IRC channel logs

2020-07-13.log

back to list of logs

<hendursaga>I was wondering that too...
<hendursaga>I'm still making my way through the Guile reference manual - how would I easily modify %base-services so that I can configure network-manager-service-type, or work around it?
<hendursaga>Never mind I found an answer here: https://github.com/pmeiyu/guix-config/blob/0f4c6dcb229c8bff4e16a2da9295f859dfd6ffb5/meiyu/systems/desktop.scm
<jgart[m]>hendursaga: see here how ambrevar does it for %base-packages https://gitlab.com/ambrevar/dotfiles/-/blob/master/.config/guix/system/default.scm#L130
<jgart[m]>I imagine you can do the same for %base-services
<hendursaga>jgart[m]: Turns out modify-services is a more compact way and fits my needs, but I forgot about ambrevar's dotfiles, I've been meaning to check them out
<jgart[m]>hendursa1: thanks for pointing that out! Found ambrevar using it here: https://gitlab.com/ambrevar/dotfiles/-/blob/master/.config/guix/system/desktop-fafafa.scm
<jgart[m]>would it be useful to have a `modify-packages` macro like `modify-services` instead of doing something like this https://gitlab.com/ambrevar/dotfiles/-/blob/master/.config/guix/system/default.scm#L130 that uses a fold procedure?
<hendursaga>Probably not? It's ambrevar, he knows his stuff :)
<NieDzejkob>modify-services needs much more plumbing, while this is nothing package-specific. You could just as well call it delete-many
<jgart[m]>or maybe delete-packages to be domain-specific
<jgart[m]>so then you can do (delete-packages nano zile %base-packages wireless-tools) instead of (fold (lambda (package l) (delete package l))
<jgart[m]> %base-packages
<jgart[m]> (list nano zile wireless-tools))
<Alontra>Hello all! What's the recommended way for installing guix (the package manager) on Debian?
<roptat>The installer script is the easiest way to do it
<roptat>See the note here: http://guix.gnu.org/manual/en/html_node/Binary-Installation.html#Binary-Installation
<roptat>if you don't like using a script, you can always follow the manual steps from the same page :)
<NieDzejkob>rekado: What do you think of the v2 of #40469? You had some comments last time :P
<Alontra>Thanks, roptat!
<raghavgururajan>nckx: Thanks so much!
<sneek>Welcome back raghavgururajan, you have 3 messages!
<sneek>raghavgururajan, nckx says: Everything's finished, but gtk+ fails to build, blocking all its dependents. I'm currently rebuilding with this patch: https://paste.debian.net/plain/1156116 .
<sneek>raghavgururajan, nckx says: To be clear: gtk+ was already part of your original request, as a dependency.
<sneek>raghavgururajan, nckx says: All done. Apply https://paste.debian.net/plain/1156120 first.
<hendursaga>So, uh, I apparently must've screwed up my system. I rebooted and when I input my encryption password, it claims that it cannot find whatever it is it needs to mount. I've booted into GuixSD installation image. What do I do next? Try to mount /dev/sda manually?
<jonsger>hm. ebook-viewer of calibre is broken :(
<hendursaga>I'm wrong, thankfully. Apparently the upgrade must've modified GRUB or something so that the error it gives me when I get the first password wrong is different.
<hendursaga>On that note, however, does anyone know how to make it so you have multiple chances at your encryption password at GRUB?
<jonsger>my system is building the world. did we have a commit which triggers hundreds of rebuilds on master or is CI broken?
<jonsger>over 800 packages and drv to build
<HenryPorter>building /gnu/store/4if6rdn4fmq6k9gm0qr7n1s921mabx3i-guix-system-tests.drv...
<HenryPorter> 48% [############################################################### ]builder for `/gnu/store/4if6rdn4fmq6k9gm0qr7n1s921mabx3i-guix-system-tests.drv' failed with exit code 1
<HenryPorter>build of /gnu/store/4if6rdn4fmq6k9gm0qr7n1s921mabx3i-guix-system-tests.drv failed
<HenryPorter>View build log at '/var/log/guix/drvs/4i/f6rdn4fmq6k9gm0qr7n1s921mabx3i-guix-system-tests.drv.bz2'.
<HenryPorter>cannot build derivation `/gnu/store/521zc5vzprxjy76bzdzl32aa3dz080i8-guix-system-tests-modules.drv': 1 dependencies couldn't be built
<HenryPorter>cannot build derivation `/gnu/store/5g9hars1x2bvn22v488aqi1yrcvlgiwp-guix-bea2134fe-modules.drv': 1 dependencies couldn't be built
<HenryPorter>cannot build derivation `/gnu/store/m89gvvg6h24q8wg9qq3w4ly9v7b0pp1g-guix-bea2134fe.drv': 1 dependencies couldn't be built
<HenryPorter>cannot build derivation `/gnu/store/njd115d2kc1s103m637ll7jfkycvdl4b-profile.drv': 1 dependencies couldn't be built
<HenryPorter>guix pull: error: build of `/gnu/store/njd115d2kc1s103m637ll7jfkycvdl4b-profile.drv' failed
<HenryPorter>I get this error message when I run "guix pull".
<HenryPorter>I tried several times but it always stop at 48%.
<hendursaga>HenryPorter: Use a pastebin next time, thanks.
<HenryPorter>Sorry, I tried to send it as a pastebin now but got "(Code: 0) There was an error with TheP(aste)B.in!"
<hendursaga>HenryPorter: You used https://paste.debian.net/ for that? See /topic
<HenryPorter>I clicked on the "Upload to TheP(aste)B.in" button.
<branjam>Not near my dev computer atm. But I have the same issue building guix-system-tests, down to the %.
<mroh>yes, <patchwork-database-configuration>: unbound variable in %resolve-variable in gnu/tests/web
<mroh>commit a143e92446 i guess
<roptat>HenryPorter, try pulling b0e7b6992f3f845e83cfbca4d700b51dba50b4d5, that's one commit before the commit that causing you issues
<HenryPorter>How do I do that?
<roptat>guix pull --commit=b0e7b6992f3f845e83cfbca4d700b51dba50b4d5,
<roptat>the issue is when you run guix pull, right?
<HenryPorter>Yes
<HenryPorter>That worked, thank you!
<roptat>I just pushed a fix, hopefully that was the only issue!
<roptat>I need to go to bed, if the issue is still there, please send a report to bug-guix@gnu.org
<branjam>Appreciate you roptat!
<apteryx>mbakke: thanks for the reply
<sneek>apteryx, you have 1 message!
<sneek>apteryx, mbakke says: merging master to core-updates or staging is OK any time
<apteryx>^ that one :-)
<PotentialUser-40>Hi guys, I install keepassxc but no icon appears to open it. I'm using the GNOME desktop. Do I need to restart my machine? I didn't have to do that when I installed IcecatGNU.
<efraim>I've noticed a lack of icons on keepassxc also, I'm on enlightenment
<jonsger>PotentialUser-40: I think you need to restart to see the menu entry in GNOME
<efraim>might need to install hicolor-icon-theme
<efraim>not sure though
<PotentialUser-40>@jonsger Thank you!
<pkill9>PotentialUser-40: i think you just needto log out and log in again
<pkill9>i think the problem is GNOME resolves the real path to the desktop files, e.g. /gnu/store/...-profile/share/applications, instead of using ~/.guix-profile/share/applications, so when you update your profile, GNOME won't see the new .desktop files because it's using the store path, not the symlink
***apteryx_ is now known as apteryx
<raghavgururajan>Hello Guix!
<raghavgururajan>sneek, later tell nckx: Could you please build wpewebkit? REPO: https://git.disroot.org/raghavgururajan/guix-wip-desktop.git ; BRANCH: temp ; COMMIT: c9a90eab37
<sneek>Will do.
<efraim>raghavgururajan: I started building it on bayfront but it says the hash is wrong for wpewebkit
<raghavgururajan>efraim: Shoot! I changed the version but forgot to update the hash.
<efraim>I updated it on my bayfront checkout to what it said the hash actually was and started the build again
<raghavgururajan>Thank you!
<efraim>it also didn't like the hplip hash
<raghavgururajan>hplip?
<efraim>I see it's already 3.20.6 on master, maybe it was updated in place?
<efraim>libwpe also failed the hash for me
<raghavgururajan>Let me know the hashes you have changed. I will also update them on my end.
<nckx>Good morning, Guix.
<sneek>Welcome back nckx, you have 1 message!
<sneek>nckx, raghavgururajan says: Could you please build wpewebkit? REPO: https://git.disroot.org/raghavgururajan/guix-wip-desktop.git ; BRANCH: temp ; COMMIT: c9a90eab37
<raghavgururajan>What? I edited libwpe but did not change the version.
<raghavgururajan>nckx: Morning!
<efraim>morning!
<nckx>raghavgururajan: I had to change the hash for wpewebkit.
<raghavgururajan>nckx: Sorry about that, I changed the version but forgot to update the hash.
<nckx>That's fine, just letting you know.
<raghavgururajan>Cool!
<rov>Hello again. I am still trying to package raylib. I've managed to get my package definition to install and raylib.h is present in the proper /gnu/store/ location, but including the header in a C project fails. I've tried looking at other C library package definitions like libpng, but can't find a significant difference.
<rov>raylib.scm: https://pastebin.com/8tWqMGkv
<rov>What else could cause this problem? Or am I missing some crucial step in defining a library package?
<g-u-i-x-s-u-c-k->nckx: Miss me?
<sneek>g-u-i-x-s-u-c-k-, you have 4 messages!
<sneek>g-u-i-x-s-u-c-k-, nckx says: You misunderstand. You're not banned for saying Guix sucks (of course it sucks; it's software), just for low-effort trolling. Why does Guix suck? Why should it be on the list? Have you asked the owners to add it yet? If you can proffer criticique instead of empty memes you're welcome to stay! 🙂
<sneek>g-u-i-x-s-u-c-k-, nckx says: You misunderstand. You're not banned for saying Guix sucks (of course it sucks; it's software), just for low-effort trolling. Why does Guix suck? Why should it be on the list? Have you asked the owners to add it yet? If you can proffer criticique instead of empty memes you're welcome to stay! 🙂
<sneek>g-u-i-x-s-u-c-k-, nckx says: Being boring for attention doesn't work. Being interesting does.
<sneek>g-u-i-x-s-u-c-k-, nckx says: Being boring for attention doesn't work. Being interesting does.
<NieDzejkob>rov: could you paste the exact commands you're trying to use to use the library?
<NieDzejkob>also, try this for the tests: (add-before 'check 'set-CC (lambda _ (setenv "CC" ,(cc-for-target))))
<rov>Well, compiling a C program that #includes raylib.h fails with the error message "raylib.h: No such file or directory"
<NieDzejkob>rov: did you install the package or something?
<NieDzejkob>or is the C program also a package?
<rov>The package is installed, the C program is a project I want to use raylib for.
<nckx>raghavgururajan: I get ‘2 dependencies couldn't be built’ and don't have time to look into it now, but I'm also building it on bayfront. It's a bit slower but we'll see.
<raghavgururajan>nckx: I see. By any chance you saw the names of those deps?
<raghavgururajan>nckx: efraim already started a buid on bayfront.
<leoprikler>rov: does it work if you compile with -I/gnu/store/.../include?
<NieDzejkob>rov: works for me: https://paste.debian.net/1156209/https://paste.debian.net/1156209/
<NieDzejkob>ouch, pasted the link twice by accident
<rov>leoprikler: > does it work...
<rov>No, doesn't
<rov>NieDzejkob: Yeah, your example works, weirdly enough. I must be messing up somewhere else.
<leoprikler>what are you doing that doesn't work?
<rov>I tried following your advice of using the -I flag with the explicit path to the header.
<NieDzejkob>rov: that's why I asked for the exact commands you're using
<efraim>raghavgururajan: looks like wpewebkit finished building on bayfront
<leoprikler>not path to header, just path to include directory
<rov>> 'not path to head, just [...]' Yeah sorry, that's what I meant.
<rov>NieDzejkob: I basically just used gcc test.c -o test
<rov>I assumed running 'guix package --install-from-file=raylib.scm' should allow me to do so.
<NieDzejkob>rov: could you paste the output of 'guix package -I' ?
<NieDzejkob>(that's an uppercase i)
<rov> https://paste.debian.net/1156213/
<NieDzejkob>where's the compiler? O_o
<NieDzejkob>the programs that provides and consumes an environment variable need to be in the same profile
<rov> https://paste.debian.net/1156214/
<NieDzejkob>try restarting your shell
<NieDzejkob>it probably told you "consider running these lines to update env vars"
<rov>Okay. Still the same problem.
<NieDzejkob>what is the value of $C_INCLUDE_PATH?
<NieDzejkob>are the headers you need present in ~/.guix-profile/include ?
<rov>I ran the lines guix package --install recommends and it works :/
<rov>Should read the output more carefully :D
<ss2`>hi there.
<ss2`>I'm wondering, if there's anyone willing to help? I'm pretty new to Guix, and am trying to play a bit with the settings.
<ss2`>But so far, I just haven't got bluetooth running yet.
<ss2`>Trying to read through the docs and have put `(service bluetooth-service (auto-enable? #t)) into the services list. And that it isn't correct how I'm doing it yet.
<apteryx>ss2`: hello there! Feel free to ask, there are many people willing to help around here.
<apteryx>what do you mean by 'it isn't correct'? Does your system reconfigure OK, but bluetooth isn't usable?
<apteryx>the auto-enable? switch is used to make bluetooth active automatically when you boot your machine
<ss2`>it says: error: auto-enable?: unbound variable.
<NieDzejkob>could you link the doc page you read that on?
<ss2`> https://guix.gnu.org/manual/en/html_node/Desktop-Services.html
<apteryx>You have one parenthesis set too many: (bluetooth-service #:auto-enable? #t)
<apteryx>^ I use the above
<apteryx>if you want to use (service ...)
<ss2`>guix system reconfigure is running now. Looks good.
<apteryx>I think you'd need to pass it a (bluetooth-configuration (auto-enable? #t)) configuration object as its last argument
<ss2`>Yes, I'd like to have bluetooth load a boot.
<NieDzejkob>as in, you could have either (bluetooth-service #:auto-enable? #t), or (service bluetooth-service (bluetooth-configuration (auto-enable? #t))), as two equivalent ways of doing this
<apteryx>ss2`: also, as explained in the doc, you'll want your user to be part of the "lp" group to have access to bluetooth devices.
<apteryx>by passing (user-account (name "your-user-name) (group "users) (supplementary-groups '("lp" ...) ...)
<ss2`>I'm part of the group, and rebooted. So bluetooth must be running now, only that it throws an error: g-dbus-error-quar:.
<ss2`>I'll have to find a usb stick first to copy and paste the errors. :)
<ss2`>anyway, I've set up a laptop with Xfce, and so far, there isn't much that I've changed within that system yet.
<NieDzejkob>you don't have network access on there?!
<NieDzejkob>you could use something like termbin.com
<NieDzejkob>(nc is available in the netcat package)
<ss2`>yes I do, but haven't loaded much stuff in it yet. I'm just experimenting and trying to learn the basics of guix.
<ss2`>just realised I can't auto mount usb devices..
<efraim>if anyone was curious, python-on-guile isn't a drop-in replacement for python-wrapper in micropython
<ss2`>sorry if I've got some easy questions, but do I need to enable udisks aswell?
<ss2`>for auto-mounting in Xfce?
<apteryx>ss2`: I use 'udiskctl mount -b /dev/sdX'
<apteryx>sorry, udisksctl
<ss2`>sure!
<raghavgururajan>efraim: Yay! \o/
<raghavgururajan>efraim: Now, how do I get the substituties of wpewebkit and its dependencies, that has been built on bayfront?
<civodul>'lo!
<raghavgururajan>o/
<mmalter>Hello, it is the first time I install guix and I was wondering about the typical size requirements for /gnu. I am partitioning right now and I don't wish to use LVM.
<mmalter>I heard guix required quite a lot of space. Would 100G be considered comfortable?
<NieDzejkob>why do you want to split the partitions?
<NieDzejkob> /gnu is somewhat like a cache, the bigger you make it, the happier you'll be
<rndd[m]>Hi everyone, https://ci.guix.gnu.org/search/latest/image?query=spec:guix-master+status:success+system:x86_64-linux+hurd-barebones-disk-image returns 504
<apteryx>mmalter: it should be fine, but you'll have to run 'guix gc' every now and then
<mmalter>NieDzejkob: I see. I just wanted to be sure that if guix went berserk on the filesystem, it would not have any impact on other parts of the system. I am going to run guix on top of debian.
<mmalter>apteryx: ok that sounds reasonable
<mmalter>Thank you all!
<apteryx>you might consider using Btrfs with zstd compression, to give you about 30% extra space
<jonsger>is there a texlive meta package? or do I need to resolve every dependency by hand?
<apteryx>texlive-base IIRC
<jonsger>`texlive` provides all
<mmalter>apteryx: Damn that btrfs even supports things I only saw in ZFS. I guees I'll give it a shot. I don't really see the downside.
<blendergeek>If a build of some package is failing, how do I use my own package definition (not from the repo) so I can figure out how to fix it?
<NieDzejkob>blendergeek: for hacking on the guix channel itself, see info '(guix) Building from Git' and the following page
<NieDzejkob>though you could try to put the definition in a file and do guix build -f file.scm
<NieDzejkob>make sure that the file itself returns the package
<NieDzejkob>(i. e. put the variable name on the last line)
<blendergeek>Thank you! the 'putting the variable name on the last line' seems to have solved my problems for now. I'm still learning lisp/functional programming while try to hack on guix so this is fun.
<ss2`>alright, thanks for the help. Just figured that I'm missing some drivers. Something I'll take care of.
<ss2`>Is there maybe a tutorial around on writing some guix configs? I'm slightly stuck every now and then, although I am reading the info manual.
<raghavgururajan>nckx: When you are available, could you please build ffmpeg, qtbase, sdl, sdl2, bluez and pulseaudio? Commit is c608933f2b
<ArneBab>can I easily setup a local substitute mirror? I’d like my childhurd to not always re-download packages.
<civodul>new mailing list! https://lists.gnu.org/mailman/listinfo/guix-science
<civodul>ArneBab: ci.guix does not provide substitutes for GNU/Hurd yet
<civodul>but janneke has been working hard to make that possible
<civodul>and childhurds are here for that reason
<civodul>i think rekado added childhurds to the build farm, but now we need to figure out the Last Details™ to make that work :-)
<ArneBab>I see, yes. Can I have a package mirror?
<janneke>ArneBab: i simply add my host (laptop) to the guix-daemon's substitute-urls in my childhurd os
<janneke>i have built some, but they are not official and still change often -- in fact, if we're lucky and fix https://bugs.gnu.org/42327 that will trigger a hurd world rebuild
<ArneBab>how do you serve them?
<janneke>i run guix-publish on my laptop
<rndd[m]>Where i can write if i notice broken links on any guix web page? 0_0
<roptat>bug-guix@gnu.org
<rndd[m]>Thanks
<roptat>or is it guix-bug? I can never remember
<leoprikler>bug-guix
<apteryx>roptat: I have the same problem, but it's bug-guix ;-)
<leoprikler>but guix-patches ;)
<civodul>roptat: thanks for fixing the patchwork test yesterday, apologies!
<roptat>no problem :)
<civodul>janneke: i'm looking at your latest "guix build -s" bug report!
<roptat>We need more commiters outside of europe though :p
<civodul>true!
<janneke>civodul: oh, great! ... i noticed that the command line differs, so it /should/ be possible to avoid a (non-Hurd) world...
***florian1 is now known as pelzflorian
<janneke>... rebuild
<pelzflorian>hi this is a test
<pelzflorian>it works i am online
<janneke>civodul: ah, great -- i didn't realise it was a "real" bug; beautiful!!!
*janneke feels confident to praise before testing...going to do that right now!
***florian1 is now known as pelzflorian
*jackhill subscribes to guix-science
<NieDzejkob>raghavgururajan, nckx: I've started the build of those packages on bayfront
<raghavgururajan>NieDzejkob, Thank you! How do I use substitute from bayfront?
<raghavgururajan>I have ssh access, but not sure how transfer substitutes to my machine.
***sneek_ is now known as sneek
<janneke>civodul: $ guix build --system=i586-gnu hello
<janneke>/gnu/store/803q5wapfnmr91ag8d9dzwabkbdxz3ay-hello-2.10
<janneke>ootb, on master (with cheated substitutes -- but that's also nice, right?
<janneke>i remember having to attempt "gawk" more than a few times, but yeah...
*janneke goes afk for a bit
<NieDzejkob>raghavgururajan: "ssh bayfront.guix.gnu.org cat /etc/guix/signing-key.pub | sudo guix archive --authorize", then "guix build --substitute-urls='https://bayfront.guix.gnu.org' stuff" has worked for me before
<raghavgururajan>NieDzejkob, Perfect!
<raghavgururajan>NieDzejkob, Does it matter who builds it? I mean all the /gnu/store is published?
<NieDzejkob>yeah
<NieDzejkob>only qtbase is left now. and maybe its dependants
<PotentialUser-40>Is the tor browser not available in the guix store. I only see the "tor" package, but that doesn't seem to be the browser.
<leoprikler>No, it's not
<raghavgururajan>NieDzejkob, I still get .0% substitutes available (0 out of 1) for ffmpeg
<PotentialUser-40>@leoprikler Thank you.
<roptat>PotentialUser-40, we don't have the tor browser packaged in guix, the closest you can get is to configure icecat to use tor, in the proxy settings
<roptat>if you're on the guix system, you'll also need to configure the tor service
<NieDzejkob>raghavgururajan: builds for ffmpeg and sdl2 failed, both due to a hash mismatch for hplip
<NieDzejkob>would you have any trouble with starting the follow-up build on bayfront on your own?
<hendursaga>PotentialUser-40: Just download TBB from an official source.
<blendergeek>I am trying to build the Nextcloud desktop client for GNU Guix. The cmake build system complained that it wanted QTWebEngine. After giving it that, Qt5WebEngine says " Could not find a package configuration file provided by "Qt5Quick"
<blendergeek> (requested version 5.14.2) with any of the following names:
<blendergeek> Qt5QuickConfig.cmake
<blendergeek> qt5quick-config.cmake"
<hendursaga> https://lists.gnu.org/archive/html/guix-science gives a 404 error, is there no archives or just (temporarily?) down?
<raghavgururajan>NieDzejkob: I would like to bootstrap the cloned guix repo, on bayfront. I usually do it inside the env with guix and emacs. How do I make sure the references for this bootstarap won't get garbage collected?
<raghavgururajan>Like /pre-inst-env stops working after GC process.
<leoprikler>blendergeek: you probably want qtdeclarative as an input as well
<nckx>NieDzejkob: Thank you. raghavgururajan: Have you tried building things on bayfront again? It seems more usable now. Would save you a half-day round-trip time.
<jackhill>blendergeek: did you try qtquickcontrols or qtquickcontrols2? The latter has a Qt5QuickControls2.pc pkg-config file.
<jackhill>oh, I now see leoprikler already replied with a more accurate answer :)
<efraim>I believe hplip's source tarball may have been changed in place upstream
<NieDzejkob>raghavgururajan: you can do guix environment guix -p ~/.guix-env to register a GC root
<blendergeek>leoprikler: Thank you! That seems to have solved the qtquick problem!
<efraim>ok, I keep on getting that the hplip tarballs are identical
<efraim>raghavgururajan: go ahead and cherry-pick ef0e7a5d332f0947adb3556cc5a8c8836eac0b6c from guix master to your branch
<drakonis>i recall hearing about guix using sticky bit for the binary
<drakonis>so it can actually not require the daemon for builds
<mfg>Hi guix :)
<ramajd>I have two custom channels on my machine, one for my personal packages, and the other one for a series of corporate packages. now, when I want to use a package located in my personal channel as an input for another package located in corporate channel, I receive error during `guix pull`, indicating "no code for module ~S". can anyone help me on this matter?
<efraim>ramajd: you need a file like this http://git.genenetwork.org/guix-bioinformatics/guix-bioinformatics/src/branch/master/.guix-channel in your corporate channel so it will "pull in" your personal channel during guix pull
<ramajd>efraim: I'll give this a try, thank you
<blendergeek>Is there documentation for the 'search-patches' function that 'patches' seems to use in the origin for the source to a package?
<cbaines>blendergeek, there's some documentation with the definitions: http://git.savannah.gnu.org/cgit/guix.git/tree/gnu/packages.scm#n92
<pkill9>could guix eventually run reasonably well on low-power hardware like raspberry pi?
<pkill9>since guix's computations take a long time on that I find
<pkill9>is that a limitation of guile, or can it be optimized?
<drakonis>depends on which rpi you're running it on
<vagrantc>i've run it on similarly underpowered hardware...
<vagrantc>it's not fun ... offloading is pretty essential
<drakonis>guix takes a while even on beefy hardware
<pkill9>yea
<pkill9>how much can that be improved?
<drakonis>imagine a dinky rpi
<pkill9>without upgrading hardware
<vagrantc>but it's technically possible for the patient with offloading and limited expectations about what to install
<drakonis>guile 3, yes?
<vagrantc>even with offloading, a fair amount of computation happens locally just for guix pul
<drakonis>rather, it'd be like running gentoo all again
<pkill9>yea the guix pull takes a while
<pkill9>the computation part that is
<mmalter>what is so long? solving the graph?
<apteryx>I think one of the think taking a a long time is the translation process of the info pages. It is a slow, sequential process as noted by civodul in previous conversations.
<apteryx>another think is simply loading the many files that make up guix
<apteryx>if it was all compiled down to a single binary it could be much faster. This would need native compilation work on the Guile side.
<blendergeek>cbaines: Thank you! I was able to read the documentation and figure out how to make it work!
<cbaines>blendergeek, that's good :)
<pkill9>has anyone built rust's cargo nightly version?
<pkill9>i am trying to build something that requires cargo nightly
<spudpnds>So I'm totally a guix noob, but an avid emacs user, I'm curious, do folks manage their melpa packages with guix? If so is there an example config I can look at for this?
<leoprikler>Yes. Some packages are already included in Guix proper, others you can get through guix import
<leoprikler>You can generate a list of packages on the fly by stuffing (elpa->guix-package "my-package" 'melpa) into eval
<spudpnds>I see that in the invoking-guix-import doc.
<leoprikler>Or rather a single package – for the list you need to wrap that in a map.
<leoprikler>like so:
<leoprikler>(map (cute eval <> (current-module)) `(,(elpa->guix...) ,(elpa->guix...))
<spudpnds>*woosh* I'll come back once I've learned a bit of scheme.
<spudpnds>;)
<leoprikler>It's quite advanced to be fair
<spudpnds>I think I'll continue to manage my emacs packages with use-package until I get more advanced.
<leoprikler>If you want it to be simple, you just copy the output of `guix import` into a file and call it a day.
<leoprikler>e.g. `guix import -a melpa use-package > use-package.scm; guix install -f use-package.scm`
<leoprikler>oops, guix import elpa of course :)
<hendursaga>spudpnds: You can also check out straight.el
<spudpnds>hendursaga: cool, I know a lot of folks are switching to it. I've yet to be burned by a melpa package update gone awry. ... yet
<blendergeek>I just finally got nextcloud built for Guix but a unit test is failing. Is there an easy way to tell Guix to ignore the failing unit tests?
<leoprikler>Well, there is a way, but it may or may not be easy depending on the build system.
<leoprikler>Basically, you patch out the call to the breaking test-cases. I think lots of packages have a 'fix-breaking-tests phase for that.
<lfam>You should also consider the nature of the test failure. It might indicate that something is really wrong
<leoprikler>Okay, it's 'disable-broken-test(s)
<Guest87457>Hello
<bdju>so I've updated minetest without rebooting but if launched from wofi, it launches an older version. I seem to have to launch it from the shell where I updated packages. is there a way to fix that without logging out or rebooting?
<vagrantc>wofi?
<bdju>it's like rofi/dmenu, an application launcher
<bdju>but for wayland
<mfg>bdju: i'd guess this isn't possible, because shells only inherit environments and yo can't alter the environment of parent shells...
<lfam>It works for dmenu
<lfam>I assume that wofi caches things and you'll need to look up how to make it refresh the cache
<lfam>It *should* work because PATH should contain entries like '~/.guix-profile/bin/foo', and when you upgrade foo, that path remains the same
<lfam>But it's likely that wofi is caching the symlink resolution and not noticing that it changed
<Guest87457>I can't set my resolution past 720p. Is there some driver I need to install? If so how do I do that. I'm using an NVida card btw.
<lfam>What did you try Guest87457?
<leoprikler>Which card to be exact?
<Guest87457>How do I find out the exact card. It doesn't show in "details".
<Guest87457>lspci?
<leoprikler>yep
<Guest87457>VGA compatible controller: NVIDIA Corporation GP108 [GeForce GT 1030] (rev a1)
<leoprikler>Hmm, that's a rather new one.
<Guest87457>Oh shoot...
<Guest87457>should I have bought an older one?
<leoprikler>Well, I get to 1080p with my GT 710 (NV106)
<leoprikler>can you paste the output of xrandr?
<leoprikler>(if you haven't installed it yet `guix environment --ad-hoc xrandr -- xrandr`)
<leoprikler>Okay, so having a look at the feature matrix, GP108 (which should translate to NV138) should still be supported
<leoprikler> https://nouveau.freedesktop.org/wiki/FeatureMatrix/
***calher1 is now known as KE0VVT
<slimjim>hi all, I'm a new user trying out Guix on a foreign distro, I'm really excited about it but I've had some trouble getting a few of the features to work - is this an appropriate place to ask for help?
<qyliss>Yes :)
<slimjim>great!
<slimjim>so one of my main interests is 'guix pack' as a way to distribute precanned environments to coworkers
<slimjim>so I was trying something with some of the python packages. I got it working in docker but having trouble with squashfs/singularity
<slimjim>so, if I generate a pack with this:
<slimjim>guix pack -f docker python2-cryptography python2 bash coreutils -S /bin=bin
<slimjim>load it in docker and then docker run -ti <image> /bin/python2
<slimjim>I can "import cryptography" fine
<slimjim>but if I guix pack -f squashfs python2-cryptography python2 bash coreutils
<slimjim>then singularity exec PATH /bin/python2
<slimjim>python2 crashes "No module named sysconfig"
<Guest87457>@leoprikler so I just type "xrandr" in the terminal?
<civodul>slimjim: weird, is it Singularity 2.x or 3.x?
<leoprikler>either that (if you have it installed) or the full guix environment thing if you don't
<Guest87457>?
<Guest87457>okay installing xrandr
<Guest87457>what was the paste link again?
<leoprikler>you can use any paste you like
<leoprikler>we here mostly you debpaste tho
<Guest87457>okay can you give me the link?
<slimjim>civodul: well, I was using the singularity that comes with guix, which is 2.6.1, but I was having trouble running that on my host at all (Ubuntu 18.04 x86_64)
<leoprikler> http://paste.debian.net/
<slimjim>so I tried singularity from the ubuntu mirrors, which is 2.4.2
<Guest87457> https://paste.debian.net/1156279/
<civodul>slimjim: ok, for my part i tested mostly with 2.6 and 3.0
<leoprikler>Hmm, so it recognizes the availability of those larger sizes
<leoprikler>what happens if you try to set it to 1920x1080?
<Guest87457>It logs me out.
<Guest87457>And doesn't adjust the resolution.
<slimjim>civodul: when trying to run with the singularity in guix, I get "Failed to resolve path to /var/singularity/mnt/container: No such file or directory" trying to run any singularity command with the generated iamge
<civodul>ok
<Guest87457>I have it hooked up to 4k off brand TV via HDMI. Is maybe that effecting it?
<civodul>slimjim: for the record we have a system test here that tests an image from "guix pack -f squashfs" against singularity 2.6: https://git.savannah.gnu.org/cgit/guix.git/tree/gnu/tests/singularity.scm
<civodul>could you report the issue to bug-guix@gnu.org?
<civodul>make sure to specify the versions you use and the command+output of "singularity exec"
<slimjim>ok. I tried singularity exec /bin/python2, and I also tried building the image with guix -f singularity --entry-point=bin/python2 and then singularity run
<leoprikler>Guest87457: Assuming your TV works like a typical monitor and doesn't do anything malicious on its own, it should not make a difference.
<leoprikler>Then again, I'm not very trusting of TVs.
<lfam>Guest87457: How are you trying to change the resolution?
<Guest87457>I have a monitor I can test it on.
<lfam>I mean, what command you using?
<lfam>What program?
<Guest87457>Via Displays>Resolution in the Gui
<lfam>In GNOME?
<slimjim>civodul: I also couldn't get a relocatable tarball with "pack -R" to work. I think the problem in both the squashfs + relocatable tarball might stem from a failure to set PYTHONPATH appropriately
<Guest87457>yes, gnome, but I have had the same issue using XKDE
<slimjim>the docker result is the only one that actually seems to have the correct PYTHONPATH
<lfam>You should try using xrandr directly
<Guest87457>oh okay, how would I do that?
<civodul>slimjim: ah, -R and -RR are mostly useful for raw tarballs
<civodul>you can pass it alongside with '-f squashfs', but it doesn't bring anything
<slimjim>civodul: yeah that's what I was doing with -R, I was trying to test the raw tarball too
<slimjim>I wasn't using it with squashfs
<civodul>ah sorry
<slimjim>so docker works, but I'd really like to use singularity so I can avoid the "using docker is basically the same as being root" problem
<leoprikler>xrandr --size WxH
<civodul>slimjim: ah yes, for the raw tarball, you have to explicitly ". ./etc/profile" after unpacking so that PYTHONPATH & co. are set
<civodul>and you probably need "-S /etc=etc" too
<lfam>I'm not sure what the difference is between --size and --mode, but I always use --mode
<leoprikler>mode has autocompletion :)
<leoprikler>and also requires --output
<leoprikler>btw. make sure to keep an eye on /var/log/messages as you do so
<slimjim>civodul: ooo ok, adding etc and sourcing the profile got it working in the tarball -R case
<leoprikler>(ideally connect from a different pc via ssh and use tail -f)
<slimjim>so now it's just singularity I'm having issues with
<slimjim>and I can submit a but to bug-guix@gnu.org, is there a particulary format you're looking for?
<civodul>slimjim: no, just make sure it fully describes the issue :-)
<civodul>i think the -R case would deserve some fix too
*civodul -> zZz
<civodul>later!
<mroh>civodul: o/
<hendursaga>Where are the source files/tarballs stored for downloaded/installed packages, as a shortcut?
<leoprikler>all files land in /gnu/store or whatever else you configure the store directory as
<Guest87457>@leoprikler when I entered this command "sudo xrandr --output --mode 3840x2160" it gave me the error "xrandr: unrecognized option '3840x2160'"
<leoprikler>no sudo
<Guest87457>Okay, I removed sudo, got the same error.
<lfam>You need to specify the output