IRC channel logs

2019-06-13.log

back to list of logs

<nckx>Yup.
<nckx>I often spout nonsense but never on #guix.
<quiliro>when i run it i get the message i3: Cannot open display
<nckx>You'd need something like ‘startx’ which AFAIK — and spot the theme here — Guix doesn't [yet] provide.
<nckx>Probably since all effort has gone into getting GDM to work.
<quiliro>you said you do not have it
<nckx>quiliro: I don't?
<quiliro>i missed your answer about videos, photos and the sort
<quiliro>how do you handle that?
<nckx>I don't follow. I don't use GDM. But development effort has been focused on it. By others.
<quiliro>you do not watch ovies or photos?
<quiliro>s/ovies/movies/
<ngz>I wonder in what file scintilla (www.scintilla.org) should go? Maybe text-editors.scm, or maybe gtk.scm. Any advice on that?
<nckx>quiliro: I'm still confused. I use X. I use i3. I am talking to you from HexChat and wasting time in IceCat. But I do not use GDM.
<nckx>GDM is a display manager, which I'll grant you isn't the world's least confusing name, but it just means that it pops up a Login: box and starts your WM.
<nckx>ngz: qscintilla is in qt.scm. I would have said text-editors, but I think there's no clear winner. Is a component a text-editor(.scm)? Is it merely a gtk(.scm) plug-in? &c. 🤷
<nckx>Objectively correct answer™: if one of those 2 files is already huge, pick the other ;-)
<nckx>I scintilla itself supports many plug-ins, add-ons, or anything someone might want to package: create a new scintilla.scm file.
<nckx>* If
<nckx>Perso'pinion.
<ngz>hmmm
<nckx>Naming is hard.
<sebboh>hi all. If it is cheap enough to do so, may I recommend distributing .sig files for the .iso files, too, not just the .xz files. Because, by default, xz deletes the archive after extraction. So, I can't check my signature now I think.
<ngz>text-editor.scm it will be.
<quiliro>nckx: oh! i understand now....how do you starty i3 without login manager, from logind?
<sebboh>my first impulse was for an independent scintilla.scm file since, what, like, a dozen (??) projects or so use it as a library. But then I haven't looked inside the other two options.
<sebboh>quiliro: personally I just log in at the console then type startx. I have a .xsession file which starts i3 for me. (Actually it's stumpwm these days but the differences aren't so important at this time.)
<nckx>quiliro: I use SLIM, it's what GuixSD used to use and I never switched. I like running everything from my .xsession (dbus, ssh-agent, user Shep' &c.) even though it's not the cool new RedHat way to do it.
<nckx>sebboh: startx works on Guix System now? 😮
<nckx>I am so out of touch.
<sebboh>nckx: oh! heck, hold on, I don't know. Ignore me. I was talking about my other systems. So sorry..
<sebboh>The reasons not to do that are obvious now. :)
<nckx>sebboh: Heh. I personally prefer many small modules too, but there's really no clear rule here.
*nckx still (mostly) nonsense-free in #guix, phew.
<nckx>I mean, there's an ‘xorg-start-command’ command, but I don't think it's a drop-in replacement for a package. 🤷 again.
<nckx>s/command/procedure/
<nckx>sebboh: Amen.
<quiliro>sebboh: i like your take...how do you install that?
<quiliro>you could have startx run at login
<quiliro>maybe it would not be good in case there is a problem with video
<sebboh>quiliro: sorry, I don't know. I'm a guix newbie.
<sebboh>so far, I've only installed it on a VM at work. Which is hyperV.. so I can either put up with the terrible gui app or try to put the same microsoft-provided xorg-rdp magic into it that I put into my debian VMs.. but like, my first package should be a simpler one, ya know? :P Anyway I'm installing it on hardware at home tonight.
<quiliro>but if you installed it on a vm you did the trick...how did you do it?
<nckx>quiliro: They were talking about non-Guix systems.
<nckx>(startx on ~)
<quiliro>nckx: ok
<nckx>scheme@(guix-user)> (xorg-start-command)
<nckx>$1 = #<<program-file> name: "startx" gexp: #<gexp 3547900> guile: #f path: ("/gnu/store/…
<nckx>That's ‘not something we can build’ according to guix build (and it should know), but I wonder if it could be made into, for example, a special-file, or wrapped in a ‘startx’ package.
<quiliro>nckx: how is your config without gdm?
*nckx is trying the former option but stock waiting for build slots while their kernel builds.
<nckx>*u
<nckx>quiliro: …pleasant? Extremely weird & custom in general? An abomination of badly-commented/indented/written code that noone must ever see? All 3.
<quiliro>haha
<quiliro>i want to try it to see if gdm is the problem
<quiliro>but probably it is not that
<quiliro>it is probably the video driver
<quiliro>but this problem is with several video cards
<nckx>Here's the slim- part, https://paste.debian.net/1087671/, but as I said before it's probably not suitable for copy-pasting. I don't use %base- or %desktop services, etc, and probably do other weird stuff elsewhere that I forgot.
<nckx>You'd have to remove gdm from %desktop-services, for one.
<nckx>quiliro: It's very strange that you're the only one that seems to have this problem.
<nckx>If you want to restrict video drivers like I do in that snippet, you can do so, but don't ask me the syntax to customise gdm-service-type.
<nckx>Search for modify-services in the manual if you feel brave.
<nckx>It links to examples too.
<quiliro>i will use this disk on another machine that works perfectly
<quiliro>tomorow i will report
<quiliro>what data do i need to report?
<quiliro>i have only messages with the error i reported before
<quiliro>/var/log/messages
<quiliro>i do not have other data on this hardware
<quiliro>maybe i could install that disk on these two hardwares too, in order to test
<nckx>No /var/log/Xorg.*.log?
<nckx>quiliro: You could do that.
<quiliro>none
<nckx>Unless there's a very obvious error in your log (‘oops, couldn't frob the foo, please do bar’), other people on the mailing list will probably ask you to try to reproduce on other hardware anyway.
<quiliro>found this on the archwiki via openchrome driver page:
<quiliro>If your machine freeze at startup (GDM) or after login (SLiM), try adding the XAA option XaaNoImageWriteRect. Note that this only applies if you are using the XAA acceleration method (configured by the AccelMethod option). Since 0.2.906, the default acceleration method is EXA.
<quiliro>
<quiliro>how can i make the driver use exa acceleration or not use acceleration?
<nckx>How does the wiki tell you to do it?
<quiliro>but i wonder why the other machines have the same problem
<quiliro>on xorg.org
<quiliro>file
*nckx found out how to add a (untested) startx to Guix \o/
<quiliro>but we do not use that or do we?
<nckx>quiliro: In my last paste, you'll see an %xorg.conf variable ;-)
<nckx>It's just a string I (define…) further up.
<quiliro>does it have the standard format?
<nckx>You can (define %xorg.conf "Section \"FooBar\" … <the rest, may include newlines>")
<nckx>quiliro: It's raw Xorg.conf goodness.
*nckx (service special-files-service-type `(("/bin/startx" ,(xorg-start-command))) if anyone's wondering.
<nckx>quiliro: Note the escaping of \" to make it a valid Guile string.
<nckx>And you'll need to modify-services the gdm service to add the (xorg-configuration (xorg-configuration (extra-config (list %xorg.conf)))) from my snippet to yours.
<nckx>It's not trivial but it works. If it doesn't, you've made an error somewhere 😛
*nckx → sleepy.
<nckx>Good night.
<quiliro>nckx: thanks
<Marlin1113>hi guix
<Marlin1113>i'm trying to package openspades
<Marlin1113> https://github.com/yvt/openspades
<bavier`>Marlin1113: looks cool
<bavier`>the license may need some discussion if we'd like to include it in Guix
<Marlin1113>i think it's gpl v3
<bavier`>I see on http://openspades.yvt.jp/ some exception for the audio files
<Marlin1113>oh
<Marlin1113>asset issues
<Marlin1113>i don't think that's usually an issue if the code is free tho
<vagrantc>as long as it works without the assets...
<Marlin1113>oh
<Marlin1113>i think it probably does
<bavier`>the "official" flatpak includes a bunch of proprietary stuff
<bavier`>see https://github.com/yvt/openspades/blob/master/Resources/downloadpak.sh
<bavier`>but yea, if the game otherwise works, shouldn't be a problem
<bavier`>though https://github.com/yvt/openspades/blob/master/Resources/PakLocation.txt says that "pak000-Nonfree.pak" are *essential* game assets
<bavier`>doesn't sound good
<Marlin1113>crap
<Marlin1113>i guess i'll see if i'll port it later
<Marlin1113>if it's so dependent on proprietary stuff i might as well stop playin it
<bavier`>it's strange too, because the sound files, which have the strictest redistribution terms of the bundle, are copyright by the same person who's developing the gplv3 game
<c2V0a3U>hai.. is there a way to mount usb without root/sudo, like a udiskctl does ?
<sebboh>it sounds strange. It's a card game? I'm just speculating here; I didn't even look at the repo.. Could the essential assets be replaced? You know, on principle. :)
<bavier`>sebboh: it's a first-person voxel-based shooter
<bavier`>sebboh: I suppose they could be replaced, but would require quite a bit of work
<Marlin1113>the game itself is pretty gud
<Marlin1113>better than the original even
<Marlin1113>the original died because the company didn't succeed and didn't want to turn it into free software
<Marlin1113>so they just held it in and the game died
<Marlin1113>but there was an older version that people got the source, so there's still some of it left
<bavier`>Marlin1113: I'm trying to build it manually right now, just to see how things go
<sebboh>bavier`: oh, I thought it was a card game. Carry on!
<Marlin1113>because of the spades? :P
<sebboh>Well, yes.
<sebboh>I'm in text only mode. :)
<sebboh>"Now in 3D! Open Spa DES, wherein you fight your way out of an evil day spa with encryption!"
<Marlin1113>sounds good to me
<Marlin1113>i was thinking about ways to make a profit using 100% free software. Saw some people talking about how it's hard to
<Marlin1113>i might try to experiment with it sometime
<sebboh>Here's the least pragmatic thing I've said all week (probably): instead of using free software to make money, try to use it to save the world so effectively that you no longer need money.
<sebboh>And with that, I'll sign out for a bit.
<Marlin1113>finding a viable and fair way to make money with free software would be great. At least people wouldn't always resort to making their stuff proprietary in order to get some money
<Marlin1113>hey, nckx, i wasn't able to use your answer for the emacs daemon thing
<recj>i need openssl to build something but it says openssl already exists in guix/licenses
<recj>what should i do
<bavier`>recj: usually we put a 'license:' prefix on the (guix licenses) module import
<recj>sounds like now is the time to do my grep and check for that
<recj>and how it's done
<recj>almost done with this, i just need to get rid of the check phase because this package doesn't have it
<recj> https://paste.debian.net/1087680/
<recj>says there's a syntax error with deleting it but i don't see how
<recj>never mind, i had to many parens
<recj>*too
<recj>now i am getting a classic
<recj>it can't create a directory in /usr/
<bavier`>Marlin[m]: see https://github.com/yvt/openspades/issues/424
<efraim>interesting, pkgsrc requires a C compiler in $PATH
<civodul>Hello Guix!
<efraim>Hi!
<user_oreloznog>hi \o
<mfg>Hi, when i `guix gc -C' after removing some generations with `guix package -d' then programs installed in older generations which are still present in the current generation are not deleted?
<mfg>Because it seemed that some programs broke after i removed some old generations (maybe a coincidence)
<civodul>hi mfg
<civodul>indeed, old generations stick around unless you delete them, and everything they refer to is still available
<civodul>what programs broke?
<Marlin[m]><bavier` "Marlin: see https://github.com/y"> Oh
<mfg>civodul: it was sbcl-next - the dependency webkitgtk was gone it took me a while to get it working again, i suppose because i'm new to guix :D.
<mfg>But now it works again :)
<mfg>Another thing i was wondering about is: how difficult is it to update the android tools? i need android-9.0.0-r42 and in the repos is 7.1.2-x. Is it just updating android-platform-version?
<civodul>mfg: then you've probably hit this sbcl-related bug: https://issues.guix.gnu.org/issue/33848
<civodul>re android tools, i don't know!
<civodul>you could try and email guix-devel if it's not as smooth as you'd like
<apteryx>hello!
<mfg>civodul: ahhh, yes that explains it :D
<mfg>but before i can try to update the android things i have to solve the makeinfo problem. i can't build guix because makeinfo fails for the russian and spanish translation, is there any workaround for that?
<civodul>mfg: did you run ./bootstrap?
<mfg>ah yes, i had to git clean -fdx and redo everything now it works
<civodul>cool
***ym555_ is now known as ym555
<mfg>how do i use guix repl to inspect default variables like base-services and such? or is there another, preferred way for this?
<nckx>Marlin[m]: I'm going to need a bit more info than that ☺ I didn't paste my full configuration but those are the relevant parts of a working set-up.
*nckx → AFK now though.
*nckx lied for a few minutes.
<nckx>mfg: ,use (gnu services foo)
<mfg>thx nckx
<nckx>At that point you're already grepping the source to find the module, so I think most people just ‘grep define.*%foo’.
<mfg>I'm now trying to build the android things. I get a warning: ACL for archive imports seems to be uninitialized. How do I initialize this? because afterwards it fails with a permission denied error
<nckx>mfg: guix archive --authorize < the-key.pub
<mfg>what key? i didn't make one
<nckx>find ~/guix.git -name "*.pub"
<nckx>It's also stored somewhere in the guix profile…
<mfg>ah so it's in the repo?
<mfg>which one of them is the correct one? ci.guix.gnu.org.pub, i guess?
<nckx>mfg: Well, that depends on the server you're using. If you don't know, you're using the default, which is indeed ci.guix.gnu.org.
<mfg>yep i'm using the default one ...
<nckx>After running the command above, you can cat /etc/guix/acl ‘just to make sure’, even though that's not technically needed.
<mfg>good to know nevertheless :P
<nckx>(Now, I don't quite see how an empty ACL would lead to that error; it just prevents you from using substitutes. But now I must go.)
<Marlin[m]>nckx: i just copied the files. Do i need to run something to run something to make them be used?
<bendersteed>hello people
<bendersteed>i'm trying to package some stuff and I get this http://paste.debian.net/1087708 when trying to download
<mfg>what mailing list should i contact with questions regarding packaging?
<bendersteed>guix-devel I think
<mfg>kk
<civodul>bendersteed: the 'guix-download' procedure takes a string (or list of strings), but you're passing it a <git-reference> record
<civodul>i'd recommend simply running "guix build -S go-github-com-burntsushi-locker"
<bendersteed>civodul: ah ok thanks, apparently that's what guix-devel-download-package does in emacs
<bendersteed>guix-devel-download-package-source*
<civodul>oh C-c . s ?
<bendersteed>maybe this is a bug?
<civodul>probably
<bendersteed>C-c . d
<civodul>ok
<civodul>yeah i think you should report it
<bendersteed>ok thanks, will do so now
<bendersteed>just in reference it's already there https://gitlab.com/emacs-guix/emacs-guix/issues/7
<civodul>awesome
<civodul>sneek_: seen alezost
<sneek_>I last saw alezost on Oct 23 at 02:58 pm UTC, saying: civodul: hello.
<civodul>uh, a long time ago
<civodul>there's been recent activity though
<civodul>bendersteed: maybe you should ping in there?
<bendersteed>from the discussion it seems that because emacs-guix depends on guix download it accepts just strings
<bendersteed>so there should probably be a non-trivial solution
<bendersteed>now I'm researching how git fetch method works in guix, I'm pretty new to this
<bendersteed>maybe we can have a function that returns the appropriate string
<civodul>bendersteed: it should use package-source-derivation instead of 'guix download'
<civodul>(package-source-derivation store the-package)
<civodul>that works for any <origin>, regardless of its download method
*civodul merges master in core-updates
<civodul> https://hal.archives-ouvertes.fr/hal-02135891 <- feat. Guix
<mbakke>civodul: How did the merge go? In related news, I think 'staging' is finally good to go.
<civodul>mbakke: yay for staging!
<civodul>the merge went well, modulo a couple of adjustments
<civodul>i'm now running "make as-derivation", which will take a while
<civodul>i'll push when it completes
<civodul>(if it completes :-))
<recj>starting phase `install'
<recj>mkdir: cannot create directory ‘/usr’: Permission denied
<recj>any idea what i should do in this case for a package ?
<nckx>Marlin[m]: Well, you have to start ‘shepherd’ as your user. This is best done automatically when you log in. How you do so depends on your environment. I use .xsession to start my WM, so that's where I start the Shepherd.
<str1ngs>recj: check the Makefile possible it does not use DESTDIR
<mbakke>civodul: Sounds great :)
<str1ngs>recj: you might have to substitute prefix. whatever variable is used in the Makefile
<nckx>str1ngs: Small correction, if that: prefix= (or its BSD counterpart PREFIX=) is righter than DESTDIR=, so try those first.
<recj>I see
<nckx>As in more likely to do the right thing.
<recj>ok
<str1ngs>it's subjective, you should check the Makefile regardless
<nckx>str1ngs: ?
<nckx>Packages differ and some are downright buggy, but it's not subjective. ☺
<str1ngs>make variable are subjective. there is no defined standard for this
<nckx>Not accurate. DESTDIR= is for staging only, PREFIX= for the final location.
<recj>it does not use DESTDIR
<str1ngs>again, if it's a Makefile there is no standard for these variables
<nckx>recj: And prefix or similar?
<recj>nope
<nckx>str1ngs: Again, that's plain false (unless you redefine standard to fit your terms) but I'm not going to argue this.
<str1ngs>probably hardcodes /usr . so you'll have to substitute is my guess
<str1ngs>nckx: please understand how make works. they are simply variables or no variables. there is not standard as to if those variables are used or set. ultimately you will need to manual check what Makefile . just because autotools and some Makefile adhere to known variables. There is not reason they can or will exist.
<nckx>OK. You've made the same point repeatedly.
<nckx>To address more productive matters: any luck, recj? ☺
<str1ngs>we can agree to disagree then :)
<kmicu>You can be both wrong too. ;)
<ItsMarlin>Hello friends
<ItsMarlin>You can both agree in being my friends
<ItsMarlin>:D
<str1ngs>hello friend :)
<nckx>But knowing what the word ’standards’ means makes me objectively superiour and entitles me to special perks & privileges ☹
<nckx>Noo.
*nckx buys everyone a botsnack.
<ItsMarlin>nckx: if i start the daemon with X, it'll be the same as starting emacs normally
*kmicu sees that ItsMarlin moved from ‘guys’ to ‘friends’. That’s kind.
<ItsMarlin>can't i run it at boot?
<nckx>kmicu++
<ItsMarlin>kmicu: :)))
<nckx>ItsMarlin: Could you explain why ‘it'll be the same as starting emacs normally’? That's not clear.
<nckx>What do you really want?
<nckx>What do you desire from emacs?
<kmicu>Everything!
<ItsMarlin>Yes
<ItsMarlin>haha
<ItsMarlin>well, i use exwm
<nckx>That's just plain emacs tho'.
<ItsMarlin>yes
<nckx>(Everything, not exwm.)
<ItsMarlin>but i'd like to run emacs' daemon at boot
<ItsMarlin>so that exwm doesn't take too long to start
<ItsMarlin>exwm already launches emacs at login, so running the daemon at login wouldn't help much
<nckx>ItsMarlin: So you want your system shepherd to start an emacs daemon as your user so when your it starts your display manager as root the exwm it spawns will connect to that daemon. I hope you concede that's not an every-day use case 😃
*kmicu saw a setup where Emacs is PID1.
<nckx>Assuming exwm actually does the right thing to find and talk to that daemon, I *guess* I'd add a shepherd service that just uses ‘su’.
<str1ngs>init=/path/to/emacs can achieve that :P
<nckx>kmicu: I had that nightmare too.
*nckx ♥ emacs but c'mon.
<nckx>str1ngs: Run everything as root! Solves all those pesky permission errors.
<ItsMarlin>emacs best kernel
<kmicu>ItsMarlin: why exwm takes so long to start?
<nckx>Windows 9x was on to something.
<kmicu>Moving to deamon is another can of worms cuz deamon cannot see everything and you cannot easily refersh its environment when entering custom guix shells.
<kmicu>Maybe it’s better to improve Emacs config so it could load under 2 seconds? (E.g. with use-package and lazy loading).
<nckx>☝ ☝ ☝
*nckx switched to the daemon to save ~4 seconds of load time to check my mails but actually knowing how to write a proper .emacs would have saved /me the trouble.
<str1ngs>I just don't restart emacs :P
*str1ngs emacs-uptime -> 3 days, 15 hours, 47 minutes, 26 seconds
<nckx>ItsMarlin: I'd prototype it with a simple stupid ‘system "su" "…" ’ system service (s⁶), then see if it really solves your problem and doesn't introduce new ones.
<nckx>str1ngs: Holy trucknutz, TIL emacs-uptime.
*kmicu saw folks moving to Emacs’ daemon mode in NixOS (there’s a service for that) but then almost everything is broken when you’re in a (--pure) nix-shell.
<nckx>Thanks for that useless thing I will now call regularly 😃
<str1ngs>the fact that emacs is lisp state machine avoid generally having to restart. though there are good reasons too. but still
*nckx ‘4 hours, 34 minutes, 47 seconds’ — amateur.
<nckx>kmicu: What do you mean by ‘daemon cannot see everything’?
<str1ngs>DISPLAY for one
<nckx>Oh, that kind of see, good point.
*nckx was thinking of files.
<str1ngs>also XSESSION variables, maybe things like dbus as well
<nckx>(Although I guess profiles won't work at all, obvs.)
<rvgn>Hello Guix!
<ItsMarlin>who is this guix you're talking about?
<kmicu>The Guix Community.
<kmicu>rvgn: ( ^_^)/
<nckx>kmicu: Note that that only a daemon started outside of the session would suffer from that, not ‘daemon’ in general, but thanks for thinking of that.
<rvgn>Referring to all folks in this Guix community.
<nckx>sneek_: Who is Guix?
<sneek_>guix is a functional package management tool for the GNU system
<rvgn>* ItsMarlin
<nckx>Is this like Who is John Galt because then I'm just out.
<kmicu>nckx: that’s too but when it’s started in a session but I jump into a pure shell with some deps then those are invisible in daemonized Emacs’s path.
<rvgn>I have doubt. How to use App-Image in Guix? I downloaded an appimage and could not open it.
*nckx resists eshell! 🙈
<recj>nope nckx no luck so far
<kmicu>nckx: some folks write ton of code to avoid restarting Emacs from within a (pure) custom shell/environment https://www.tweag.io/posts/2019-03-28-introducing-lorri.html
<nckx>rvgn: This is yet another container format of some kind, AFAICT?
<kmicu>rvgn: those trendy formats are not supported yet. They assume a distro with FHS. They are like a half-baked Nix/Guix. 😺
<nckx>
*rvgn just realised that "fuse" is required to use AppImage
<nckx>You'll presumably have to package appimagerunner, or however it works, in Guix.
<nckx>rvgn: We have that.
*nckx made up that name.
<rvgn>nckx Kind of container/package format that is distro independent
*rvgn rvgn installs *fuse*
<nckx>I.e. they support both Ubuntu and Red Hat. Cool.
<nckx>if </usr/bin/foo> else </bin/foo>
*kmicu adds a star to ‘distro independent*’ if that distro follows FHS.
<nckx>Do I get VC funding now.
<kmicu>IF? Add some more IFs and we can start an AI startup!
<rvgn>kmicu what's FHS?
<nckx>Binary blob hacks like these won't work on Guix, I'm afraid, without (possibly involved) hacks. Like an FHS service.
<kmicu>rvgn: https://en.wikipedia.org/wiki/Filesystem_Hierarchy_Standard
*nckx isn't really afraid.
*nckx feels safe and cosy.
<nckx>kmicu: You'll need to install the proprietary GPU drivers to run more than one if don't you know anything about deep learns.
<ItsMarlin>nckx: join the eshell gang
<nckx>I always knew it was a gang!
<nckx>ItsMarlin: Sure, convince me ☺
<nckx>I like my termite but I'm open-minded. Go.
<rvgn>kmicu cool
<ItsMarlin>u can run emacs commands from it uwu
<ItsMarlin>and use lisp for making shell scripts
<ItsMarlin>shell scripts with a language that makes sense, gud huh
<nckx>Well, I already have emacs, bound to Mod+e, I'm not an animal.
<ItsMarlin>u a
<nckx>ItsMarlin: Guile scripts check mate atheist.
<ItsMarlin>u already have eshell too *
*nckx has to go, now, sorry.
<ItsMarlin>oh no
<kmicu>nckx: eshell is nice but even pipes lose data there so better avoid if for BIG DATA.
<rvgn>kmicu nckx Is guix FHS compliant?
<kmicu>rvgn: It is not. By design. Everything should be under /gnu/store/.
<rvgn>okay
<kmicu>That’s why to support Flatpaks, Snaps, AppImages Guix needs a service emulating FHS. That’s possible but not implemented yet.
<sammich>what is FHS?
<kmicu>sammich: a standard followed by traditional distros so we can assume that certain things are in certain places https://en.wikipedia.org/wiki/Filesystem_Hierarchy_Standard
<rvgn>kmicu Will guix be able to form a new standard?
<kmicu>(It also makes systems not really reproducible and makes detecting reproduciblity issues more difficult. That’s why Nix or Guix do not follow FHS.)
<kmicu>rvgn: heh, I doubt it. It’s like moving away from QWERTY.
<rvgn>kmicu Like reforming the FHS standard
<sammich>i see, what would the daemon have to be doing for the afformentioned package toolkits have to be doing exactly? woudl it be like a FUSE with your environments FHS or osmething?
<rvgn>I see.
<sammich>there is one too many have to be doing in there
<cap>Greetings. I installed icecat but there semes to be no font provided, so that I have a bunch auf font placeholder infront of me. Does anyone have a hint for me, how to fix that?
<kmicu>cap: install some font and ‘fc-cache -f’ to refresh the font cache. Restart IceCat and fonts should be there.
<kmicu>sammich: Guix could provide a service to emulate FHS like https://nixos.org/nixpkgs/manual/#sec-fhs-environments and then we could run all those Flatpaks/Snaps/etc there.
<sammich>hot dog, thats a pretty cool idea
*rvgn gonna reconfigure system with fuse and see whether appimage works.
<kmicu>rvgn: I whish you luck but that should not work. NixOS uses FHS to run AppImages https://github.com/NixOS/nixpkgs/blob/19.03/pkgs/tools/package-management/appimage-run/default.nix#L5
<cap>kmicu: thx. I installed font-dejavu and fontconfig and run the command, but the problem is still there. Do you have another idea?
<kmicu>cap: did you refresh the cache with ‘fc-cache -f’?
<kmicu>cap: to clarify, are you on Guix System or did you install IceCat from Guix on a non Guix System distro?
<g_bor[m]>cap: hello cap, tomorrow I intend to share my config with sway. I will do some packaging before to have a better experiece, and split my manifest to get a better overview of what belongs to the wm.
<cap>kmicu: Yes, I did. I'm on GuixSD and installed icecat via config.scm. I'm using the slim desktop manager and removed gdm. Also I'm using i3.
<cap>g_bor[m]: Thanks a lot!
<kmicu>cap: another ideas are: you could uninstall fontconfig that’s not required adn could shadow system fontconfig. You could run ‘fc-cache -r’ and run it with sudo (or as root) (and restart IceCat after that).
<cap>kmicu: I installed fontconfig because I did not have the command fc-cache.
<kmicu>Hmm. I had the same issue in the past and I don’t recall not having fc-cache.
<kmicu>Maybe root has it and I used that account. Don’t remember.
<cap>kmicu: Well… as root I have fonts. I also noticed that I can't use wget as non-priv. How do I install packages systemwide via config?
<rekado>nckx, janas: I want to use mpd (on a home server), but I only have it installed and configured it manually to see if it’s a possible solution (it is). I’m not currently using it and I know rather little about it.
<civodul>mbakke: libpng as 1 (one!) test failure on core-updates, which breaks "make as-derivation"
<civodul>s/as/has/
<civodul>the "good news" is that it's not deterministic :-)
<civodul>it built on the second try...
<g_bor[m]>cap: what does wget tell you?
<cap>that it wasn't found
<g_bor[m]>Also with fontconfig: these is a profile hook on Guix System to get that working when installed in a user profile. When you list it in the operating system packages, it gets installed, but a systemwide config is most probably not provided. One way to get around that would be to create a service providing the fontconfig etc files. Does that make sense?
<g_bor[m]>cap: is it listed on the packages field on your operating system? And have you ran reconfigure? If yes, then it might be something with the path. It should reference the system profile somewhere.
<cap>g_bor[m]: I don't quite understand. I installed it as root. I will try to reconf
<bgardner>cap: Did you install it as root or add it to your config.scm?
<cap>bgardner: Both so I shouldn't need to reconf instantly
<g_bor[m]>cap: I am confused.
<bgardner>cap: Yes, you do - reconfigure makes it visible to non-root users, installing it as root made it visible only to root.
<cap>I wrote it to the conf and installed it via guix install, cuz I didn't wanted to wait for the rebuild process.
<bgardner>cap: And installing as root was not needed, just add it to config.scm and reconfigure.
<cap>Now I got it. Thx
<bgardner>cap: My pleasure. And I'm not expert, I just recognize your predicament because I did it to myself before.
<jonsger>sneek_: later tell playX: llvm version 8.0.0 is now available on guix master :)
<sneek_>Got it.
<sammich>jonsger: yippee thanks for merging core-updates
<jonsger>sammich: only staging was merged and it was not me, it was mbakke :P+
<bgardner>ccccccgvhgcjdvvtuklghnbvctrdgkeknfdvvthjnklc
<bgardner>ccccccgvhgcjjhvtuntucegjjhgtfbufhehicnjnebhn
<bgardner>ccccccgvhgcjvinrngcnkuljlkvvhigifnnvfnurjjih
<sammich>mbakke: thanks goes to its rightful owner then >:)
<nckx>bgardner: Are you generating store hashes as a service?
<nckx>Your PRNG seems off.
<bgardner>Oh crap, sorry. No, hardware key - I'm just an idiot
<cap>Unfortunately I have already a different problem: I'm trying to use password-store. I installed it, gpg and pinentry-gtk2 via config but when I try to generate a key, it tells me that pinentry wasn't fount.
<bgardner>You need to configure gpg-agent.conf to indicate your preferred pinentry
<nckx>cap: e.g. ‘pinentry-program /home/nckx/.guix-profile/bin/pinentry-tty’
<pkill9>is there a guide on creating a service that you can use a configuration with?
<cap>thankyou.
<cap>nckx: I used /run/current-system/profile/bin/pinentry-gtk-2 is that bad style?
<nckx>cap: No, that's fine, it depends entirely on where you installed it (system vs. user profile).
<nckx>cap: Installing all your ‘day-to-day, user’ packages into your system configuration is, however, but only slightly. It means you'll have to rebuild/download all packages to update your system, which could block/delay security updates. But it's not ‘bad’-bad. It's up to you.
*nckx just watched some Marvel films and can't read ‘cap!’ any other way.
<pkill9>i can't seem to run the functions that return the parts of the configuration record
<pkill9>e.g ERROR: Wrong type to apply: #<syntax-transformer fhs-configuration?>
<civodul>pkill9: this means fhs-configuration? was used before being defined
<civodul>you probably have to move its use after the record definition
<pkill9>ohh, thanks!
<pkill9>being able to change your system config and instantly test it in a VM is niice
<civodul>isn't it? :-)
<cap>nckx: thanks for the advise
<civodul>yay for the merge! thumbs up mbakke!
<quiliro>hello guix
<quiliro>i have tested the installation on this hard disk on another machine where guix system is functioning perfectly
<quiliro>this disk installation (the one i am writing from) has the same problem on this machine that it has on the other machine
<quiliro>the problem is with installing i3
<quiliro>video problem
<quiliro> http://ix.io/1L2l is my config file
<kmicu>Shouldn’t your initrd-modules extend %base-initrd-modules?
<quiliro>with a barebone config file, i have no problem
*kmicu hopes that (file-system-label "raiz") is not a typo for raid.
<quiliro>haha
<quiliro>kmicu: it is just the label
<quiliro>of the root
<quiliro>raiz = root in spanish
*kmicu prepares for the next hailstorm.
<kmicu>quiliro: whab about extending %base-initrd-modules?
<kmicu>initrd-modules has default %base-initrd-modules so I recommend (initrd-modules (cons "patasomething" %base-initrd-modules)) instead.
<quiliro>kmicu: what is the syntax if i have 2 modules?: (initrd-modules (cons* "patasomething" "anothermodule" %base-initrd-modules))
<quiliro>is that correct?
<kmicu>Yep, cons*.
<kmicu>( https://www.gnu.org/software/guile/manual/guile.html#List-Constructors )
<sammich>How can i get the directory of binary from the build step in a cmake-build-system package? I need it to be able to run the tests
<rvgn>Is there a way to make guix automatically set paths and environment variables when new packages are installed?
<kmicu>rvgn: are you asking about Guix, the package manager or Guix System, the operating system?
<rvgn>the former
<quiliro> (initrd-modules (cons* "pata_via" "pata_acpi"
<quiliro> %base-initrd-modules))
<quiliro>is that correct?
<rvgn>kmicu For example, when I install some packages, at the end of the process, it shows a "hint" section to export paths and environment variables.
<kmicu>rvgn: Guix, the package manager cannot do such things cuz it has no access to an operating system’s piping. Every distro has different ways to setup envars or paths. THat’s why we love Guix System where services can do all those things.
<kmicu>So ideally we should say (services something) in system config and that something should be setup properly after guix system reconfigure step. No user intervention required.
<quiliro>kmicu: thank you for the manual...i will read it now
<rvgn>kmicu How to do that with guix system? For example. everytime I do "guix pull" as "root", it will "hint" me to "export" a path. How to automate this process?
<kmicu>quiliro: looks correct. guix system reconfigur should not complain about it.
<kmicu>rvgn: Guix System service could set that path already. Could you share that hinted export thing here?
<rvgn>kmicu Also when I install a package, let's say flatpak, it will hint to set xdg_data_dir paths. Is there a way to make this automatic?
<rvgn>kmicu sure, just a sec.
<kmicu>Yes, create a Flatpack service.
<kmicu>Flatpak*
<rvgn>kmicu export PATH="/root/.config/guix/current/bin${PATH:+:}$PATH"
<kmicu>(Or a package wrapper.)
<kmicu>Guix System shows you that after guix pull?
<rvgn>I have to export that in my terminal everytime after running guix pull. But it does not hint anything if run guix pull as user.
<rvgn>yes
<kmicu>Let me check on a fresh system.
<rvgn>kmicu thanks
<rvgn>kmicu https://upload.disroot.org/r/vof6sepq#oALVQqKoaKw1HGmtn9+kVQUzDeyiGU6dSO1jiBr4ny0=
<quiliro>i will test by rebooting
<quiliro>thank you
<rvgn> https://disroot.org/upload/Pa4FCDNiPaQxF_cp/Screenshot%20from%202019-06-13%2013-22-56.png
<jahb>hey #guix i keep getting a configure error when updating guix packages on a foreign distro; happened on various packages over time: "checking whether we are cross compiling... configure: error: in /tmp/..." and then "configure: error: cannot run C compiled programs.\nIf you meant to cross compile, use `--host'."
<jahb>is it me do you think?
<efraim>is /tmp mounted noexec?
<jahb>yes
<jahb>goddamn it I keep forgetting
<jahb>thank you
<efraim>:)
<nly>hello, is there any canonical way to filter package-inputs?
*kmicu still watches guix pull grafting all git things.
<NEET`>Can anyone tell me what the default console font on the guix installation is? I quite like it
<rubic88>Hi Guix. I just got my first native guix installed on a ThinkPad X220 (yay!).
<jackhill>rubic88: congrats!
<NEET`>Nice rubic88
<rubic88>I want to read some stuff from a usb drive. Should I mount it manually? mount /dev/sdb1 /mnt/usb
<kmicu>NEET`: %default-console-font points to LatGrkCyr-8x16
<NEET`>kmicu: Thank you! How did you find this out exactly?
<NEET`>I mean, like what file did you look at or whatever
<jackhill>rubic88: I use gnome, and the gnome-files mounting helper, which uses udisks (I think) works for me. If something like that works for you could do that as well. Otherwise, manually mounting should work.
<jackhill>I like using the desktop mounting stuff because it doesn't require me get elevated permissions.
<rubic88>Manual mount worked for now. I'll add auto-mount later. Thanks.
<jackhill>cool
<rubic88>Really happy to get wifi working once I replaced the bios (with coreboot) and the Lenovo wifi chip.
<quiliro>i have tested http://ix.io/1LHa configuration file for system reconfigure but I have video problems. It is respawning video....Please check what could be wrong with it.
<quiliro>i get error gdm_session_set_environment_variable: assertion 'value != NULL' failed
<quiliro>on /var/log/messages
<raingloom>hey, is there an equivalent to pacman's ignore setting? I accidentally updated to Blender 2.8 which is unusable on this machine so I wanna keep it at 2.79b and i'd prefer not having to set that in a config file instead of using a command line flag each time.
<raingloom>s/not having to//
<pkill9>raingloom: there is an environemnt variable for setting default commandline flags
<raingloom>pkill9, i guessss that could work, thanks!
<pkill9>GUIX_BUILD_OPTIONS is the one, https://www.gnu.org/software/guix/manual/en/guix.html#index-GUIX_005fBUILD_005fOPTIONS
<pkill9>you could create config file functionality by setting an alias for guix to point to a script that sets that environment variable
<bavier> raingloom: or you can pin blender to a previous version of guix, see "Inferiors" in the manual
<kmicu>rvgn: yes, I see that message somewhere in the middle of guix pull messages. No idea why Guix System shows that ‘may be needed’ stuff.
<pkill9>bavier: would that stop `guix package -u` from upgrading blender?
<pkill9>if you install an inferior package
<kmicu>quiliro: you should find Xorg log and share it. W/o that it’s impossible to guess what’s wrong. Or at least desrcibe your hardware.
<quiliro>kmicu: it is not a question of hardware. the disk boots in the same way on other hardware where another system works well
<kmicu>But there is different hardware (including video cards) there?
<bavier>pkill9: I think so, the local profile manifest tracks the provenance of each package
<raingloom>pkill9, ok so I set it to "--do-not-upgrade=blender" and guix upgrade complains about that being an "unrecognized option" (both with and without the "=")
<quiliro>kmicu: i have tested on 3 different machines
<quiliro>each has different video cards
<kmicu>quiliro: but still the same disk and the same config?
<quiliro>and the systems that work, if i change them to other machines, they work well here
<kmicu>quiliro: Is /home/* on that disk fresh? Did you format it?
<quiliro>no
<kmicu>What is that working ‘the system’? Is this a Guix System or another distro?
<quiliro>it has nothing to do with home
<quiliro>Guix System...not a foreign distro
<kmicu>Not booting Xorg can be broken b/c of config files in home. I saw that many times.
<pkill9>raingloom: I don't get that error hmm
<kmicu>quiliro: Do you remember the version of that working Guix System?
<quiliro>kmicu: impossible....it does not login to the user yet
<kmicu>quiliro: possible. Login to user is not important.
<quiliro>kmicu: no...but i updated it recently...after 1.0
<quiliro>it is slower than before...but works
<kmicu>quiliro: so that working Guix System was something before 1.0?
<quiliro>was 0.16 but i updated it
<quiliro>and kept working
<kmicu>But it was working correctly?
<quiliro>was and is
<pkill9>raingloom: does it work if you run like `GUIX_BUILD_OPTIONS="--do-not-upgrade=blender" guix package --upgrade`?
<bavier>raingloom: 'guix upgrade' does not support that option
<kmicu>quiliro: it is? Do you have a working Guix System on that computer (with graphical session)?
<pkill9>oh there are those new shortcuts
<pkill9>yea i get the same error when using guix upgrade
<pkill9>raingloom: it worked with `guix package --upgrade` which is the same as `guix upgrade`
<pkill9>s/worked/works
<raingloom>pkill9, yup, that one works
<quiliro>kmicu: i have a disk with a working updated GSD gnome system.....i also have this disk that is a new GSD 1.0.1 installation but it will not have good video when using gdm
<quiliro>is that a better explanation?
<raingloom>welp, this is far from consistent or smooth, guess i'll make a bug report.....
<quiliro>i can brin my disk tomorrow to test and verify versions on this hardware
<quiliro>s/brin/bring/
<kmicu>quiliro: is that working GNOME on a Guix System older than 1.0.1?
<quiliro>kmicu: it is updated
<kmicu>quiliro: updated to 1.0.1?
<quiliro>guix pull
<quiliro>sudo guix system reconfigure desktop.scm
<kmicu>And config is the same on a working system and not-working system?
<quiliro>not the same
<kmicu>Could you test that old working config on the new broken system?
<quiliro>i could copy that system's config file and reconfigure this one with htat file
<kmicu>That would be very helpful to find the issue.
<quiliro>i could do that tomorrow
<quiliro>or maybe take this disk with me and test it there today
<quiliro>kmicu: yes
<quiliro>will tell you later...bye
<kmicu>Sure, take your time.
<kmicu>Cya (*^▽^)/
<bavier>oh, someone made llvm/clang@8 packages :)
*kmicu thinks there’s a stale GDM cache somewhere in ~/.* and that could be quiliro’s issue.
<daf1234>How do I install zathura plugins with guix?
<daf1234>zathura is looking in /gnu/store/...-zathura-x.x.x/lib/zathura/
<daf1234>the one plugin i installed went under /gnu/store/...-zathura-pdf-mupdf-x.x.x/lib/zathura
<daf1234>should i create a symlink there manually or something else?
<cbaines>daf1234, the zathura package apparently works with a search path: ZATHURA_PLUGINS_PATH
<cbaines>so if you install zathura, and some plugins in the same profile, you should see that set
<cbaines>then hopefully the plugins work?
<daf1234>awesome, that works. thanks!
<cbaines>great :)
<jb55>does anyone know of a tool like diffoscope but for auditing dependency source code changes? I want something where I upgrade my deps and get a shortlog of all the changes across all of my dependencies.
*kmicu is surprised too see that https://emba.gnu.org/ exists. If Emacs can then Guix can too.
<pkill9>support for running downloaded binaries, like appimages, coming soon: https://gitlab.com/pkill-9/guix-packages-free/blob/master/pkill9/services/fhs.scm
<pkill9>coming soon to a guix near you