>The problem comes from the jami-daemon service; running this outside of shepherd it's OK, but for some reason inside shepherd it locks its execution and I must reboot as everything on my system comes to a halt.
<nckx>nij: Comment it out & see.
<nij>nckx: thanks! lemme try that!
<tissevert>so that means I could have NetworkManager handling my connection even if I deactivated GDM and all the graphic stuff ?
<nckx>NM is not a graphical desktop thing.
<nckx>It's intended to run on headless servers if you want to.
<nij>That seems to work! I'm rebooting. See you shortly ^^
<nckx>Zweet. See you!
<nckx>So erm
<nckx>don't install minetest.
<nckx>Time will cease all meaning.
<nckx>I can't delete old worlds though. Has anyone noticed the same?
<apteryx>ah, exec-command is probably ill-chosen; it seems to do exec over the current execution process
<raghavgururajan>rekado_: If you, civodul and others plan to conference about this over jitsi or something, can I join the discussion? I would like to pitch-in and brainstorm. :)
*nckx hands apteryx a fork.
<rekado_>raghavgururajan: the assembly mailing list is open to those who endorse the social contract. Outside of that I don’t think I can persuade myself to make time for discussions on GNU and the way things came to be this way :)
<raghavgururajan>rekado_: Cool!
<nckx>Ew, ‘gnu.’ hist-completes to gnu dot something else in my browser.
*rekado_ clutches pearls
*nckx formats hard drive.
<roptat>hi guix!
<apteryx>ah, I reconfigured and I can still type this. Progress!
<avalenn>hi guix!
<roptat>since there are more people around now than yesterday evening: you can help translating Guix to your native language: can you reach 100% like French and Slovak? :D
*nckx is.
<roptat>I'd love to see more 100% :)
<nckx>I'm at 8% now, and the insane amount of work you must have put in to get to 100 is staggering.
<nckx>Still, I'm aiming to get most if not all done by the 18th. When will they be merged?
<roptat>when ready, pretty close to the release
<nckx>With 3 days to go that's a wide margin of error, but I'll hurry :)
<roptat>to be honest, I didn't start from scratch on that component
<roptat>someone else had started (and managed to get to 100% on a 0.x version)
<mdevos>Can I just start translating, or do I need to register, or be vetted, or something?
<roptat>you need an account on the platform, that's all
<nckx>Any Internet rando can start.
<nckx>Is the actual correct URL that works.
<nckx>But I'll get there, don't put down other work to help.
<roptat>German, Portuguese (Brazil), Spanish and Danish are pretty close to being complete :)
<apteryx>is it not possible to use sleep in a shepherd service? It seems to wreak havoc (can't communicate with shepherd anymore)
<roptat>doesn't that make shepherd itself sleep?
<roptat>I don't think it spawns a process or thread for running services, it's all in the main loop
<apteryx>I'm guessing it blocks everything, but I'd expect it to not sleep forever, as the sleep value I used is 20 s
<raghavgururajan>> * nckx formats hard drive.
*raghavgururajan formats nckx
<roptat>hey no, I need nckx for the translation!
<roptat>efraim, does "stop" work at all?
<apteryx>efraim: oh, so primitive-fork might be key to have it working?
<apteryx>efraim: what happens when the value returned from primitive-fork is non-zero? It exits without doing anything else than returning the pid (in your code) ?
<efraim>stop does work, no clue about the magic in it, to me it seems like it shouldn't work but I've been using it for months
<roptat>ah, I see, start returns the pid, so stop knows which pid to kill
<efraim>now that sounds like something I've heard before
<roptat>fork creates two processes at the same program point, and returns 0 in the child, and the pid of the child in the parent
<roptat>if it's 0, it's the child, so that's where you do the work
<rhou[m]>is it possible to call the gnu-build-system outside of the daemon?
<roptat>if it's not, it's the parent, and you return the pid of the child, so you can kill it with "stop"
<roptat>rhou[m], it's not meant to, but sure, it has a public API you can call from any context
<rhou[m]>roptat: I just thought it might sens to call the same build instructions for development
<roptat>ah, I see
<roptat>not sure exactly how you can make it work properly, you'd have to pass the same keyword arguments as the build system does
<roptat>whereas, you could use --with-sources to use the same package with your development sources
<rhou[m]><roptat "whereas, you could use --with-so"> can I pass a folder to this?
<nij>nckx: nmtui is now working! thanks so much :-)
<roptat>rhou[m], I think so
<roptat>or you could have a local guix.scm you build from, whose source would be something like (local-file (dirname (current-source-file)) #:recursive? #t) (from memory)
<mdevos>you should probably filter out ".git" directories
<roptat>that won't have any impact though
<roptat>we're not computing a sha256 anyway
<rhou[m]><roptat "or you could have a local guix.s"> you mean like an inherited package description?
<roptat>oh if it's a package that already exists in guix, it's easier to use --with-source
<mdevos>.git directories can be large
<roptat>oh sure
<roptat>although, local changes to uncommited files (like having a Makefile) might affect the build
<rhou[m]><roptat "or you could have a local guix.s"> ah you mean by using the daemon?
<roptat>no, I meant a guix.scm file in the source directory
<rhou[m]>but offloading the build to the daemon?
<roptat>with that local file, you'd build with "guix build -f guix.scm"
<rhou[m]>what about the sha?
<roptat>if you use local-file you don't need to declare it
<rhou[m]>ah perfect, will look into it
<rhou[m]>is there any info about this in the manual?
<roptat>I'm not sure what's the magic behind, I think the procedure computes the sha256 in advance, gives it to the daemon, and then the daemon is happy to check it matches
<roptat>not sure, efraim uses that method
<roptat>he demonstrated it at the last guix days
<rhou[m]>was it recorded?
<roptat>I'm afraid it was not recorder
<rhou[m]>no problem, thanks for the inputs anyway!
<wonko7>Hi all!
<wonko7>I'm failing at installing grub
<wonko7>so first things first, /boot isn't a thing? I setup a non encrypted /boot but the kernel & initrd are still going to live in /gnu/store?
<roptat>right, no need for a separate /boot
<roptat>you can encrypt the whole disk
<wonko7>yes, but grub isn't prompting me for a password, just failing at finding the kernel
<wonko7>where should efi be mounted during install? under /boot/efi, or under /mnt/guix/boot/efi?
<wonko7>my current iteration is with the latter
<roptat>during the install, it should be /mnt/boot/efi
<roptat>(well, if you mount the root partition for guix at /mnt, which the guided installer does for you)
<mdevos>For $SCHOOL, we have a ‘groupswork’ (too lazy to look up proper English word), and we need to collaborate. $PERSON suggested $PROPIETARY for video/chat and $PROPRIETARY for live editing on a single document. I suggested Jami and <>. My suggestion was accepted (-:. No complaints so far.
<roptat>great :D
<roptat>I remember all the students I tought to in France were using framapad too for collaboration, and I didn't even have to suggest it ^^
<roptat>wonko7, are you using the guided installer, or are you manually installing guix?
<asdf`>@mdevos I have now pulled guix and modified file-systems.scm, but I don't know where to execute `make`.
<roptat>asdf`, you'll have to first run "./bootstrap" then "./configure --localstatedir=/var --sysconfdir=/etc"
<roptat>then you should be able to run make
<mdevos>asdf: do you mean "git pull" and modify "gnu/SOMETHING/file-systems.scm"?
<wonko7>roptat: manually, I'm giving it another shot
<roptat>wonko7, would you like to share the configuration you wrote, maybe we can help you spot the issue?
<asdf`>roptat: thanks
<asdf`>mdevos: exactly
<roptat>asdf`, of course, in a guix environment: "guix environment guix --pure" should work even on foreign distro
<roptat>asdf`, it's all described here:
<asdf`>I am running guix, but I created a pure guix environment anyway
<roptat>the main drawback of the pure environment is you can't use other stuff anymore, so I'd limit it to running bootstrap and configure
<roptat>then an unpure environment works well
<roptat>and it lets you use your editor, etc
<asdf`>roptat: seems like I skipped the line with `./bootstrap` when I read the page.
<wonko7>roptat: thanks;
<roptat>oh not sure if encrypted btrfs is supported
<wonko7>oh :/
<roptat>well, I never tried, and I know the situation improved lately, so maybe it is
<roptat>but I can't spot anything that's obviously wrong
<nckx>roptat: Is there an option to use a monospace font for Weblate input fields?
<roptat>nckx, good question, I don't know
<nckx>mdevos: Zoetjes.
<nckx>I know I could muck about with my browser, but I'm surprised it's not a prominent option.
<roptat>nckx, there is no mention of "monospace" in the documentation
<asdf`>roptat: I have the environment in a terminal in emacs, so it does not prevent me from doing anything in other frames. But thank you for the heads up! I guess it will be a while till it finishes compiling. I'll be back with more question when it fails or - if it happens to work - I'll have a closer look at Cheers!
<roptat>yay :)
<nckx>roptat: Thanks for looking.
<nckx>I'll live, usually by copying the added indentation from German, a similarly expansionist language.
<apteryx>efraim: I'm still trying to understand the example you provided; so there's a thread with a loop running forever; and there's this other clause where it returns its pid. I'm guessing that shepherd might retry its start proc until it gets a non-zero value for the pid?
<efraim>apteryx: with the fork the child is always inside the loop and the parent gets the PID
<efraim>so it should never get a 0 for the pid
<roptat>nckx, do you have an example where it would be nice to have?
<roptat>I'll attach it to a feature request on their bug tracker :)
*mdevos afk
<nckx>roptat: - I don't know why it looks misaligned, maybe the German was wrong to begin with, but monospace would make it much easier to be consistent.
<nefix>hello! I'm trying to package Zig (, but it's failing at the install phase. It uses cmake-build-system and it seems to try to create a $out/lib directory to copy stuff there. It's returning a permission denied, do you know why is doing that? Should I create the lib directory before? And if so, how could I do this? Thanks!
<roptat>nckx, as a workaround, you can use the browser's console, and set a monospace font (there are two elements you need to set that on)
<rekado_>nefix: it’s that a build system wouldn’t create the target directory, but yes, you can add a build phase that creates the directory first
<nckx>That's what I meant with browser tricks above, but I thought this was supposed to be a more accessible interface.
<rekado_>nefix: something like this: (arguments (#:phases (modify-phases %standard-phases (add-before 'install 'prepare-install (lambda* (#:key outputs #:allow-other-keys) (mkdir-p (string-append (assoc-ref outputs "out") "/lib")))))))
<nckx>roptat: ☝
<roptat>right, it doesn't seem to exist
<nefix>rekado_: thanks! :D
<PotentialUser-28>I used "guix pack -R emacs" to package emacs into a tarball.
<PotentialUser-28>I used "guix pack -R emacs" to package emacs into a tarball.
<PotentialUser-28>I used "guix pack -R emacs" to package emacs into a tarball.
<PotentialUser-28>I used "guix pack -R emacs". But I cannot run sudo in this emac's shell. I get /usr/bin/sudo must be owned by uid 0. In emacs shell and in dired it is owned by 'nobody'. But outside emacs its owned by root. Can someone point me in the right direction?
<nefix>Another quesion: is there a way to cache phases? I've found myself a lots of times having errors in the check / install phase and having to make a change, waiting for the thing to compile, see that I've missed out something and doing it again, so I end up wasting a lot of time re-compling things that succeed without changing the source / inputs
<leoprikler>PotentialUser-28: perhaps this is a result of packing it into a tarball? I don't think you can setuid root stuff that goes into a tarball.
*nckx likes & subscribes.
<nckx>Copy & paste is now completely broken on my Sway. Rebooting would probably fix it but I can't. Next few days will be fun.
<mdevos>nefix: Currently not, but it would be a very useful feature to have!
<PotentialUser-28>leoprikler: Thanks. But still confused. I am trying to run is my foreign distros /usr/bin/sudo. Why would the tar packaged emacs see my foreign distro's file system differenlty.
<mdevos>guix pack -R does some bind mounting I think (not sure)
<mdevos>maybe that's relevant
<nefix>mdevos: oh :( indeed it would be really useful, it would save lot of time!
<leoprikler>nefix w.r.t. zig, there are already packages waiting in the ML, perhaps try those out?
<nefix>leoprikler: I haven't understood anything you just have said xD
<nefix>oooh I see now :)
<leoprikler>It's like my professor once said: "Manchmal erspart einem wochenlanges Implementieren einen Nachmittag in der Bibliothek."
<avalenn>good one leoprikler, I needed a translator but it made me laugh
<nckx>leoprikler: I'm going to steal that.
<wonko7>ok, I got a step further, grub is loading the kernel
<wonko7>then I get a few messages about iwlwifi & bluetooth not loading, and then nothing
<wonko7>so X isn't starting but no errors on that, and crtl+alt+f2 isn't opening a new tty login
<wonko7>am I supposed to add something to config.scm to get a console non x login prompt?
<leoprikler>wonko7: try a config without X first. X sadly has the ability to interfere with the lifetime of consoles
<leoprikler>once you get a terminal/ssh running, you can reconfigure from there
<wonko7>leoprikler: thanks, what should I have in services? nothing?
<leoprikler>nothing, %base-services or %base-services + ssh
<wonko7>do I need to "guix system init" after updating config? or is there something quicker? (booting from usb)
<Noisytoot>What is GNU Assembly?
<Noisytoot>Guix is listed on its software page:
<avalenn>Noisytoot this message resume a bit what it is:
<avalenn>the blog post on the site explains also the history of the idea
<mothacehe>configured cuirass to build core packages for ppcle64 using QEMU, all failing: :(
<civodul>mothacehe: d'oh
<distopico>I think we can create a fork of Guix and have OSI Guix aka and GNU Guix
<asdf`>I have compiled and run the test suite on my local copy of guix with modified file-systems.scm. 25 tests were skipped and there are 2 XFAIL, which is actually good if I understand it correctly. Now I'm not sure what I should replace the etc in `./pre-inst-env guix system ETCETERA` with.
<nckx>distopico: You've got it backwards, I think :) GNU Guix is, I don't know what ‘OSI Guix’ would look like... Guix has always skewed heavily towards the Free-not-Open crowd.
<nckx>Not that they're not welcome, but we tend to aim higher than just open source.
<civodul>mothacehe: looking at, the issue appears to be that the libgc build times out
<civodul>/gnu/store/s5qp9sqk40xm2mv9mxv0mmy3hmvv8rkp-libgc-8.0.4.drv specifically
<lfam>Hey roptat, are you around?
<mothacehe>true, but I also found Guile build timing out
<civodul>i'm reconfiguring my laptop to give it a spin
<civodul>(via emulation)
<civodul>i guess this is gonna take a while...
<mothacehe>9 hours at least, unless your laptop is faster than the build farm :p
<Noisytoot>distopico, Is affiliated with the OSI?
<Noisytoot>A fork of Guix would not be good
<lfam>No, the two groups are not affiliated with in any way
<lfam>With each other
<distopico>not sure, well sounds to me like an OSI initiative, even are taken as example of governance Gnome
<civodul>mothacehe: i'll have substitutes though?
<lfam>Perhaps it sounds that way, but you're mistaken, distopico.
<mothacehe>cbaines: any clue why patchwork test is suddenly failing:
<Noisytoot>Could someone apply
<distopico>Although I think Debian can be a good example of governance (although it lacks a lot, for example having more female maintainers), but I like Debian woman, we should have Gnu Woman
<Noisytoot>distopico, Gender doesn't matter
<distopico>Well, I think that is the problem that many people have against RMS
*Noisytoot supports RMS
<nckx>distopico: You mean GNU? OSI is something else, IMO (so many TLA) not really something worth emulating or improving. Regardless, it already exists. GNU is what we care about, so GNU is what needs to be saved.
<mothacehe>civodul: almost none I fear, the build dependencies are not fetched on the main node for now
<civodul>ah :-/
<civodul>intermediate build results were not retrieved on the head node?
<mothacehe>no only the final derivation is fetched using ensure-path RPC, once the build is successfully completed
<roptat>lfam, I am
<lfam>roptat: I just wanted to let you know, I'm keeping a list of pending patches that are being held due to the string freeze
<roptat>lfam, oh cool, thanks for that!
<distopico>thinks like that claim about non-inclusion in GNU, I don't see anything different of inclusion there only men, no trans, but anyway now the user end needs to understand about FS, Gnu vs vs OSI and so on
<lfam>Well, the idea is that we want to improve the situation. That's why a new group was formed, to avoid the problems of the old group
<roptat>distopico, which is probably the result of the history of GNU
<Noisytoot>lfam, What are the problems with GNU?
<civodul>comrades! this is #guix! :-)
<civodul>there's a #gnuassembly channel if you want to discuss it further
<distopico>s/user end/end user/
<roptat>oh, nice :)
<civodul>(i'm there too :-))
<nckx>I missed Noisytoot's question: < affiliated with the OSI?> No, not at all. Not sure why ‘OSI’ is being mentioned at all. It's a different group from GNU. But this is off-topic for #guix, I suggest going to #gnuassembly to clear up such misunderstandings.
<morgansmith>what's the way I'm supposed to install fonts? Put them in my X11 config? Also any suggestions for fonts? I just want full coverage so I stop seeing the "that's not in your font" symbols.
<sneek>Welcome back morgansmith, you have 1 message!
<sneek>morgansmith, efraim says: I checked the bug, I also don't see my response email to it. Perhaps it got lost in a mail queue, it was supposed to close the bug.
<nckx>lfam says it well. :)
<nckx>morgansmith: Just ‘guix install font-<thing>’ should just work. Possibly followed by ‘fc-cache -rv’ to flush caches.
<nckx>Eeevil stateful caches.
<nckx>morgansmith: But if you want to keep them close to your X11 configuration in /etc, you can add them to your operating-system's packages field too.
<cbaines>mothacehe, regarding the patchwork system test, I don't think I've changed anything recently at least
<cbaines>I'm not sure if it's linked from anywhere, but the Guix Data Service does support looking at system tests through time now, e.g.
<cbaines>unfortunately the build farm I'm using for system tests hasn't got the resources to keep up, and there's not data available from, so there's not much help there
<mothacehe>the failure seems to be cause by memory issue, but once increasing the VM memory, I'm having another error locally
<morgansmith>nckx: Thanks :)
<mothacehe>strange, I'll investigate more tomorrow
<mothacehe>I'd like to have all system tests passing before the release
<mothacehe>I asked danny and rekado_ help on the two other failures: nfs-root-fs and ldap
<morgansmith>Did we ever figure something out for triaging bugs? Looks like 46496 is a duplicate of 47325 which I fixed :) (sorta. It'd be nice to merge the nano and non-nano toolchains like everyone else does)
<morgansmith>so I get that you should never hardlink a directory so you can avoid infinite loops, but why would you ever symlink over doing a hardlink for a file? symlinking in ~/.guix-profile I think makes sense because we want it to be transparent that stuff isn't actually there. But when doing stuff in a read only filesystem like /gnu/store, wouldn't hardlinking files always be better then symlinking?
<roptat>isn't that what we do when deduplicating?
<morgansmith>roptat: I just found that while searching the manual for "hard link". I was just curious because in one of my patches one of my hardlinks was changed to a symlink by the committer
<avalenn>morgansmith with symlinkings you have the notion of reference
<roptat>ok, not sure
<avalenn>this is how you know for example that a derivation needs an other one, GC in Guix make heavy use of symlinks
<morgansmith>avalenn: I'm not really sure what you mean... In operating systems almost everything is some kind of reference?
<morgansmith>I mean symlink vs hardlink just moves the reference from the kernel level down to the filesystem level
<avalenn>when you do hardlink both files are strictly equivalent you can delete one without impact on the other
<avalenn>with symlink you have the main file and the symlink is only a pointer to the other
<morgansmith>avalenn: aren't dependencies bad? Isn't that an argument for how hardlinks would be better?
<avalenn>in Guix you want the dependencies to be explicit
<avalenn>and the unit of managing the dependencies is not the file, it is a group of files called derivation
<efraim>looks like the powerpc64le folks forgot to build mesa :p
<avalenn>if you replace symlinks by hardlinks you could inadvertently delete a whole derivation without knowing it is used by an other
<cbaines>the hardlinks work at the level of individual files within the store
<cbaines>avalenn, I think you're using "derivation" in place of what I'd call a "derivation output" or "store item", I'm not sure what the cannonical term is
<avalenn>cbaines you're right I was wondering about the right term too
<morgansmith>How would replacing symlinks with hardlinks cause the deletion of a derivation output? I get that there are build time dependencies and runtime dependencies, but I don't think you need to rely on symlinks to ensure that your runtime dependencies still exist.
<avalenn>I don't know how to explain it on a simple manner.
<morgansmith>efraim: Why did you change the header link to symlink when you did the nanolib patch? I'm very confused about hardlinks vs symlinks right now :P
<cbaines>on the question of symlinking files in /gnu/store, symlinks are quite different to just having a file there or using hardlinks, so that could and would break things that read files in /gnu/store
<efraim>stylistic reasons? I don't think we use hard links in any of the packages
<cbaines>that's not to say you can't have symlinks in /gnu/store, just that you wouldn't use them in place of hard links for general deduplication
<morgansmith>efraim: Well the static libraries in that patch are still hardlinked :P
<efraim>looks like I didn't touch the other link in the commit and didn't change the commit message
<efraim>git-grep says there are a couple of links but a lot more symlinks
<morgansmith>wait, does the deduplication step actually convert symlinks to hardlinks? The manual makes it sound like it's just identical files that get hardlinked
<avalenn>no, it doesn't convert symlinks
<morgansmith>oh I see. roptat thought that might be what happens and then cbaines said no that would be very bad
<avalenn>symlinks are just files whose content is the name of an other file, they are small they don't need to be deduplicated
<morgansmith>Is it very bad because some applications will be expecting a symlink and fail if they find a file? In my experience usually applications usually are expecting files, not symlinks
<avalenn>kernel resolves symlinks so most application don't see the difference
<morgansmith>I'm not saying hardlinks would be amazing and change everything. I'm sure the benefit to using them would be very small. But I'm still trying to figure out why you wouldn't just hardlink everything to get that small benifit
<asdf`>I have no idea how to continue and test my modifications to add better exfat support. If anyone wants to have a look, I uploaded a diff: I've got to go. Thanks for the help so far. I'll probably check in again in the next days. Cheers!
<morgansmith>efraim: Could you close bug 46496? It's a duplicate of the nanolib one. I'm not sure if debbugs would let me do it but also I would think only comitters should close bugs.
<lfam>morgansmith: Anyone can control the bugs
<lfam>It's just a matter of sending email, as described here:
<morgansmith>lfam: But *should* anyone control the bugs is my question
<lfam>So, you could send a message that includes "close 46496" to <>
<lfam>Yeah, I think it's okay
<lfam>On other platforms, users can open and close bugs too
<morgansmith>I guess that makes sense. But I should probably close it in a way that emails the subscribed people about the fix. Is there a specific format of message I should send that's outline in the gnu standards or something?
<morgansmith>Like I know it'll email the people that the bug is closed, but I want to add an informative message
<bdju>I've changed my keyboard layout in my config.scm and I want to make sure it really worked before my next reboot or else I may have trouble logging in at the TTY. what can I do to make the new layout take effect in a TTY without rebooting? I tried `sudo herd restart term-tty2` and then switching to tty2 and the layout is still wrong.
<morgansmith>when doing a reconfigure it adds a new option to the grub boot. so worst comes to worst, you can pick the second last grub boot option and get back to a known safe state
<bdju>it's more than just inconvenience. I had software-level dvorak and now my keyboard itself is in workman layout, so it essentially types gibberish. I don't have an easy way to make my keyboard qwerty again so that the dvorak at least types normally. I'd have to get an actual different keyboard
<bdju>I changed my actual keyboard since last reboot and there is no safe generation anymore, I haven't quit my sway session since changing layouts and keyboards some months ago
<morgansmith>"Since turning it off and on again does not seem to be a valid solution, we are transferring you to a level 2 technician. Please hold. Your call is important to us."
<morgansmith>I mean there is still an inherent risk, but you could set one of your ttys to autologin
<bdju>hm that's an interesting idea. I would be fine if sway started itself
<lfam>morgansmith: You could send your summary to the bug, and then close it. That way, the subscribers will get your message and the bug will be closed
<morgansmith>Personally, I have a tty autologin and then the end of my .profile is
<morgansmith>if [ "$(tty)" = "/dev/tty7" ]; then
<morgansmith> sx
<bdju>what is sx?
<bdju>does that launch your environment of choice on tty7?
<morgansmith>it's some suckless startx thing. how do you start your gui? Just put your gui start command there
<bdju>I do `exec sway` after I log in
<bdju>not sure how to set up tty autologin, and also I'd be a little nervous to have something like that set up. unless it launched sway and then locked the screen so I still needed my password or something. hmm
<bdju>I'm not in a situation where someone using my machine physically is a big concern, but I still hesitate to make things less secure
<morgansmith>bdju: Ya security is an interesting problem. I've got disk encryption so I just rely on the disk password instead of an account password
<bdju>I almost set up FDE on this install but I had never used it before and was nervous about losing access to my data. a while after I ended up setting up FDE on my pinephone and am a bit more comfortable with it now. I don't think I can add it without doing a reinstall at this point, though.
<morgansmith>Maybe sway has some config file that you can use to launch a screen locker? It'd be annoying when you manually start sway from a tty but ideal in an autologin scenario. Maybe you could make it only launch a screen locker if you're on your autologin tty
<morgansmith>^ for auto sway on login
<nij>While there are guix commandlines, can I invoke them in the repl?
<nij>In particular, I wish to `guix install` a package in the repl.. so I can test if the package is well-formed while developing.
<bdju>morgansmith: thank you for the link! I also just mashed keys to figure out what I have to hit for my user and password in case it is screwed up after a reboot. wrote myself a little translation. so now I should be okay even in the worst-case
<morgansmith>bdju: I wish you luck. This is my autologin service btw (username is a variable for me. please use your own username). The first 6 ttys already have services so I put one on the 7th. I think usually most GUIs are on the 7th anyways so it all works out. (service mingetty-service-type (mingetty-configuration (tty "tty7") (auto-login username)))
<morgansmith>lfam: Thanks. I will do this. The gnu-standards just says to thank them and gives no guidance beyond that so I'll do that too.
<lfam>I think you should feel free to do it as you see fit :)
<mdevos>Are there other translators for Dutch currently on #guix?
<jonsger>mdevos: I think nckx
<mdevos>nckx, do you know how to translate "root file system"?
<mdevos>At <>, I see ‘de wortel van het bestandssysteem’
<mdevos>ik ga "root file system" -> de wortel van het bestandssysteem aan de ‘glossary’ toevoegen
<Noisytoot>Could someone apply
<nij`>Hello! I ran `guix system reconfigure config.scm` as root.. it didn't output anything, but it seems that it's working hard for almost 1 hour. Is this normal?
<sneek>nij`, you have 1 message!
<sneek>nij`, nckx says: Most use a log-in manager. I use (#:auto-login? #t) so I never see it. I don't use the TTYs unless debugging kernel work.
<roptat>nij`, no, there might be an issue somewhere
<roptat>do you use only packages from guix, or do you have your own definitions/other channels?
<roptat>I know guix can be silent when it's busy figuring out a dependency loop (which it can of course not figure out, so at some points it runs out of memory)
<Noisytoot> adds node-once, a dependency of
<lfam>nckx: Check your messages
<nij`>roptat I am using another channel.. it starts from n..
<nij`>And I'm using an inferior to pin down the version of the kernel.
<roptat>oh that's it!
<nij`>4 of my memories are working 100% (due to htop)
<roptat>the inferior is building the kernel silently
<nij`>:-( isn't there a way for me to see what it's doing
<civodul>would be sweet to have a service!
<nij`>it's scary in that i don't even know if it's responsive
<roptat>I've seen that before, but inferiors are too hard for me to understand
***nij` is now known as nij
<roptat>I think the output from the inferior is being ignored
<nij>Will working in a repl make it better?
<nij>I was actually wondering if I can do everything `$ guix xyz ..` in a repl, rather than cli.
<nij>For example, can I `guix system reconfigure 123.scm` in a guix repl?
<nij>I expect the experience will be more native.
<roptat>nij, there was and which seem related
<roptat>not really easier, but you can use the code from (guix scripts system)
<roptat>but since it's the same code that's being run from the CLI, it won't be any different
<nij>You mean (use-modules (guix scripts system)) and then (guix system reconfigure "123.scm")?
<roptat>something like that, yeah
<roptat>you'd need to find the right procedure of course
<morgansmith>I have installed font-gnu-freefont in both my system profile and my user profile. I've run fc-cache -rv. I've restarted my applications. I still cannot view a unicode penguin. I am very sad
<apteryx>are the standard output & error streams in a Shepherd service (start/stop procedures) supposed to end in /var/log/messages? They don't seem to.
<roptat>I don't think they are
<apteryx>I end up having to copy paste the code at the REPL to debug things, which is inconvenient at best
<roptat>need to go, bbl
<nij>morgansmith: 🐧🐧🐧🐧🐧🐧🐧🐧🐧🐧🐧🐧🐧🐧🐧🐧🐧🐧🐧🐧 :-(
<morgansmith>nij: My pain and suffering is unimaginable
<nij>maybe the system doesn't know the right path to fetch to cute unicodes?
<mdevos>Definition of daemon: <>
<mdevos>This seems to apply perfectly to daemons in operating systems
<mdevos>Is anyone opposed to the terminology "waking a daemon" and "bringing a daemon to rest"?
<mdevos>(Replacing "starting a daemon" and "stopping a daemon"?)
<bluekeys>Hi guix.
<morgansmith>mdevos: I think if we changed herd start/stop/restart then there would be riots in the streets. Also you'd have to consider how clear that terminology is in different cultures and different languages. I'd bet that it's worse but I don't know.
<bavier[m]>mdevos: that terminology sounds fun.
<morgansmith>It does sound really fun though. I do agree
<mdevos>sneek: later tell nckx: Tenzij je het ermee oneens bent, ga ik "start a daemon" vertalen als "een daimon wakker maken" en "stop a daemon" met "een daimon naar bed brengen"
<mdevos>or "bringing a daemon to bed"? As in, it is bed time for the daemon? (Not a reference to $ACTIVITIES. Does not have the same associations in Dutch.)
<morgansmith>Ok so qutebrowser will show me the penguin but nothing else will. How the heck do I install a darn font?
<morgansmith>Can I get a confirmation of some fonts that have this penguin 🐧?
<nckx>morgansmith: font-google-noto has it.
<sneek>Welcome back nckx, you have 1 message!
<sneek>nckx, mdevos says: Tenzij je het ermee oneens bent, ga ik "start a daemon" vertalen als "een daimon wakker maken" en "stop a daemon" met "een daimon naar bed brengen"
<nckx>mdevos: ‘onderstoppen’ or gtfo.
<mdevos>nckx, wat is gtfo
<mdevos>een daimon onderstoppen klinkt goed
<nckx>Just a meme I'd best not expand. Consider it ^W'd.
<nckx>I'm looking forward to waking up to a completely butchered Dutch translation tomorrow.
<mdevos>nckx: bij ‘Try adding them to the
<mdevos>@code{initrd-modules} field’
<mdevos>dat zou ik vertalen als ‘probeer ze toe te voegen aan het @code{initrd-modules} veld’. Maar
<mdevos>strikt genomen is dat een samenstelling, dus ...
<mdevos>nevermind, I've found a work-around
<mdevos>‘het veld @code{initrd-modules}’
<nckx>A hyphen as in @code{initrd-modules}-veld is, I think, standaardtaal.
<mdevos>maar je zegt toch wel ‘ik ga de stad Amsterdam bezoeken’, en niet ‘ik ga de Amsterdamstad bezoeken’
<mdevos>dus zou ik ‘het veld @code{initrd-modules}’ gebruiken
<nckx>I think you're right.
<nckx>Let me stop flipping between tabs at high speed and actually think about it for a second.
<nckx>Yes, you're right.
<apteryx>nckx: thanks for being my prime source of entertainment
<mdevos>maar ik zou wel @code{operating-system}-declaratie schrijven. Hmm. Ik denk dat ik het nu doorheb, maar vraag me niet de theorie uit te leggen
<nckx>mdevos: Yep. What you are feeling is entirely normal & healthy.
<nckx>Tomorrow you will see the sun rise, wonder where your 8h of sleep went, why the wall is covered in red string and strangely humid Groeneboekjepages, and why you managed to translate 54 strings.
<mdevos>nckx: I can relate
<mdevos>Locale modifiers are really bonkers. ‘Choose your locale's modifier. The most frequent modifier is euro. It indicates that you want to use Euro as the currency symbol.’
<nckx>In every piece of code that deals with locales there's a clearly visible point where the developer threw up their hands, cried ‘fuck it! it's all random!’, and was Enlightened.
<mdevos>Gewoon strings vertalen gaat min of meer
<nckx>Tekenreeksen my germanic friend.
<mdevos>maar ‘currency symbols’ are a poor fit for locale's
<nckx>Due to an unfortunate incident at my nephew's birthday party two years ago I do not use the word ‘strings’ in Dutch ever again.
<nckx>But I'm veering off-topic in a clear attempt to postpone translating strings...
*nckx → that.
<mdevos>Begrijpelijk. Ik had ook ooit eens zo'n incident. Miscommunicatie ...
<nckx>apteryx: (◍•ᴗ•◍)
<mdevos>weet je hoe "network device" genoemd wordt?
<mdevos>Op eerste zicht zou ik "netwerkapparaat" gebruiken, maar dan hoor ik in gedachten de leerkracht Nederlands strafpunten voor loperwerkwoorden geven ...
<nckx>Hm, I'm not aware of any more colourful commonly-understood synonyms. I used draadloze netwerkadapter for wireless network card elsewhere.
<nckx>I don't work in IT, I don't know what's hep with the jargon.
<mdevos>misschien kan ik "netwerkadapter" gebruiken
<mdevos>dat is meer specifiek dan netwerkapparaat
<nckx>Apart from ‘use English every 3rd word’ which is exactly what I'm trying to avoid.
<mdevos>maar sluit ethernetpoorten niet uit
<nckx>mdevos: That may be good or band.
<mdevos>wat kan goed of slecht zijn
<nckx>If the original means everything on the network, ‘like printers & stuff’. I don't know.
<mdevos>‘ The install process requires Internet access. Please select a network device.’
<mdevos>is ok hier
<nckx>It's perfect.
<nckx>Don't take my responding in English as anything more than laziness to switch languages on the fly. On-topic Dutch is fine in #guix.
<nckx>If anything, there should be more of it.
<mdevos>‘Powering technology’, ‘Waiting for technology ~a to be powered.’. Now that's vague
<nckx>Yes, the English used in the installer is a different flavour from that used elsewhere. Enjoy it? Although in this case I think it's terminology inherited from connman.
<nckx>Technology can be ‘WLAN’, ‘bluetooth’, etc. Powering means, IIRC, ‘connecting’ (everything from rfkill unblock to ‘you can surf cats now’).
<mdevos>What's the difference between ‘Connecting to ~a, please wait.’ and ‘ Connection in progress ’?
<mdevos>is dat niet beide ‘aan het verbinden’?
<mdevos>het enige verschil dat ik zie, is dat ook vermeld wordt met _wat_ verbonden wordt
<nckx>That's it.
<Noisytoot>nckx, Can you surf cats of bluetooth?
<nckx>These are all just strings that someone happened to type whilst coding. You'll encounter ones that differ only in a trailing full stop, or three that mean the same thing.
<nckx>Noisytoot: No, only kittens.
<lfam>"Connection in progress" is kind of weird to me
<lfam>That's not idiomatic, at least in US english
<nckx>I'm guessing this is another installer string?
<mdevos>I'm wondering if ‘Separate /home partition’ should be different, less technical
<Noisytoot>Could someone apply
<nckx>I mean this entirely literally: it's clear that it was almost entirely written (a heroic effort!) by a single non-native speaker.
<mdevos>punt for later
<nckx>s/later/nckx/ sneaky fellow.
<mdevos>(I translated it; it's the change to the source string that is punt)
<nckx>mdevos: Thanks for helping, by the way. I was becoming less & less hopeful of reaching 100% before release.
<mdevos>‘ Please select the file-system type for this partition.’
<nckx>I hope to translate the primary parts of the Web site if time permits, so the both Dutch-speaking GNU/Linux users that don't read everything in English can hit a Dutch download button.
<mdevos>is het ok als ik Please verwijder? ‘Kies een bestandssysteemtype voor deze partitie’
<nckx>That says ‘Download’. I didn't think this through.
<nckx>mdevos: I've always omitted it.
<nckx>French does too, and it sounds odd in Dutch.
<mdevos>‘Doe dit alsjeblieft!’ ‘Als het even kan, zou je ...?’ ‘Ik smeek het je!’
<mdevos>I don't need my computer to beg
<nckx>‘Toe nou. Klik hier. Je weet dat je't wil. Een keertje.’
<nckx>I'd run.
<mdevos>Once or twice is ok and even funny, but all the time?
<nckx>Oh, side note: I've consistently used je/jouw everywhere (which actually goes against my own solidly Flemish instincts).
<nckx>Just in case you're uwing up the place.
<mdevos>nckx: I'm using je/jouw
<mdevos>je/jouw is meer informeel, wat ook past bij de communicatiestijl van guix (op de mailinglijst ten minste)
<mdevos>De eerste keer dat ik ‘Ludo, wat vind je van mijn patch’ (of iets dergelijks) zag ...
<mdevos>eergh partitieterminologie
<mdevos>logical partition --> logische partitie? Hmmm
<nckx>Really? :) Where were you coming from?
<nckx>mdevos: ‘logische partitie’ seems to be the accepted term.
<mdevos>Technically flemish, but with Dutch-speaking Dutch parent and a Walloon parent with a French parent
<mdevos>Ik verwijder de puntjes na ‘Logical partition creation error.’
<nckx>mdevos: I only now see your <nckx, do you know how to translate "root file system"?> question. Yes! Feel free to add things to the glossary. The fewer English loan words (within reason) the better IMO.
<mdevos>-> ‘Fout bij het aanmaken van een logische partitie’
<nckx>Sounds as good as translated computer Dutch can sound IMO.
<nckx>I use ‘mislukt’ as well.
<mdevos>wist je dat je ‘verboden termen’ kunt toevoegen aan de glossary
<mdevos>encryption -> encryptie: Verboden! Uitleg: anglicisme
<nckx>mdevos: I just wondered if you were used to formal mailing lists (is there such a thing? Please senpai, consider my patch this month), but nice to learn about your background :)
<nckx>Wist ik niet, bedankt voor de tip.
<mdevos>ik ken geen formele maillijsten
<nckx>‘Genereren’ heb ik tot dusver kunnen vermijden.
<mdevos>maar wel op school geleerd dat je telkens u en uw moet schrijven in mails, en ook zulke aansprekingen
<nckx>Thought it might be more clear when used next to ‘generaties’; we'll see.
<nckx>I still write U when I'm in a bad mood, and I'm not even old (they insisted).
<mdevos>‘U’: de hoofdletter U? Dat is oud.
<mdevos>‘ Please enter the encrypted label’
<mdevos>labels aren't encrypted I think?
<nckx>I agree. Lemme see.
<mdevos>‘Voer het label van de versluitelde schijf in.’?
<nckx>It's good not to assume all source strings make sense.
<mdevos>Ik zet het op 'suggest'
<mdevos>to samenstellen, or not to samenstellen? partitiegrootte/grootte van partitite, partitienaam/ naam van partitie
<Ikosit>Is there still an qt4 package?
<lfam>Ikosit: No, it was removed a couple weeks ago
<mdevos>Hmm... ‘mounting point’
<mdevos> Please enter the desired mounting point for this partition. Leave this field empty if you don't want to set a mounting point.
<lfam>Ikosit: However, you can find it in the guix-past channel:
<mdevos>mount wordt vertaald met aankoppelen
<nckx>Another case where the English needs fixing too. ‘koppelpunt’ is what I was going to suggest. If you think ‘meh’, I do agree, nothing better comes to mind at once.
<mdevos>Bij ‘koppelpunt’ denk ik aan koppeltjes
<mdevos>koppelen is symmetrisch
<mdevos>aankoppelen is asymmetrisch
<Ikosit>lfam: I want to package qucs, which requires qt@4 and wanted to push it to the guix repo
<lfam>Ikosit: When I removed Qt 4, I also removed qucs
<nckx>I prefer using ‘van’ in your previous question but it's subjective (I'd type partitienaam, then see 4-syllable partitiegrootte, and go back & change both).
<Ikosit>So it existed
<lfam>Qt 4 was abandoned upstream more than 5 years ago, and we won't be reintroducing it in Guix
<lfam>Yes, there were packages for qucs and quc-s
<lfam>I recommend taking those packages from our Git history, and adding them to guix-past
<nckx>mdevos: I guess that makes sense. I certainly don't have a better suggestion.
<nckx>I'm going to be a bit less active in this channel in a vain attempt to reduce my task-switching overhead but I'll read it every few minutes.
<mdevos>‘Weet je zeker dat je alles wat op deze schijf staat wilt wissen?’
<lfam>Ikosit: It would be great to add the packages to guix-past. You are not the only person who has missed them
<Ikosit>lfam: All of them?
<mdevos>I dislike the practice of dropping t from je [...] wilt [infinitief]?
<lfam>Ikosit: Well, the ones that you asked about: qucs
<Ikosit>lfam: I'll put it on my todo list
<mdevos>Errr, ‘ Delete disk’
<lfam>Ikosit: The other packages that were removed in that commit are probably not as widely used
<mdevos>kan de installer Guix Europe opbellen om iemand te sturen die mijn harde schijf kan vernietigen of zo?
<nckx>mdevos: We'll miss these charming installer strings one day.
<mdevos>nckx: ?
<nckx>(‘Delete’ from the list? I don't know.)
<nckx>mdevos: Was the ‘t’ in reference to one of my translations?
<mdevos>‘Guided partitioning’. Geassisteerde partitionering???
<mdevos>nckx: niet dat ik weet
<mdevos>soms past "als je wil" wel. Maar ik hou van de t
<mdevos>ik heb geen idee hoe ‘partitioning’ te vertalen
<nckx>Because I'd totally write ‘je wil [inf]’ but I don't think I did yet.
<mdevos>partitioneren, partionatie?
<mdevos>Handmatige configuratie van partities.
<nckx>I'd go for something like ‘verdelen in partities’ and avoid partitioneren like the plague.
<mdevos>dat is goed
<nckx>Want vandaag is je schijfje een taart, en iedereen krijgt een partje. Feest!
<mdevos>mount point -> aankoppelpunt klinkt nu slecht
<mdevos>(point -> punt)
<mdevos>Maar er is wel zoiets als het drielandenpunt?
<mdevos>ik laat het nu maar
<mdevos> <>
<mdevos>t'is allebei goed
<mdevos>bootstrap binary --> ???
<mdevos>‘ Raw build system with direct store access’. Raw -> ruw, rauw, basis, ???
<mdevos>nckx, wat vind je van command -> roepwoord? Beetje een neologisme weliswaar
<nckx>Bijzonder neo. Ik zou't eerder bij de meer conventionele ‘opdracht’ houden.
<mdevos>maar bij opdracht denk ik aan $LEERKRACHT geeft een opdracht aan $LEERLING
<mdevos>Hier, bepaal deze integralen maar.
<nckx>‘Opdrachtprompt’ e.d. 't Is in ieder geval ingeburgerd. En hoe behandel jij je PC dan? Als een gelijke?
<nckx>(Niet dat ik ijver voor ‘prompt’, was maar 'n voorbeeld uit de wereld om ons heen.)
<mdevos>dat is waar ... Ik verwees wel niet naar de hoger > lager relatie
<mdevos>hé, dat klopt eigenlijk wel. Ik geef mijn computer de opdracht iets te doen voor mij
<mdevos>je zult een aantal vertalingen moeten aanpassen vrees ik
<mdevos>voltooid deelwoord updaten: geüpdatet
<nckx>Don't get me wrong: ‘roepwoord’ reminds me of such legends where knowing someone's name gives you power over them, so it's hella appealing.
<mdevos>nckx: that's also what I was thinking of
<mdevos>Ik ken alle roepwoorden van mijn computer! Ik ben heer en meester over mijn computer! Mwuhahaha
<nckx>Not sure I want my computer to know what I shout at it most.
<nij>Hello! My guix system does not play any sound.. even though in pulsemixer it shows it ain't muted. I have (service pulseaudio-service-type) in config.scm (which is run reconfigured on).. what could be going on?
<Noisytoot>Could someone apply
<nckx>Hi nij.
<mdevos>Ik moet er nu wel mee ophouden voor vandaag. Doei!
<nckx>Alvast extreem megahard bedankt echt waar.
<roptat>I reached 50 percent of the cookbook today, just finished translating the packaging tutorial :)
<nij>nckx: hi :)
<roptat>I'll stop there, hopefully I can finish it by the release
<nij>Mee ophouden vandaag Doei, er nu Ik moet.
<nckx>Noisytoot: I forgot to answer when you posted it half an hour ago but my telepathic answer remains the same: I'll merge it in two weeks (those two words occur in that order in the manual so it must be right), but really hope someone who actually uses Node does so before.
<Noisytoot>nckx, Is it 2 weeks from now, or 2 weeks from when I posted it to guix-patches?
<nckx>nij: It's a total guess, but does /etc/asound.conf exist on your system?
<nckx>From posting, so 9 days, but don't shoot me if it's more.
<nckx>Actually don't shoot me like ever thanks.
<nij>nckx: no there's not /etc/asound*
<jeko>Hi 😅️
<nij>hi jeko
<nckx>Still from the ‘let's see if this flies’ school of debugging: try adding (service alsa-service-type) to your system services.
<nckx>I see some Pulse compatibility options in the file it creates.
<nckx>nij: ☝
<nckx>jeko: guix install hello
<nij>nckx: lemme try
<nij>(nckx you've been saving my lives so much recently! Thanks a lot ^^)
<nij>It made my guix experience much more enjoyable than if you weren't there.
<nckx>nij: If that does work, and you'd have the time to report a bug against the manual (or even the service -- I'm not sure if it should hard-depend on ALSA or not), I'd be double chuffed.
<nckx>nij: (◍•ᴗ•◍) ...thank you...
<nckx>I'm happy to help, but let's make it the last time this issue needs it.
<nij>which bug do you mean?
<nckx>That pulse silently (heh) fails if you don't add alsa.
<nckx>At least that's my current hypothesis.
<nij>oh hmm ok no i did that, rebooted, saw /etc.asound.conf, but still no sound @@
<nij>@@ = googly eyes
<nckx>‘It works here.’ I wonder why...
<nij>wait, it soundly has sound!
*nckx doesn't use %desktop-services just in case you were wondering.
<nckx>I don't know what that is but its sounds very soundy.
<nij>It has happened before.. weirdly. A few months ago when I played with guix, the sound sometimes worked, but sometimes not.
<nckx>What doesn't?
<nij>sound.. sometimes it makes sound but sometimes its broken
<nckx>So every programme you can think of plays sound now?
<nckx>That's going to be fun to debug :-/
<nij>yeah hm
<lfam>Sound problems?
<nij>lemme keep on with this! maybe it was just loading slowly
<nij>dadadadada lfam
<lfam>I fixed sound issues using Guix programs on other distros, in recent months
<nckx>This is Guix System but that's great news too!
<nckx>Maybe it's related after all.
<nij>lemme reboot and test again
<nckx>I've a similar (but different-sounding) problem when I disconnect my Bluetooth headphones: the Pulse audio sink actually *blocks*, so anything that tries to play sound just waits. Video players, browsers, ... freeze on the first frame. When I reconnect the 'phones they start playing again.
<nckx>It's odd.
<jeko>I am trying to create a Guix System definition with an Emacs init.el file (inside a .emacs.d directory) ready to use so I tried to set an extra-special-file but the resulting file inside the store is write protected so Emacs can't configure itself
<nckx>Also relatively new (less than a year).
<jeko>do you know other alternatives?
<jeko>well, I am actually trying with skeletons… just remembered about it after I sent my message haha
<jeko>(irc duck…)
<nij>nckx: Rebooted, launched my browser and youtube, no sound.
<nij>AS last time, I turned on `pavucontrol`, which makes sound.
<nij>THe browser is still silent. Turned it off and back on again. Now it has sound.
<nij>I remember a few months ago similar things happened too. It seems that pavucontrol did invoke some sound related setting.
<nckx>Hm. ‘Turn it off & on again’ is supposed to be a joke.
<nij>didn't get it 😅
<nckx>It's not good that it's the actual fix.
<nij>I will reboot again and wait for 5 min.
<nij>If the same thing happened, I can be sure pavucontrol has some magic power.
<nckx>nij: I'm glad you have a relatively easy work-around, but would you mind filing a bug? Does pavucontrol invoked from the terminal the first time after boot print anything?
<nckx>Good plan.
<lfam>nckx: That sounds *exactly* like the bug I fixed
<lfam>The bug being that applications couldn't find the ALSA plugin for using pulseaudio
<lfam>It meant that only one application could use sound at a time
<lfam>Well, not exactly
<lfam>But similar :)
<nij>lfam: the bug that's bugging me now?
<lfam>I don't know the details of your problem nij
<nij>!!! Same thing happened again after waiting for 3 min.
<nij>lfam: As I logged into my user, the sound isn't available until I invoke `pavucontrol`.
<lfam>Before you do that, is pulseaudio running?
<nij>I've waited for three minutes this time. No sound in the beginning either. But immediately after pavucontrol is launched, problem solved.
<lfam>`ps aux | grep pulseaudio`
<nij>lemme reboot again and find it out
<philipmcgrath[m]>When refactoring some packages into a new file: (1) Should you copy the whole header from the old file, or start fresh? (2) Does the old file need to reexport the packages or something for compatability?
<lfam>philipmcgrath[m]: You'll need to move the relevant copyright lines to the new file
<lfam>Assuming that's the "header" you meant
<philipmcgrath[m]>Yes, though I don't know exactly which ones are "relevant."
<lfam>In terms of exporting variables, no, there are no compatibility workarounds to perform
<nij>lfam: just rebooted. `ps aux | grep pulseaudio` => grep --color=auto pulseaudio
<lfam>nij: So, that's a "no"
<nij>?! oh really i thought that means its on?!
<lfam>It's showing you your grep command
<lfam>Not a pulseaudio daemon
<nij>sorry the output is actually
<nij>nij 368 0.0 ...... S+ 17:41 0:00 grep --color=auto pulseaudio
<nij>oh! this is grep
<lfam>philipmcgrath[m]: It's annoying, but you can use `git log` and `git blame` to see who owns copyright on which parts of the file, and then re-do the headers appropriately. It's a good reason to not do this kind of refactoring work, IMO :)
<nij>now lemme invoke pavucontrol
<philipmcgrath[m]>lfam: Makes sense :)
<nij>Errr now pulseaudio is on!
<nij>lfam: so what do you think this problem is @@?
<lfam>nij: Basically, any program that makes sound should start pulseaudio on demand. If it doesn't, then you'll need to install the ALSA / Pulseaudio compatibility plugins
<nij>plugins .. to that app?
<lfam>So, `guix install alsa-plugins alsa-plugins:pulseaudio`
<lfam>No, they are plugins to ALSA
<nij>i see lemme try
<lfam>Which programs are you trying to use, that don't work?
<lfam>Qutebrowser is qtwebengine, which depends on pulseaudio, so I'd expect it to start the pulseaudio daemon on-demand
<nij>hmm ;(
<lfam>I would call it a bug in qutebrowser / qtwebengine
<nij>lemme see if alsa-plugins solve the issue
<nij>got it, rebooted, launched qb, still no pulseaudio
<nij>turned pavucontrol on, now sound is available, pulseaudio on
<lfam>That makes sense
<nij>kill pavucontrol, pulseaudio still on, but sound off from qb!
<nij>very weird
<lfam>Pulseaudio is typically started "on demand". So, pavucontrol started it
<lfam>I guess that something is wrong with qutebrowser
<lfam>If you start qutebrowser from the terminal, and try to play sound, does it print any warnings or errors about this?
<nij>good idea lemme reboot and try again
<lfam>You don't need to reboot
<lfam>You can just quit pulseaudio
<nij>ALSA lib pcm_dmix.c ... unable to open slave
<nij>PcmOpen: default, No such file or directory
<lfam>Please share the entire error messages on <>
<lfam>Hm, too bad it doesn't say which file or directory it is looking for
<jeko>Can I use skeletons for root ?
<cj>Helo! I would like to know if there are any disadvantages to using Guix compared to Debian. In Guix you can use old libraries for compatibility, but wouldn't that open a security hole?
<jeko>seems I can't but maybe I miss something
<lfam>nij: Well, I googled "qutebrowser pulseaudio" and there are a lot of discussions about it. Maybe you can read those discussions and find a fix
<lfam>cj: What do you mean, "use old libraries for compatibility"?
<nij>hmm never happened on my arch machine, but sure i'll do that
<jeko>cj: you can use guix package manager inside a debian distribution
<lfam>nij: Yeah, but you still might learn what's going on.
<lfam>"Googling" is what I would do, anyways
<cj>A recent software that only has compatibility with an old library and with vulnerabilities. In Debian the libraries are updated and corrected.
<nij>lfam: yeah.. sorry i didn't realize it's a problem of qb.
<lfam>cj: Guix does security updates:
<lfam>If you have a question, I'll do my best to answer it
<jeko>cj: actually, Guix provides me with more up to date package versions as my Ubuntu does.
<jeko>cj: depends on the software you use I guess
<cj>Yes, I understand that libraries are more up-to-date, but I was wondering if it allows using old and vulnerable libraries. Sorry I am unable to follow the discussion normally, it is because I am using a translator.