IRC channel logs


back to list of logs

***stikonas_ is now known as stikonas
<leoprikler>is there any reason `guix gc -D <path>` takes ages deleting links?
<brettgilio>leoprikler: if I had to guess it takes a long time checking canonical paths for dead objects and can be more exhaustive than regular GC
<brettgilio>are you on an SSD?
<leoprikler>of course not
<leoprikler>but I passed a direct store path that was immediately deleted
<brettgilio>then that is a factor too. disk spinning is going to slow down this process by a lot
<leoprikler>it gives me some "updating windows" vibes in terms of how slow it is, though ;)
<leoprikler>especially since i've not asked guix gc to optimize anything and yet it writes:
<leoprikler>[wait for it, I have to scroll back a little]
<leoprikler>deleting `/gnu/store/trash'
<leoprikler>deleting unused links...
<leoprikler>note: currently hard linking saves 71613.57 MiB
<leoprikler>I can maybe understand cleaning the trash, but I don't expect searching for dead links on -D
<xelxebar>Tried groveling around in guix/scripts/gc.scm but can't really figure out what -D ends up calling. Also, grepping for the string "unused" finds nothing under guix/, so I'm confused.
<xelxebar>Looks like it boils down to some guix/store.scm(run-gc) call, but that procedure is a bit gnarly.
***catonano_ is now known as catonano
***KE0VVT_ is now known as KE0VVT
<raghavgururajan>Hello Guix!
<apteryx>hello raghav!
***zender_ is now known as zender
***ezzzc675 is now known as ezzzc67
<raghavgururajan>Oh hey, is back online. 🕺
<ryanprior>I'm realizing that I'm going to be chasing my tail for some time packaging Hugo
<ryanprior>I started 5 months ago, and most of that time I haven't been working on it daily, maybe 10 days out of those 5 months I've touched it.
<ryanprior>But in that amount of time the dependencies have changed significantly, various new ones added, some that I've already packaged removed.
<ryanprior>So now I need to go through and compare so that I don't waste effort, and update my plan of attack to account for the new additions =D
<ryanprior>And I should try to finish so that I don't let another 5 months go by with another round of dependency changes!
<xelxebar>ryanprior: The pain is real. Had a similar issue packaging some TeX stuff.
<xelxebar>Unfortunately, that eventually ran into a bug in the texlive package itself and so stalled out.
<ZombieChicken>how large is texlive anyways?
<Brendan[m]2>Some news:
<ZombieChicken>Amnesia: The Dark Descent?
<Brendan[m]2>yep, im a little confused by the article. I'm not sure if they have freed all non functional data with it, just the source code.
<ZombieChicken>I'm just reminded of how broken Amnesia is. The speedrun of it is... something
<Brendan[m]2>Looks like it actually depends on a proprietary component FBX SDK . but that may be replacable
<Brendan[m]2>and their are no assets, so its basically nonfunctional xD
<efraim>which error is 139?
***iyzsong-- is now known as iyzsong-w
<raghavgururajan>efraim, do you know where to find the public key of bayfront?
<raghavgururajan>to download substitutes
<efraim>was about to ask for guix publish or for ssh
<efraim>it should be in etc/substitutes in the git repo, let me check my notes, it may use the same key as berlin
<efraim>it doesn't seem to be in the repo and it isn't the same as berlin. the canonical place to get it of course is /etc/guix/ from bayfront but I can upload a copy if you want
<raghavgururajan>efraim: Thank you!
<civodul>Hello Guix!
<raghavgururajan>civodul, o/
<janneke>Hello civodul!
<Brendan[m]2>Why does magit reverse the order of commits when rebasing? everything else like logging has the most recent commit at the top, then suddenly when rebasing it's all backwards.
<efraim>I need a bash for binutils-final on powerpc
<civodul>Brendan[m]2: i think it's really "git rebase -i" under the hood
<Brendan[m]2>ok well why does git do it backwards then ;)?
<jlicht>Brendan[m]2: because you are looking at a 'plan', not a history. It is still a bit confusing though
<raghavgururajan>civodul: is back online. Could change it back on the #guix's subject line?
<raghavgururajan>*could you
***ChanServ sets mode: +o civodul
***civodul changes topic to 'GNU Guix | get Guix at | videos: | bugs & patches: | paste: | Guix in high-performance computing: | This channel's logged: | 1.1.0 is out!'
<civodul>done :-)
***amfl_ is now known as amfl
<civodul>apteryx: fun, i had been investigating the failure on and off without success :-)
<civodul>(just saw your commit)
<Lurkki[m]>Does anyone have a tutorial or config file for running guix on raspberry pi?
<civodul>Lurkki[m]: i don't have that but maybe you can take a look at
<civodul>it contains an intro on how to get Guix System running on ARMv7
<Brendan[m]2>omg im so happy i just got past all these errors and made this service i made work!
<civodul>heh well done, Brendan[m]2 :-)
<Brendan[m]2>I think there is no refresh-er for programs hosted on was anyone planning to add that?
*rotty is trying to deploy a guix install to the (emulated) HDD of a hoster (hetzner), and is stuck with "waiting for partition "6d9ee61d-...-ba8b-3b1d8ccbb548" to appear"
<rotty>a partition with that UUID exists, according to `bklid` run from the rescue system
<guix-vits>rotty: IDK, but: is there any 'device-mapper' things in the setup?
<roptat>hi guix!
<mroh>Hi roptat!
<civodul>hey rotty!
<civodul>rotty: could it be that your initrd lacks a driver for the HDD?
<rotty>civodul: ¡hola!
<rotty>I think it is due to my VM using IDE emulation, while hetzner uses virtio
<civodul>rotty: virtio modules are part of the default initrd:
<bdju> nearly been a month since the last post on this. anyone have a clue why the problem mentioned at the end is happening?
<civodul>bdju: no idea, but it looks like these patches fell through the cracks!
<civodul>Brendan[m]2: this is from you, right? ↑
<mfg>Hi o/
<mfg>had to reboot sry
<mfg>does someone use sway with swaybg?
<bdju>I do
<mfg>nice! it doesn't load any backgorund image here... i get "couldn't recognize file format" for every format which has a loader in gdk-pixbuf installed... is there anything special to do?
<bdju>the lines I have are just like `output monitor-name bg /path/to/picture.jpg fill`
<bdju>and I have two, one for each monitor
<bdju>I didn't have to do anything besides just add those lines to the sway config
<bdju>I have swayidle and swaylock in my manifest file and sway itself in my system package list. not sure if any of that would affect wallpapers working
<mfg>i'll see if i get it to work
<mfg>and you don't have the swaybg package installed, don't you?
<apteryx>civodul: some network things sometimes behave differently in the build container compared to --container --network, even they only access the loopback interface
<apteryx>it's probaby good enough for most uses and lacking some bits that stuff like the requires to work correctly, I don't know.
<apteryx>civodul: the next big failure for me on core-updates is mesa. I don't yet have a clue as to what it might be caused by.
<guix-vits>Lurkki[m]: Hello. Any successes with Raspberry?
<civodul>apteryx: ok
<civodul>apteryx: i won't look into Mesa but i'd still like to understand that issue
<civodul>esp. since it work outside the build env
<apteryx>I'd like too! There are a couple of things that fail strictly in the build environment if you grep the code base.
<civodul>yeah, that usually falls into one of two classes: uses of undeclared deps, or networking issues
<Lurkki[m]>guix-vits: i won't be trying it now since i can't find any information on it, maybe i'll try later
<civodul>i also noticed there are many tests being skipped because of a missing dependency on socat
<apteryx>Perhaps will magically work with socat present?
<civodul>i hoped so, but no :-)
<apteryx>probably not, because socat was not present in a 'guix build -C --network' env
<apteryx>where it works
<civodul>i'm trying things in the background, we'll see...
<apteryx>ok! good luck :-)
<bdju>mfg: I don't have a swaybg package specifically anywhere that I see
<mfg>that's really interesting, because since verison 1.1.<smth> sway uses swaybg to set the background. i can also see how sway invokes it but that's exactly the point where the error occurs
<apteryx>how do I get the size of a system closure?
<apteryx>I guess 'guix system build os.scm', then 'guix size /gnu/store/os-profile' ?
<apteryx>I think it'd be nice to have our default templates know how to mount NTFS and NFS file systems as simple users if ntfs-3g and nfs-utils don't add much to the closure size.
<apteryx>this would resolve #39670.
<civodul>rekado: can we reboot build node .158? its guix-daemon cannot be started due (indirectly) to the fact that PID 1 is too old
<civodul>"no binding `%precious-signals' in module (shepherd service)"
<mfg>bdju: sigh, just noticed that swaybg is a dependecy of sway so you have it :D and i don't need to specify it manually
<mfg>So it seems to be a bug in gdk-pixbuf 2.40.0, rebuilding swaybg without support for it works
<mfg>No, my environment was just messed up :| i hate it...
<mfg>when does guix add the LIBRARY_PATH variable to a environments /etc/profile file? i know that it only adds vars which are used by the packages in the profile
<mfg>but i don't understand how exactly it works
<leoprikler>mfg: look at [native-]search-path
<mfg>will do
<leoprikler>if there's a package, that has it [either directly or through propagation], it will set those paths accordingly
<mfg>now it makes sense, thank you!
<civodul>janneke: i just looked at the image types patch by mothacehe:
<civodul>perhaps you'll want to take a look too
<civodul>it looks very nice, definitely an improvement over what we have
<civodul>today's patch review session led me to apply 2-month old patches
<civodul>challenge to y'all: can you find older patches to apply? :-)
*raghavgururajan peeps in
<mothacehe>civodul: thanks a lot for having a look, glad you like it :)
<sneek>Welcome back mothacehe, you have 1 message!
<sneek>mothacehe, zimoun says: is awesome ! Thanks!
<zacts>does guix support macppc 32 bit?
<efraim>zacts: almost! I have a WIP branch that I'm testing on core-updates
<efraim>unsuprisingly building packages is slow
<zacts>oh nice!
<zacts>efraim: are you crossbuilding the packages, or are you building them natively?
<efraim>building them natively
<zacts>cool. I'm wondering if qemu could be used for this
<zacts>like cross building within qemu-macppc on-top of x86_64 to speed up the process
<efraim>I'm using an ibook G4 with 1.5 GB of ram and the original HDD. one of these days I'm going to tear it open and put in the m.2 disk in an ide adapter I bought months ago
<zacts>efraim: nice
<zacts>I'd love to get a gnu/linux distro to run on an iMac G3 with up-to-date packages
<efraim>how much ram can that machine take?
<zacts>efraim: I can't remember. I don't currently still own my machine, but I've considered getting one.
<zacts>I want to use it mainly for typing in vim or emacs on console mode.
<zacts>kind of like a typewriter
<zacts>I just think really well that way with no distractions.
<zacts>for writing
<zacts>the last I used my iMac G3 with gnu/linux it had issues with audio / video due to lack of ram
<zacts>so I couldn't even play youtube
<zacts>(from what I remember that is)
<roptat>btw guix can use qemu transparently for compiling for another architecture, with no cross-building involved (except for the bootstrap binaries of course)
<zacts>oh nice
<roptat>see qemu-binfmt in the manual :)
<zacts>thanks roptat
<nly>how to disable guix for a user? or limit a user on the system, disk, cpu usage, some packages?
<apteryx>civodul: seen while building a system on a recent master:
<apteryx>would anyone know what most file managers such as Thunar use to auto mount file systems? udisks?
<bdju>is there a way to specify an older package version in my manifest if I want to downgrade, or is it not that simple?
<apteryx>yes, using an inferior.
<bdju>I tried the thing with an @ and that didn't work. not sure what an inferior is
<apteryx>see "info '(guix) Inferiors'"
<apteryx>basically it'll allow you to use a different version of Guix for that package
<bdju>thanks. it looks like kind of a pain so not sure if I'll do it or not. but now I know.
<joshuaBPMan>Hey guixoids!
<str1ngs>sneek: later tell jenneke. Hello, I pushed a fix to emacsy where the console no longer spams a backtrace when using C-g . see .
<sneek>Got it.
<rhou[m]>what is the status of the x86_64-w64-mingw32 cross compilation target?
<civodul>apteryx: re (shepherd service), that's OK: it's because we autoload it and it's not available at compile time
<civodul>str1ngs: i think you should have written "janneke" without a period
<str1ngs>civodul: ah thanks!
<str1ngs>sneek: later tell jenneke Hello, I pushed a fix to emacsy where the console no longer spams a backtrace when using C-g . see .
<sneek>Will do.
<rekado>civodul: I’ll reboot .158 now
<rekado>hmm, it hasn’t returned to life yet
<rekado>ah, there it is
<rekado>civodul: don’t you also have root access on berlin?
<rekado>root can log on to all local build nodes
<civodul>rekado: yes i do! the other day i had forgotten i could log on local nodes that way
<civodul>i've uncommented it from machines.scm
<rekado>what annoys me about online conferences or meetings is that it’s difficult to really *meet* individuals
<rekado>you can’t bump into them and stop for a chat
<rekado>initiating a conversation by selecting a person from a roster has a very different feel to it
<rekado>so I wonder if this already exists: a 2D game like environment with moving avatars and 2D audio
<rekado>you could walk to different locations where you hear the local sound; you can walk elsewhere to talk to other people, and the limitations of 2D positional audio mean that there’s a natural limit to what you can hear and what you cannot
<rekado>you could also become a ghost if you don’t want people to talk to you, but this means that you can only listen to audio in the designated conference rooms.
<rekado>OpenAL has the right abstraction for positional audio
<rekado>something like this certainly already exists, no…?
<rekado>I imagine it would be lovely for future Guix Days
<leoprikler>I think the real shit these days, pardon my language, would be a 3D game environment with 3D audio ;)
<rekado>3D is so 2000
<rekado>2D is where it’s at, yo!
<rekado>okay, let’s compromise: 2D graphics with 3D audio
<leoprikler>nah, we'll just have a VR meeting room
<leoprikler>2D graphics with 3D audio sounds like a very interesting niche
<rekado>I tried VR once (someone here bought an expensive system for … ‘work’) and was so very underwhelmed. What was exciting, though, were the many possibilities of getting injured in the R while being in the VR.
<rekado>3D audio would make the 2D pop!
<leoprikler>I could imagine that to be useful in visual novels
<leoprikler>to imply that things happen off screen for example
<civodul>rekado: what you describe sounds like a great idea, i like it!
<civodul>my experience with interactions over BBB has been rather frustrating
<rekado>civodul: one thing that causes me more headache than the implementation details is how much closer the component of motion moves online meetings to physical meetings — and thus opens up new possibilities for harassment and abuse.
<civodul>yeah, that's always a possibility...
<civodul>on the upside it can be very easy to kick the offender
<rekado>tricky to get it right, but perhaps lessons can be drawn from other systems like mumble
<civodul>yeah dunno
<civodul>i'm surprised it doesn't exist yet
<civodul>but there's this thing in some 3D proprietary games i think
<civodul>where the media reported that people demonstrated or whatever in some virtual space
<rekado>all audio needs to go to the servers to provide different mixes for different listeners, so whatever is said there can be retained anyway; makes it easy to *react* to harassment
<rekado>I’m surprised that there are no Guile bindings for OpenAL
<civodul>true, what are all the Guile game developers doin'? :-)
<rekado>probably distracted by yet another Lisp game jam
<rekado>ts ts ts … focus, people!
*janneke eh whut
<civodul>perhaps you could throw OpenAL at nyacc's FFI thingie
<rekado>perhaps SDL has similarly useful abstractions
<aitzkora>I'm seeking which guix packet contains the classic linux "column" command : what is the command to use to find that guix package -A ?
<terpri>for my part: writing C# without actually knowing C#, and porting essentially a game engine to godot. not much room for guile there, alas, beyond using guix for development and guile for some internal tools
<terpri>(although i'm presently a game-development tools developer rather than a game developer per se)
<rekado>aitzkora: it’s probably util-linux
<rekado>terpri: oh, Guix is useful for C# development?
<rekado>(is that Mono?)
<aitzkora>rekado : thanks!
<apteryx>I'm experimenting with having mount.nfs and mount.ntfs-3g setuid-root out-of-the-box for desktop users, any idea how we could make this work:
<terpri>rekado, beyond having mono available...not really, but i'm stubborn :p
<apteryx>currently it says: guix system: error: more than one target service of type 'setuid-program'
<civodul>apteryx: yup, there's already an instance of setuid-program-service-type in %base-services
<civodul>so you just need to extend it
<terpri>that reminds me, i should really try to get afs working on guix (for hcoop; also used by many american universities), now that the nfs service exists as a model
<apteryx>ah, can't they "combine" themselves, like some other services?
<civodul>i thought there was a helper to add setuid programs, but apparently not
<apteryx>there's a field on the operating system
<civodul>so you could use 'simple-service' to extend it
<efraim>aitzkora: I normally run 'which column' to see if I have it, then 'readlink $(which column)' to find the package. if that doesn't work 'find /gnu/store -name column'
<civodul>terpri: re game-development tools devs, that reminds me of Lispers who often are Lisp implementation developers :-)
<terpri>indeed :)
<terpri>incidentally, iiuc c# is becoming the preferred language for godot (the developers acknowledged that having their own mediocre language was a mistake), so good c# support will probably be more important for free games in the future
<davexunit>oh they finally acknowledged that? that's good
<davexunit>they were very adamant about it back in the day
<terpri>davexunit, i don't know if there's a blog post or anything, but that's what i was told by $client who tracks such things
<apteryx>civodul: re services, thanks that helped!
<terpri>(it's also possible godot is trying to become "language-neutral" or something, i haven't looked into it in depth, but in either case c# would be important for porting games off of a certain extremely popular proprietary game engine)
<aitzkora>efraim : But if it is not yet installed, how to find the package (I use guix as a package manager not as a distro)
<terpri>aitzkora, on a foreign distro, i believe you want the util-linux package
<terpri>(on guix system, column is in util-linux-with-udev, with-udev sounds like something you might not want on a foreign distro although i'm not sure what it signifies)
<terpri>aitzkora, 'guix search' is helpful for finding packages, but i don't think there's an equivalent to, e.g., debian's apt-file where you can ask what packages contain bin/column
<aitzkora>@terpri : thats was my question... sadly it seems there is not that kind of equivalent...thanks anyway
<terpri>aitzkora, there will be eventually as it's not terribly difficult to build a tool like that, it just doesn't exist presently, afaik (might have missed some news)
<cbaines>Any ideas why some old revision of Guix doesn't like en_GB.utf8 as the locale, but accepts en_US.UTF-8 ?
<terpri>(hacky way: build/install all packages and generate a 'locate' database of files under /gnu/store)
<roptat>I think I was able to send a RTM_GETLINK request to my kernel, but I get an empty answer :/
<apteryx>anyone knowns how to reopen a Debbugs issue? I tried unarchive followed by reopen, but it seems unsuccessful.
<civodul>cbaines: could be that you're using glibc-utf8-locales and it doesn't have en_GB?
<cbaines>civodul, I'm using the glibc-locales package, it's seems that 2.28 doesn't have the .utf8 locales, just the .UTF-8 ones
<cbaines>I also had the Guix Data Service use the .utf8 prefix, which is causing trouble now when loading older revisions...
<cbaines>I can't work out whether I should code for both, or just migrate to .UTF-8 which seems more available
<cbaines>Ah, I've found some reference to normalized codeset... maybe I should just migrate the code and data to .UTF-8 then, as that seems more compatible
<civodul>the normalized codeset is .utf8 though, isn't it?
<civodul>2.28 in Guix was 2 years ago?
<civodul>a bit less maybe
<cbaines>Yeah, obviously (setlocale ...) and whatever that does doesn't handle normalization of the codeset, otherwise things like UtF--8 should work too
<cbaines>The revision is from 2019-02-01, so not that long ago
<roptat>ok, I managed to get exactly the same sendto from "ip l" and my netlink library, but still no answer from the kernel :/
<roptat>oh, I got an answer! :D
<PotentialUser-52>Hi, i was wondering if there was a reason for .NET core to not being packaged apart from no one doing it
<PotentialUser-52>Also what is the default login manager?