<Parra>mbakke: I see now what you mean, it's a graft <raghav-gururajan>Folks! I get this error while building a package and I am not sure what it means. <raghav-gururajan>(/tmp/guix-build-tracker-new-2.3.1.drv-0/build/tests/libtracker-miner/tracker-miner-fs-test:3845): Tracker-WARNING **: 22:53:06.862: Parent 'file:///tmp/guix-build-tracker-new-2.3.1.drv-0/tracker-miner-fs-test-6F9SE0/recursive/4' not indexed yet <mbakke>Parra: right, not sure if we can get packs to ignore the non-grafts... can you file a bug report? <Parra>anyway, I think it is not the best thing to avoid grafts, because of security issues <mbakke>Parra: that is correct, using --no-grafts is unsafe <guixy>Has anyone here had success with the guix system on a bananapi m2u? <jonsger>raw-content is your friend with nginx :) *jonsger is happy about his slow but steady progress on nextcloud :) <alextee[m]>(to update from a stable release to the next rc) <alextee[m]>also ping to anyone with commit access available to merge/review my patches for helm, vl1-emulator and regrader :-) ***catonano_ is now known as catonano
<Gooberpatrol66>I'm getting "cc: command not found" when compiling a package. Isn't gcc supposed to implement the cc command? <alextee[m]>either set an environment variable before the configure phase or add it to the make flags <apteryx>I'd prefer the make flags approach (that's more succint than defining a custom phase). <apteryx>could we replace the base inputs used in builds (%final-inputs from (gnu packages commencement)) by their multi core enabled variants? (e.g. gzip --> pigz, xz -> pixz, etc) ? Seems a low hanging fruit to get a good multi core boost. <apteryx>(I was just looking at inkscape source tar.xz being decompressed at very low speed, maxing out a single core) <vagrantc>hrm... make[3]: *** No rule to make target 'gnu/packages/patches/mrustc-0.8.0-fix-variable-length-integer-receiving.patch', needed by 'all-am'. Stop. <vagrantc>and here i was hoping to have a frenzy of typo fixes ... now it looks like real work :) <lfam>I notice the manual section Binary Installation no longer suggests sourcing the profile in '~/.guix-profile/etc/profile'. Instead, it suggests sourcing '~/.config/guix/current/etc/profile' <lfam>Does that somehow get installed executables into $PATH ? <lfam>Or is this handled some other way now? <marusich>lfam, you can source either one, but the ~/.config/guix/current profile is where "guix pull" installs new versions of Guix. <marusich>By sourcing that one, you will get Guix tools on your PATH. <lfam>But not executables from anything in the other profile <marusich>I have to run - but you can source them both individually (or not) depending on your needs. Most people will probably want to source both. <zig>ArneBab: I know it is not ideal solution, but you can always call: chroot nodejs somethign... <nixo_>guix days are approaching :)) <nixo_>I just upgraded guix on a pc that have been off for some months: 1,771 new packages, 2,522 packages upgraded. Good job <civodul>nixo_: heh, it's a good way to realize just how much work has been done :-) <nixo_>civodul: yes, but then you realize from this picture it's missing the effort on binary bootstrap, making guix faster, upcoming upgrade to guile3... It's simply amazing how fast this is moving <civodul>it would seem that the 'build' phase of "guile3.0-guix" is 14% slower than that of "guix" <lxsameer>hey folks, is it possible to use guix client to do the daemon job as well ? or is there any way to run guix as a standalone tool without the daemon ? <NieDzejkob>Whoops, I just noticed that submitting packages has become routine enough for me to forget to run `guix lint'. I wonder whether it would be possible to write a pre-commit hook that would detect this <leoprikler>you can even write your hook in guile using the exec trick <htgoebel1>OT to emacs-users: Is there some integration from VCS/magit with debbugs? <htgoebel1>I want to view the log and select which patch(es) should be send to guix-patches. <htgoebel1>I currently use a shell-script, but this is still very cumbersome. <bricewge>Does the CI doesn't build test? I can't find them in the log for libtorrent-rasterbar. <nixo_>uhm, it seems like guix gc --verify=repair,contents isn't working as it should. I have a file that I know is broken (different hash on a different machine and size 0) which isn't fixed <nixo_>also, since it is alive, I cannot force delete it. Any idea? <bricewge>nixo_: I had a similar issue some days ago. I ended up reinstalling Guix... <nixo_>bricewge: uh, I hope to find an alternative. Right now I removed the broken file, and guix gc found & fixed it. But I can't be sure this was the only one <NieDzejkob>htgoebel1: I'm not an emacs user, but see `guix show emacs-debbugs' <nixo_>damn, I should have tried guix build --repair, if that worked I could have run it on everything <smithras>nixo_: does --verify=contents find the errors at least? <bricewge>I hope you find a solution other than reinstalling that way I would know how to fix it next time <nixo_>smithras: nope, it didn't report it <nixo_>but when removing the file it said path `/gnu/store/rvbn66a34vdy7hd52l4mlr6mbjdzdagw-cairo-1.16.0' was modified! expected hash `8353a17116adb4761a830efe2fc9dab1c075bae24387f72b88ce4b8f12463df8', got `6f827ed38b498bba7 <nixo_>f863f1deade256a5f056f8de2c8e1a49f99b14960019203' <nixo_>so, it did actually check the hash (and not the file existence, according to the message) <bricewge>Why does `guix build openssl` gave me `/gnu/store/...-openssl-1.1.1d` will building a package with openssl as an input uses `/gnu/store/...-openssl-1.1.1c`? <leoprikler>guix uses grafts for important security updates such as openssl-1.1.1d over 1.1.1c <leoprikler>basically this minimizes the amount of rebuilds you'll have to do <bricewge>I'm confused.I have some tests related to ssl that are failing and I'm wondering if it's the cause. <bricewge>leoprikler: But I'm building with --no-grafts <htgoebel1>niedzejkob: I know there is a `debbugs` package. I just want to know whether this integrates with VCS/magit. <htgoebel1>Trying to figure this out may take long - esp. it is known to not work. <htgoebel1>But I'd better as on help-debbugs@gnu.org, which I just discovered. <str1ngs>htgoebel1: you can use it to apply patches. that much I know <refpga>Hello, can I put my custom package definitions in config.scm and install them? <sneek>Welcome back refpga, you have 1 message. <htgoebel1>niedzejkob, str1ngs: I discovered that in magit I can at lease easily select the patches to be formatted, decide on a cover letter, etc. This is a HUGE step forward to wait I'm doing now. <htgoebel1>Now the only thing missing is how to actually send these patches easily. <str1ngs>htgoebel1: I use magit to generate patches and git send-email. <htgoebel1>str1ng: So git send-patches on the command line? <htgoebel1>How do you handle consecutive patches? With `--thread` and pass all patch-files on one call of `git send-patches`? <htgoebel1>(I'm currently sending the cover-letter, wait for the confirmation, and then send the remaining file to this bug-id. Much hazzle!) <nixo_>bricewge, smithras: sent an email with more details on my failure to guix-devel <str1ngs>htgoebel1: I send the first in the series then I send the rest in reply to the message ID returned from sending the first one. <str1ngs>htgoebel1: I don't bother with a cover letter. I try to make the patches speak for them selves <str1ngs>htgoebel1: I'm curious how to refine the process though <refpga>How can I redefine a package provided in guix packages tree? I've modified the definition but where do I put it so that the package specification in my operating-system slot in config.scm picks it up? <str1ngs>refpga: did you modify it in the tree or outside? <str1ngs>export a GUIX_PACKAGE_PATH . then you can import it as a module <refpga>I thought I'd just put it in the config.scm file above operating-system and then use it? <refpga>I mean, put the package definition. <str1ngs>but just keep in mind in will not be available to the command line guix <refpga>It doesn't update the already installed package though. <str1ngs>generally a common pitfall is to put packages in config.scm. that really you should just install into your profile <str1ngs>then it's find in config.scm. just use guix system reconfigure to update system packages <refpga>I can't install a theme for it that is outside sddm's tree. So I forked the repo and added the theme in there, and now I'm installing my version. <str1ngs>sddm has sddm-configuration where you can set the theme directory <htgoebel1>str1ng: I'll try out `--thread=shallow` next time. This will make git to create and chain message-ids <htgoebel1>(I just learned about this, never tried it out) <refpga>str1ngs: I have set the theme directory, but it just refuses to accept the Main.qml provided in that directory. The same error you get when trying to execute a file from a mounted drive on linux, "No such file or directory". I checked manually the file is there. <refpga>I don't know what guix does with the file system. Everything in .guix-profile is read only. <str1ngs>I'm not familiar enough with lddm to helpful refpga <refpga>So I guessed maybe sddm wants theme files on that read only file-system only. <NieDzejkob>refpga: The "No such file or directory" error you get when executing a non-guix binary might be because the kernel can't find the ld.so at the path specified by the binary <PotentialUser-73>Hello, I'm trying to run GuixSD in a qemu VM. I've problem having access to the network using network bridges. Is there a tutorial somewhere that helps using this networking solution ? <g_bor[m]>PotentialUser-73: are you trying to do this using libvirt? <PotentialUser-73>I would expect that using the NAT solution would give me access to the Internet too on the guest machine <g_bor[m]>What is the current network configuration of the guest? Here I have one interface on the 192.168.122 default network, I adjusted the dhcp range, and configured a static address outside that range. <g_bor[m]>I also have a bridged setup at another place, but as I am vacation I currently have no access. <PotentialUser-73>Well, thx for your vacation time then :-) I will look at jamielinux.com that apparently has a doc on libvirt and networking <bricewge>How do you usually find out if a package uses networking in their tests? <bricewge>It looks like I was fighting for some hours to pass tests that were trying in fact trying to access Internet... <g_bor[m]>PotentialUser-73: sorry I could not help. My setup is actually identical. Really might be some proxy. <g_bor[m]>liberdiko: I don't think there is a general rule you can apply here. I usually find it out reading the test logs. <bricewge>g_bor: Argh, I the answer I didn't wanted. Thanks (: <g_bor[m]>liberdiko: there might be something that I don't know tough... If you find out a better way I would be happy to hear it. <g_bor[m]>PotentialUser-73: I glad that you found out. <refpga>Hello, how can I see the hashed name of the package that is installed via config.scm? <jlicht>refpga: As a last resort, you could put an expression that evaluates to your package definition at the end of your config.scm, and simply do `guix build -f config.scm'. Don't forget to remove the expression afterwards though :-) <smithras>refpga: I think the emacs-guix interface has this information, but I"m not 100% <pkill9>does anyone know how to configure a laptop's bootloader (i think) to load an existing installation of guix from a hard drive? ive put in the hard drive of one laptop into another laptop, containing an installation of guix, and i want to run it <pkill9>the laptop ive put the hard drive into is a thinkpad x220 tablet <refpga>smithras: It does, but only for package in the profile of the user, not the packages defined through config.scm. I could install it, and see which one gets installed though, I think this would be hard if the package was older than the current release and I wanted to know the hashed name. <refpga>jlicht:I thought of doing that but I guess I did the right thing and asked on help-guix. <refpga>pkill9: Is your current OS guixSD? <pkill9>refpga: no i just want to run an already-installed guix system installation on a different laptop <pkill9>but i know nothing of the boot process, other than that the bootloader needds to be updated or something <pkill9>might just reinstall guix system <refpga>You can try booting into that secondary hard drive using BIOS options, if you just want to run it. ***ng0_ is now known as ng0
<pkill9>i dont think that can be done with uefi <refpga>I don't think UEFI causes any trouble. <pkill9>it loads grub, and grub says 'grub_file_filters' not found <refpga>You can't boot into any other disk like you would into a usb disk to install an OS? <refpga>I'm talking about the process before loading Grub <refpga>grub is installed on a disk. You probably have grub installed on both the disk (check that). <apteryx>weird. `find_library(some-var croco) ' in a CMakeLists.txt doesn't seem to be able to find libcroco.so. libcroco is in LIBRARY_PATH. <ng0>htgoebel1: out of curiosity, how's the Plasma on GuixSD story progressing? <ng0>anything new since early 2019? <pkill9>i managed to install the bootloader with grub-install <pkill9>it boots into a grub commandline with no errors <joshuaBPMan>Hello, I'm currently making a guile web website. I am currently trying to get the website to email me when someone submits a form. Any recommendation how to do this via guile? I suppose I could pass the info onto something like sendmail... <epl692>Hello, New guix user here, wondering if someone can help me track down the problem I'm having with installing the system <epl692>During the install, everything seems to roll along just fine, then it will hit a "slow" package and stop <ng0>what drakonis means is, the name and maybe the error message would help more <epl692>The download seems to stall for a package download. I am running in a virtualbox vm to try to get familiar before I try it on bare hardware. I had similar issues while trying to install within a XCP-ng virtual machines <joshuaBPMan>epl692: We need a bit more details. What are you installing on? A laptop? What kind? What model year? <epl692>current install is stalled on exo <epl692>The download goes pretty well on some package, and slow on others, eventially the connection speed will show "3KiB/s" and just stop. I am seeing some network activity on the status indicator <bavier>epl692: you could maybe stop and restart <bavier>fwiw the "exo" package just now substituted for me at 763KiB/s <epl692>I'm restarting it and trying again, this time with no window manager <drakonis1>there's no reason to remove the window manager <pkill9>and just guesswork, and it worked! <pkill9>booted my installation of guix system on another laptop <pkill9>i just ran 'grub-install' with the root directory on the boot partition, with sone flags i cant remenber, which overwrite the bootloader that pointed to a fresh installation of grub, then i copied 'grubx64.efi' from the guix installation of grub into that fresh installation of grub <bandali>hey guix, i forget, can guix weather check whether a specific package has been already built on substitute servers? <bandali>(don’t have a browser handy right now) <pkill9>bandali: `guix weather --manifest` <bavier>bandali: or `guix weather <package-name> ...` <bandali>pkill9, bavier, will try those; thank you both! <gnutec>Is anyone with problem after upgrade to kernel 5.4.13? <lfam>What kind of problem gnutec? <gnutec>lfam: The boot stop in ntpd. And the gdm does not start. <gnutec>lfam: The Guix System always show this "error" but the gnome start anyway. ***MinceR_ is now known as MinceR
<gnutec>I reboot the system, but nothing. I can't boot with kernel 5.4.13. <joshuaBPMan>Is it possible for guile to run a command...like tell it to run curl... to send an email? <enderby>hi, wondering if there's a way to always output "guix pull --news" while still actually pulling? <pkill9>is anyone going to get an mnt reform? <bavier>pkill9: I've been waiting for it to launch on crowdsupply for a while <pkill9>it looks like what i would like out of a laptop, probably won't be worth it for me to buy though <bavier>I was personally hoping the eoma68 laptop would fulfill most of the same needs. <pkill9>i was hopeful for the eoma68, but it's been delayed so many times, and the first cards are underpowered for daily use <joshuaBPMan>I don't know what they were thinking about the scroll wheel... <pkill9>they'll have touchpad available they said <joshuaBPMan>Also, I am considering adding a section to the guix cookbook about how one can try out sway for the first time. Do ya'll think this would be worth it? <pkill9>i think these projects for providing linux-friendly hardware out of the box will do good for pushing FOSS forward <pkill9>it will iron out a lot of nuisances with running linux <jonsger>nice. XMPP in chatty works. That's the mobile client for XMPP and SMS for the Librem5 :) <jonsger>testing SMS will be a bit more tricky <jojoz[m]>No one's responded to my patch-thread in over a week. Is this normal? Seriously asking, because I have no idea how many maintainers there are and what kind of workload-burden they have. <jojoz[m]>I'm concerned it has fallen into the ether, but maybe people are just busy <zig>I have seen worse :) <lfam>jojoz[m]: Which ticket is it? <pkill9>jojoz[m]: i have a patch that's fallen into the ether <lfam>I'm polishing and testing your patches now jojoz[m] <lfam>In general, there is a huge number of patches submitted, and it's really hard to keep up the pace of patch review <lfam>The project has grown considerably over the past few years and the growth doesn't seem to be slowing down <jojoz[m]>I suppose these things can happen. Oh, how burdensome the success! ;) <lfam>It's no guarantee of quick patch review, but it helps a lot if contributors follow the directions in the manual section Contributing. Even the suggestion to run `guix lint`, which mainly catches trivial formatting issues, can help speed patch review. Because if reviewers see that the submissions were not linted it can make us worry that there are deeper problems; for example, that the packages don't even build or something like that <lfam>In general, it's fun to help new contributors by handling these minor issues ourselves, teaching them how to improve their work and eventually be able to review patches themselves if they want. But sometimes it can feel like a drag if one is not in the right mood <lfam>jojoz[m]: I'm not familiar with Haskell. For ghc-llvm-hs, would "General purpose LLVM bindings for Haskell" be an okay synopsis? I akso because "General purpose LLVM bindings" seems a little to nonspecific <jojoz[m]>lfam: Oh sorry, must've missed that point about linting. I did verify that it built and worked though. I'm using this branch of guix for building my own compiler actually :3 <lfam>Also, what is "ADT" in this context? <lfam>Awesome, I'm glad to hear it :) Adding notes like that in the submission can help reviewers feel confident about reviewing things outside of their expertise <jojoz[m]>Yeah, I kind of figured there wouldn't be Haskellers among the maintainers, since Nix seems to have mostly captured that crowd. <lfam>Obviously in this case my review is pretty superficial, but if you are using the software and it works I don't feel that it's necessary to go deeper. If there are some problems with the functionality they can also be fixed later. I'm mainly checking for obvious issues like, is it the most recent release? Is the license appropriate? Are there any unused dependencies listed? etc <lfam>Well perhaps but the haskell-xyz module is growing very long :) <sirgazil>sneek: later tell joshuaBPMan I'd be interested in the information about trying out sway (either as a recipe in the cookbook or as an entry somewhere else). <jonsger>sirgazil: just add sway to lightweight-desktop.tmpl and try it <jojoz[m]>lfam: That I suppose it has! I checked in about a year ago, and back then my compiler was faaaar from buildable in Guix. Many packages missing, and the rest were old versions. Someone's been contributing quite a bit of Haskell recently it seems. <sirgazil>jonsger: What's lightweight-desktop.tmpl? <jonsger>sirgazil: it lays in the guix repo: gnu/system/examples/lightweight-desktop.tmpl there are also some other intersting ones... ***ng0_ is now known as ng0
<sirgazil>jonsger: I see, but that's not what I'm looking for. I use GNOME, but want to create another user and set sway for it to try it out. I don't know if GDM in the Guix System would just display Sway as an option when logging in or if you need to configure something. <sirgazil>That's why I'm interested in joshua's information. <gnutec>Am I the only one with problems of kernel 5.4.13? <smithras>gnutec: I would update and try, but I know that there's an problem with the latest upower-daemon so I'd rather not update my main machine <smithras>gnutec: actually let me try updating my thinkpad, I'll let you know if things go bad <gnutec>smithras: I can access guix system using kernel 5.4.12. So I can wait the next update. <nckx>gnutec: What problems exactly? <smithras>gnutec: besides the upower issue, everything seems normal on my thinkpad with 5.4.13 <sirgazil>nckx: gnutec said earlier that, "The boot stop in ntpd. And the gdm does not start." <pkill9>now that i have a laptop again, i can return to hacking guix <nckx>gnutec: You can ignore those messages, they don't seem relevant. So the only real problem is that gdm won't start. Why do you suspect the kernel and not, say, a gdm update or other parts of GNOME? <lfam>A polkit timeout, more precisely <nckx>s/messages/ntp messages/ <nckx>lfam: Do you know where/if these polkit/dbus/… errors would be logged? <lfam>I don't have any graphical Guix systems running or any GDM / GNOME systems at all <nckx>gnutec: Neither do I. Sorry, I don't use the G-desktop ☹ <lfam>Maybe soon though! It's not judgement on them from my end <gnutec>nckx: I don't. Is gdm was update too? I don't pay attention to that. <nckx>sirgazil: Sneek is polite and won't answer that. <nckx>I don't know, I only know that Raghav (maybe others) has done some work on GNOME. <sirgazil>sneek: GNOME is GNU's desktop environment. <nckx>gnutec: Doesn't that only print human-written announcement? A package update isn't a --news-worthy change. <nckx>sneek: later tell sneek: botsnakc <sirgazil>nckx: will sneek understand Russian, though? <nckx>Those scary ‘Russian bots’ you heard about during the US election? <gnutec>nckx: This is to much for me. :P *nckx → off to eat some nckxsnakc. <pkill9>nckx: it shows a list of all added and updated packages *smithras looks suspiciously at sneek *civodul prepares to push (guix self) on guile-3.0 <jlicht>I remember this not being as trivial as I hoped it would be in the past, but how could make use of e.g. guile-json in a phase? (So on the build-side of everything) <jlicht>nvm, from a discussion on the ML I noticed this will only be convenient to do once we use gexps icw 'with-extensions' <civodul>it's indeed convenient with gexps, much less with current packages <jlicht>civodul: yeah, I got stuck and then I suddenly remembered that I read something about my exact issue :-) ***sneek_ is now known as sneek
<sirgazil>sneek: Welcome back! You have lots of messages. <sneek>GNOME is the GNU desktop environment. <sneek>Last time I checked ed is the standard text editor! Ed! Ed! Ed! <NieDzejkob>sneek: later tell NieDzejkob: Now I want to reverse-engineer this parser. <sneek>NieDzejkob, you have 1 message. <sneek>NieDzejkob, NieDzejkob says: Now I want to reverse-engineer this parser. <NieDzejkob>gah! What if someone wants to use sneek to tell someone else about botsnacks!? <sneek>NieDzejkob, sirgazil says: about botsnacks. <NieDzejkob>sneek: tell sirgazil that a singular botsnack could be more problematic <gnutec>sirgazil: There is nothing in my /var/lib/gdm. <sirgazil>gnutec: But are you able to boot using a previous version of the system? <mbakke>oriansj: try running 'guix pull' first <oriansj>mbakke: they are from 8 hours ago; when I started guix pull <mbakke>oriansj: that's odd, what is the output of 'which guix' ? <NieDzejkob>a previous discussion on how oriansj doesn't want to use the generated profile file comes to mind... <mbakke>oriansj: you need to ensure that the guix from '~/.config/guix/current/bin' comes first in PATH <mbakke>also, you probably want to uninstall guix from your profile, there is no good reason to have it there <mbakke>oriansj: if you run '~/.config/guix/current/bin/guix package -u', i'm sure you'll get very different results <gnutec>nckx: sirgazil: Just removed /var/lib/gdm/, /home/[USER]/.cache and /home/[USER]/.local/share. The system will create news directorys when I reboot?? <oriansj>NieDzejkob: you are correct, I am quite ancient in regards to my environment <NieDzejkob>oriansj: well you should update that every time you pull <oriansj>NieDzejkob: especially since I am responsible for guix's future root of trust