IRC channel logs


back to list of logs

<zero21>guix system: error: symlink: Permission denied: "/var/guix/profiles/"
<atw>I emailed help-guix a while ago, but haven't seen it show up in the archive just yet. I don't suppose anyone's gotten it?
<zero21>ops dont worry looks like it has to be root only after all
<PotentialUser-69>If GuixSD uses Sheperd as the init system why does is systemd in /etc/fstab?
<lfam>PotentialUser-69: I don't see that in my /etc/fstab, but I'd guess it's a side effect of using elogind in some GuixSD configurations. elogind is a fork of systemd's logind that helps us use GNOME without systemd
<PotentialUser-69>yeah, IDK...I just ran "locate systemd" and my terminal spit out screens full of systemd directories.
<PotentialUser-69>fstab line is "none /run/systemd tmpfs mode=0755"
<lfam>Right, I'm sure it's related to elogind
<buenouanq>ok so,
<buenouanq>there is something wrong with services
<buenouanq>possibly rooted in dbus-system
<buenouanq>and I would like to know what I can do to give a proper analysis of it for you real devs
<buenouanq>last night it was suggested doing a fresh install (system init) from the supplied and verified image without doing any sort of pull or update
<lfam>Basically, a bug report should say what you expected to happen, what happened instead, and any details that could help reproduce the system. So if it's reproducible from a fresh install, or especially with `guix system vm-image`, that helps a lot
<lfam>PotentialUser-69: If you're interested, this is the code that makes that directory:
<Apteryx>what is the "module-import.drv" derivation? it takes a while to make each time I test the build of a package I'm working on.
<marusich>Apteryx, that is a derivation that exists purely to copy Guile modules into the store for further use.
<marusich>For example, any G-Expression you write that imports modules will probably create such a derivation, since the code you write needs to import those modules when it runs.
<marusich>For details, check guix/gexp.scm
<Apteryx>is this mechanism new? It's the first time I notice it.
<clacke[m]>Apteryx: I've seen it since I started using guix 2 years ago :-)
<Apteryx>ah, OK!
<Apteryx>thanks :)
<Apteryx>sneek: later tell civodul I like your new man-db profile hook: 6611 entries processed in 12.2 s :)
<sneek>Will do.
<buenouanq>what can I put in the bootloader field if I know I'm going to be initing it with --no-bootloader ?
<str1ngs>buenouanq: can you just leave that field out?
<buenouanq>No, it's a required one - I need it either not to be, or like a dummy bootloader that doesn't do anything. The latter option would be nice because then you wouldn't have to remember --no-bootloader.
<str1ngs>I think --no-bootloader is the best way. since the target system could be a real system or VM. so --no-bootloader act's as a switch as it's intended
<buenouanq>right, I don't mean that that flag should be gotten rid of
<buenouanq>just a different way to call it
<buenouanq>define (bootloader no-bootloader) and put that in the config
<brendyn>Has anyone got network connection sharing working with networkmanager on GuixSD?
<buenouanq>what is that?
<brendyn>For example my laptop has wifi, so id like to share it's connection via ethernet to a desktop. Network Manager has this feature but for me it's not working by default.
<buenouanq>there seem to be weird things happening with some services right now
<buenouanq>I'm about to go through an install and see if I can figure anything out.
<buenouanq>my ssh service won't start on it"s own is what i'm dealing with
<brendyn>hmm ok. im wondering if i need some dependencies like dnsmasq
<buenouanq>GPT PMBR size mismatch (2319631 != 3891199) will be corrected by w(rite).
<buenouanq>I get this on the flashdrive after dding the iso.
<buenouanq>and it causes refind to freeze/crash when I plug it in
<buenouanq>this is the iso.xz just now downloaded from the wobsite
<buenouanq>so I guess that puts a hault to my desire to try and figure that out
<str1ngs>buenouanq: is this a desktop? in regards to ssh?
<brendyn>buenouanq: You decompressed it right?
<buenouanq>brendyn: yes
<buenouanq>str1ngs: both my friend"s AMD desktop and a 2008 Macbook (thus rEFInd)
<buenouanq>because of the above error on the laptop I had to start with the 0.13 image and pull before installing
<str1ngs>buenouanq: I meant your using deskop-service? in regards to ssh issue
<str1ngs>err %desktop-services
<buenouanq>oh, yes
<str1ngs>I'm not sure if this resolves this but I moved ssh service higher up in my service list
<str1ngs>seemed to help. but it could just be an intermediate bug
<str1ngs>there is some service issue though with ssh and %desktop-services. dunno if it's related to network manager or shepherd
<buenouanq>I never would have imagine that the order they come in the os config would matter...
<buenouanq>is that a thing?
<str1ngs>I suspect ssh service times out waiting for network manager bring network up. that's just a guess, I have not had time to debug it.
<buenouanq>sounds plausable
<str1ngs>buenouanq: It probably doesn't matter no
<str1ngs>I mean in terms of service order. it just seemed to help
<str1ngs>buenouanq: either way I have ssh service just after (gnome-desktop-services) in my services section
<marusich>buenouanq, I have this issue for SSH also.
<marusich>I wonder if it's a bug...
<buenouanq>good to know it's not just me
<marusich>buenouanq, I am trying to reproduce the issue in a vm...
<marusich>I looked into this on my own computer a long time ago, but I gave up because there were no obvious error messages.
<buenouanq>marusich: let me know how it goes
<g_bor>Hello guix!
<g_bor>I remember having conversation about building deterministic python bytecode on python3.
<g_bor>I've just bumped into the issue again building python-minimal.
<g_bor>I don't feel I'm the python guy here, but it seems, that nix has done this
<g_bor>Could we use that?
<g_bor>I remember we have a fix for 2.7.
<catonano>I received some highly valuable feedback. I appreciate that
<ng0>hi. how do you feel about a patch that adds fish-guix into our etc directory in the guix repo? I don't feel like maintaining it anymore, and I don't see myself fixing the general fish integration (installing extensions via guix) any time soon. It's on my list, but I think in the meantime people would benefit more from something they can simply copy from the directory.
<ng0>I'd relicense it to public domain, so that it can be integrated (unless BSD-3 + GPL3 are compatible, didn't check in a long time)
<ng0>there's lots of todo's in fish-guix, including a rewrite, but I think it's better this way
<fps>there seems to be a bug in the handling of bug-guix. reporting a single bug by sending a mail to the ml causes a subscription where you get mails about all bugs
<fps>i don't want to send another bug report. i just unsubscribed ;)
<catonano>where is makeinfo in guix ?
<nckx>catonano: texinfo.
<catonano>nckx: I am trying to build g-golf and texinfo is in my ad-hhoc environment. Yet the g-golf configure can't find makeinfo
<catonano>"checking for makeinfo... no"
<catonano>nckx: thanks, anyway
<nckx>catonano: then g-golf is doing something ‘clever’ (read: stupid) and not relying on $PATH, or there's something wrong in your environment.
<nckx>You can always try running ‘makeinfo’ from a ‘guix environment’ shell to rule out the latter.
<catonano>nckx: that's exactly whhat i did. Makeinfo gets found indeed
<catonano>so probably I messed up the ad-hoc environment
<catonano>thhe ad-hoc envirrronment I used for g-golf
<catonano>is thhis line wrong ?
<catonano>guix environment automake autoconf pkg-config texinfo guile gobject-introspection glib
<nckx>Unless you forgot --ad-hoc: probably. That's the environment *for* those packages, not *with* them.
<nckx>Add --ad-hoc before the package names.
<nckx>Or it's not an ‘ad hoc environment’ :-)
<catonano>nckx: ok, now it doesn't find something else ;-)
<catonano>nckx: thanks !
<nckx>catonano: you sound tired/frustrated today. I hope it's not (too much) Guix's fault.
<nckx>catonano: yay!
<catonano>nckx: sorry to sound frustrated. If Guix has a role in my frustration, that'll be marginal ;-)
***pksadiq is now known as Guest24336
<castilma>for some reason, guix system build trys to compile linux-libre 4.14.13, even though i use and what may be the problem?
<castilma>it builds m3nrwq5b18392j3j4g56pchmjy48z4qx-linux-libre-4.14.13.
<pkill9>maybe you're using the git version of the package recipes and the build server hasn't built the newest ones?
<pkill9>probably not, but just a suggestion
<catonano>where is ldd ?
<nckx>catonano: I think glibc:bin (not checked).
<clacke[m]>not binutils?
<amz3>catonano: try that kind of search to figure out it yourself
<amz3>in debian, it's libc-bin
<nckx>clacke[m]: that's ld.
<nckx>If you're feeling lucky, patient, or both, there's always ‘ls /gnu/store/*/*bin’ which I use more often than I care to admit. It only works if it's something that's already [likely to be] installed, of course.
<clacke[m]>I know the difference, and I still thought it was in binutils. But if debian has it in libc, it should be there in guix too. :-)
<clacke[m]>fun tip: 'printf "%s\\n" /gnu/store/*/*bin/blah' is faster than ls if you have many hits, because ls stats every hit. :-)
<nckx>clacke[m]: All right. My search engine ‘helpfully’ autocorrected ldd to ld until I noticed, just double-checking.
<clacke[m]>fun tip: `'printf "%s\\n" /gnu/store/*bin/blah'` is faster than ls if you have many hits, because ls stats every hit. :-)
<nckx>I'm going to... try to remember that. Thanks.
<nckx>ACTION feels another one-line script coming on.
<clacke[m]>argh stop helping me, matrix client. thats percent s backslash n ...
<amz3>why are you using matrix clacke[m]
<nckx>clacke[m]: both show up correctly (well, identically) over IRC at least...
<clacke[m]>because I can't be bothered to set up a bouncer
<clacke[m]>and also there are native matrix channels I frequent
<amz3>oh ok
<clacke[m]>full oggcast episode on how and why I do it here: :-)
<nckx>Ding! That's why your nick sounded familiar.
<clacke[m]>nckx: interesting. so it's the reading side of the client that is being helpful then, apparently
<clacke[m]>a listener, eh
<clacke[m]>I also did an hour-long show with cwebber on the awesomeness of guix
<clacke[m]>and a quick followup on guix everyday ops
<nckx>I listened to an episode you made about either Guix or Nix (sorry, terrible memory).
<nckx>Yep, that's the one.
<nckx>I subscribed to the HPR feed last week, after thinking ‘I should subscribe to the HPR feed this week’, but it's a bit... much. I'll have to filter it somehow to get rid of the gaming on Windows stuff etc.
<nckx>And cool toolz for my Mac.
<nckx>Anyway, off-topic. Something something /gnu/store.
<clacke[m]>if that's why you ended up in this channel, I'm immensely happy about that
<str1ngs>nckx: nix kinda supports macos now
<clacke[m]>yeah, I listen to <10% of the shows!
<clacke[m]>nix works very well on MacOS and has done so for years
<clacke[m]>I've been using it for the last two years
<str1ngs>ah I thought it was a recent thing
<str1ngs>does guix support macos?
<nckx>clacke[m]: Other way round, actually. I found out about HPR on the Guix ML.
<nckx>clacke[m]: (sorry, I was being snarky. I've never owned a Mac.)
<clacke[m]>aha, glad to hear that too :-)
<clacke[m]>I looked at the commit history a while ago and it was over a decade ago that OSX was added.
<nckx>str1ngs: Guix doesn't, for reasons which were once discussed on the ML.
<str1ngs>nckx: dues to fsf reasons? or technical?
<str1ngs>mind you I could just search ML
<nckx>Mainly, IIRC, the fact that's it's very hard to do purely functionally because of all the MacOS stuff that has to be passed through.
<clacke[m]>nix supposedly runs on cygwin and freebsd too, but I haven't tried those personally
<clacke[m]>guix is a GNU affair, with GNU or Linux kernel
<clacke[m]>Yeah nix cheats to make osx work
<clacke[m]>on linux you can bootstrap nicely
<nckx>But yes, the philosophical and possible legal issues means there won't be many people as enthousastically trying to make it work as in, for example, the Nix community.
<clacke[m]>on mac it's here, take this huge blob of xcode binaries
<str1ngs>hmm w/e screw mac
<str1ngs>does anyone use offloading, seem non trivial to setup
<nckx>Then again, there are (good) people working on GNU/Chromium, so never say never.
<str1ngs>for example it will only build using root, and I get archive errors
<str1ngs>guix offload: error: corrupt input while restoring archive from #<input: channel (open) 1d73260>
<str1ngs>is the latest hurdle
<nckx>There seems to be a flurry of offloading-related ML posts lately. Don't know if it was recently broken, people are only now trying to use it at home, or both.
<ng0>what is gnu/chromium?
<nckx>(Sorry, can't help you. Can only say useless things like ‘omg Hurdle sounds like the cutest kernel ever’.)
<str1ngs>guix steeple chase!
<ng0>str1ngs: you need to update to a more recent commit, reconfigure and reboot the machines
<nckx>ng0: Meant to type Chromium on Guix while reading another window. Not a statement on Chrom*'s... ambitions, although it could have been.
<str1ngs>hmm I'll try that, but I did a reconfigure on both machines just yesterday
<nckx>And yes, I was referring to your and Marius' work :-)
<ng0>str1ngs: you need to pull
<str1ngs>I'll try with the latest pull though see if that helps. think my pull was a couple of days old
<str1ngs>I assume both machines should be in sync with pulls and reconfigures?
<str1ngs>ok, will try that out
<str1ngs>also I have to eventually figure out why I can only use root as for user.
<clacke[m]>offloading is pretty new
<pkill9>nckx: is there aproject page for getting chromium on Guix?
<pkill9>a project*
<ng0>not that new though
<str1ngs>I'm hoping it makes managing multiple machines easier. right now syncing pulls and guix version and using substitutes is tedioius
<pkill9>i assume it's just chromium with closed source stuff stripped out
<ng0>bugtracker, pkill9
<str1ngs>also invalidating substitute cache is PITA. unless I'm missing how to do it via guix
<clacke[m]>guix --listen is mentioned in the 0.14.0 release notes. someone not on the mailing list will not have noticed it until then.
<ng0>pkill9: if you want the preview, I have a public branch with it
<clacke[m]>so then several people get interested in trying it out, no surprise
<str1ngs>I need to work more on my guile/gtk web brower. been pretty busy
<str1ngs>I'm calling it nomad.
<str1ngs>I hope to add gnunet support aswell.
<catonano>str1ngs: wonderful idea
<str1ngs>I have a working pre alpha alpha. so far, I have a basic browser with a embedded geiser REPL via vte
<str1ngs> screenshot
<catonano>yes, I already saw the screenshot
<str1ngs>ah my bad
<str1ngs>it also has a mode line like thingy. that does primitive evaluation
<catonano>str1ngs: no, it's ok :-)
<catonano>whhat is a mode line thingy that does primitive evaluation ?
<str1ngs>instead of having a full blown REPL. there is text input field you can eval expressions with
<str1ngs>so say you want to go back. you can do (back)
<catonano>a field ? You mean a line like a terminal ? Or a field like the fields in the web forms ?
<str1ngs>also you can define your own functions. pretty much everything will be extensible via scheme
<catonano>wondeful really
<str1ngs>like emacs M-x input field
<catonano>mmm ok
<str1ngs>except I'm not using command functions. just expressions for now
<catonano>sonds like a great project.
<str1ngs>emacs you would do M-x back . in nomad right now it's M-x (back)
<ng0>sound interesting
<jlicht>hi #guix!
<g_bor>This python bytecode reproducibility is still bugging me...
<g_bor>Looking around I found this:
<g_bor>It seems to me, that this is very similar to our python2 fix.
<az->hi there
<az->what to do with messages
<az->warning: arbitrarily choosing /gnu/store/xfrlpdnzkbyrbqss0gqfi6fsmx84yq9c-pinentry-gnome3-1.1.0/bin/pinentry
<az->warning: collision encountered: /gnu/store/ddjiydxwqmcdq95gi8qbj8ba4c5bbz7x-jack-0.125.0/bin/jack_metro /gnu/store/yx9r3p06j9v1x7g2wmr71awy5q5x13hm-jack2-1.9.12/bin/jack_metro
<lfam>az-: In most cases, they are harmless, but we have to evaluate them on a case-by-case basis
<az->same with dconf ipiphany xdg... and many other packages
<az->what mean case-by-case? how to fix this?
<lfam>az-: The issue arises when you install into a profile, with `guix package --install`, two directories in /gnu/store that contain the same path.
<lfam>I mean they might be harmless, might not be, and we need more information to decide
<lfam>In your second example, there are two JACK packages in the profile, and they both have 'bin/jack_metro'. Only one of those 'bin/jack_metro' files will end up in your profile
<lfam>Did you mean to install both versions of JACK? Or perhaps one of them instead propagated by another package you did install?
<lfam>I'm also curious about the pinentry collision. It would be helpful to get all the warnings about these collisions
<lfam>However, I'll have to offline in a few minutes. Maybe somebody else can help you here, otherwise email <>
<az->lfam: thank you very much! i'll try to find that collision
<az->lfam: ok
<jlicht>would it be okay if I rebased my and Jan's work on the npm importer + node build system and push them to a branch on guix' git repo (e.g. "wip-node-build-system")? I would like to get it into a mergable state so people can at least properly hack it. I would also love to co-mentor a potential GSoC student if they choose to continue work on this as well :-)
<ng0>okay, I have sent patches to remove fish-guix and add the completions to Guix. Could anyone using Fish look into the patch set in the next days if it works for them?
<ng0>to my best knowledge they should just work
<lfam>jlicht: Yeah, I think it's fine. Generally, we use the "wip-" prefix to indicate that the branch's history may be rewritten (rebased, etc). Just something to keep in mind when naming it
<jlicht>lfam: exactly what I was going for, so that will be a perfect fit
<lfam>Yeah, it's necessary for this sort of long-term project
<jlicht>ng0: What can I do to test your patches?
<ng0>hm.. probably test if the addition to does what it should do
<ng0>patches should come in in the next couple of minutes
<lfam>Be sure to reply to civodul's GSoC email, jlicht :)
<jlicht>lfam: I will somewhere today :-)
<ng0>jlicht: I added --with-fish-completion-dir
<ng0>and merged fish-guix as into etc/completion/fish/
<ng0>I'll push a relicensing commit to the original repo, an informal statement is already attached to the patches
<jlicht>ng0: I am not seeing anything weird, so if the licensing thing is not an issue I can merge them later today once I have re-built guix with your changes to see if fish picks everything up
<jlicht>I had to laugh a bit seeing you update the copyright notice for /removing/ some code in `shells.scm'. When is it appropriate to update the copyright notice anyway?
<ng0>jlicht: well you could also pull from my repository
<ng0>branch 'etc-fish'
<jlicht>ng0: If that includes your two patches, I already have them locally.
<jlicht>just a matter of rebuilding on my (t)rusty t400
<ng0>fish has issues in Guix anyway, I don't expect it to pick this up if it's in /gnu/store/… but maybe this works better
<mlin>Anyone have any idea how to get `xrdb` to run at startup? I can't figure it out.
<jlicht>ng0: yeah, I found out that profile-related things for setting all manner of paths do not work yet /w fish.
<guixbuilder>Hi there! I wonder how to install libbamf on GuixSD?
<guixbuilder>Do I have to compile it manually?
<jlicht>guixbuilder: afaik we do not have it packaged yet
<atw>guixbuilder: I searched for it, but it does not appear to be packaged. You could try packaging it yourself:
<ng0>jlicht: well I made some changes.. idk.
<ng0>jlicht: I have a couple of bookmarks if you are interested
<ng0>nix solved them, but I have other priorities
<jlicht>ng0: Understandable. If you could share them, that'd be nice
<ng0>iirc they are in the fish related bug on the bugs bug tracker
<ng0>but it could be that I need to sync them up
<guixbuilder>jlicht atw: Too bad… I don't feel enough good in guile to package it myself.
<ng0>long 7 days D
<ng0>I'll send in links in the next hours
<jlicht>guixbuilder: You can always try your hand by looking at relatively simple packages like `hello', and if you get stuck, ask for help here on #guix or on the help-guix mailing list :-)
<jlicht>ng0: I have some connectivity issues, so if I am offline by then, just leave a message with sneek.
<guixbuilder>jlicht: ok, I will try.
<jlicht>guixbuilder: do you have a link to the sources of libbamf?
<ng0>jlicht: I meant I'll send them to the bug
<jlicht>ng0: also fine
<atw>looks like ?
<guixbuilder>jlicht: Yes:
<guixbuilder>jlicht: Will this work: (uri (string-append "" version "/+download/bamf-" version ".tar.gz")))) ?
<jlicht>guixbuilder: it might, but in that specific snippet you have two ")" too many at the end. If you want to share code snippets, you can use a service like and just share the whole package recipe as-is.
<nckx>guixbuilder: Even better is to replace that ‘0.5’ using (I think) version-major+minor. Examples abound in the source tree.
<marusich>buenouanq, regarding the problem of ssh-daemon not starting, do you use static-networking-service-type in your operating system declaration?
<ng0>jlicht: updated the bug
<mlin>Hi all, anyone know how to downgrade a package? Or install a specific version of it?
<atw>mlin: something like binutils@2.22 I believe
<atw> see the -i section
<marusich>mlin, to donwgrade a package, you can do any of the following: (1) use 'guix package --roll-back' or 'guix package --switch-generation' to revert to a prior generation of your profile that uses the desired version of the package, or (2) change the package definition so that it builds the older version, and then install it with "guix package -i"
<mlin>Of course, I forgot
<mlin>I might bump my question from before to you. How do I get `xrdb` to run at X startup from the graphical login screen?
<ng0>guix package -i xrdb
<mlin>ng0: Thing is, that when I run `xrdb .Xresources`, it works. I just want it to happen from X startup
<atw>mlin: maybe you need to put the command you want in your .xinitrc? then it will run after graphical login
<amz3>doesn anyone know why there is no package for guile-clutter (cc rekado)
<guixbuilder>nckx: Thank for the tip
<pkill9>how does guix handle packages that add kernel modules?
<guixbuilder>I began this script, is it a good start?
<mlin>Hmm. I made `.xinitrc` and put `xrdb ~/.Xresources` in it, but it doesn't seem to be working
<pkill9>mlin: try `xrdb -merge ~/.Xresources`
<jlicht>guixbuilder: afaik bamf needs *some* version of libwnck, so you might want to import `(gnu packages gnome)' at the top of your modules and it it to inputs.
<jlicht>and inputs (among others) is indeed the place to put dependencies :-)
<jlicht>guixbuilder: see for a short explanation on what to use when
<atw>hmm. sorry about that. maybe graphically logging in via your login manager (slim/sddm/gdm) isn't running your xinitrc? You could test this by putting something like "touch /tmp/test" in the xinitrc, logging in, and seeing if the file is created
<atw>mlin: ^
<mlin>atw: I'll have a go
<mlin>atw: No luck. The file doesn't exist
<jlicht>mlin: not sure if this is still relevant, but maybe your issue is similar to what was discussed in
<jlicht>mlin: especially 宋文武's message might offer a workable solution for now :-)
<mlin>jlicht: Looks promising, where do I put `.xsession` though
<jlicht>mlin: my uneducated guess would be in $HOME :-)
<mlin>jlicht: Ah ok, I'll try
<mlin>Hmm. So it is now trying to execute `.xsession`, but it is now saying `permission denied` for the file. Perhaps $HOME is not the place for it?
<mlin>I also cannot log in to i3, I am root
<mbakke>mlin: Is it executable?
<atw>mlin: also make sure to include the exec "$1" at the bottom
<mlin>mbakke: Ah, forgot that
<mlin>Wicked! Thanks pals, it works now
<pkill9>mbakke: would running this custom version of Guix still allow you to use the officail Guix build substitutes?
<jlicht>pkill9: guix-nonfree only includes some (nonfree!) extra packages for guix. So all packages from the normal guix repo should be unaffected an therefore still available from the 'normal' guix build servers
<guixbuilder>Using this recipe
<guixbuilder>I get this error: guix build: error: #<unspecified>: not something we can build
<atw>excellent mlin ☺
<pkill9>jlicht: i should've been more specific, the readme in that repo explains how to install a custom GuixSD, which I think uses the nonfree Linux kernel, i was wondering if official Guix build substitutes could be used with that custom build of Guix
<ng0>yes. though this is not official guix material and will not receive support inside Guix.
<pkill9>no problem
<mbakke>pkill9: You can have multiple third-party repositories in $GUIX_PACKAGE_PATH, but Hydra obviously won't have substitutes for them.
<jlicht>pkill9: it should still work, as it doesn't touch the dependency graph of the freedom respecting software
<buenouanq>marusich: no, I'm using %desktop-services
<buenouanq>and gnome
<jlicht>guixbuilder: I'm not sure if the "gnome" dependency is correct.
<pkill9>the question for me is how much you can replace dependencies provided in the official repo with custom ones and still use substitutes built using dependencies from the official repo's recipes
<guixbuilder>jlicht: Should I list every dependency contained in ?
<pkill9>(which is more a question generally related to Guix, doesn't neccessarily have to be nonfree patches)
<ng0>pkill9: come in and find out
<jlicht>guixbuilder: I do not get the error you are talking about when using your package recipe btw. My build complains about pkg-config being missing, so you could try to add pkg-config to the `native-inputs' of your package.
<ng0>pkill9: for some modifications it will be good to have an offloading machine, but in general it just works
<jlicht>guixbuilder: and finding out which depencies are required is usually a combination of reading the documentation of the package you are trying to build, some trial and error and indeed sometimes a hamfisted approach of just including everything referenced in the build scripts of the package.
<ng0>you only get substitutes from the "official" buildfarm for substitutes it builds
<ng0>the moment you change something you will have local builds
<ng0>for example my st-full package is build, not (yet) from substitutes
<ng0>and if I will ever get substitutes for it, it will be from my own substitutes server
<jlicht>pkill9: to add to ng0's comments: if the output of is changed to include your custom packages, you will most likely need to build locally
<pkill9>ok thanks for answering, i'll test in a VM
<ng0>pkill9: that repo just contains chromium. you could take the chromium from my pretest/chromium branch and modify it the way you want it to re-apply on a local guix repo branch
<guixbuilder>Anyone knows what this means: "ice-9/boot-9.scm:2795:6: no code for module (gnu packages libwnck)" ?
<ng0>pkill9: oops. my repo ->
<jlicht>guixbuilder: It means that there is no module for (gnu packages libwnck) ;-). If you were to issue `guix package --show=libwnck', it shows among other things the location of the package
<jlicht>in libwnck's case, the location is (gnu packages gnome), so if you have that module imported, you can just refer to the `libwnck' library by name.
<jlicht>in case it is not clear: guix modules are guile modules, which can each define and export many (or one, or none) gnu guix packages. There is usually some sort of 'theme' to each module, but this is more of a convention than a technological choice.
<guixbuilder>Does this version give you the same error as me (not something we can build)?
<jlicht>guixbuilder: could you share (via paste) your exact error message?
<jlicht>I am getting an unbound variable error for "dbusmenu"
<guixbuilder>I have nothing more than "guix build: error: #<unspecified>: not something we can build"
<guixbuilder>Even when increasing verbosity
<jlicht>guixbuilder: hmm, I have no experience with that error. You could try to email the help-guix mailing list
<guixbuilder>jlicht: Didn't I forget a field in the recipe?
<jlicht>guixbuilder: not that I can see, no. You should take care to import the right modules for each of the mentioned packages though. For exapmle, the desktop-file-utils package is located in the (gnu packages freedesktop modules), which you did not yet import.
<jlicht>I have to leave now, but I can definitely recommend to have a look at the guix manual, as well as some of Pjotr Prins' notes:
<jlicht>good luck guixbuilder!
<mlin>jlicht: I appreciate you sending those notes, they're useful
<mlin>Has anyone packaged Firefox 57 at all? I know this is off topic for a libre distro, but if anyone could send over a recipe I'd be glad.
<mlin>amz3: indeed
<amz3>mlin: that said, there is bug about chromium in debbugs
<mlin>amz3: wow. I don't use Chromium - due to it being mostly non-free, but I would like to use Firefox 57 since I don't mind it much
<ng0>mlin: that reminded me, I do have firefox but I only started it as a base for torbrowser, which referencing the message exchanges I had with torproject can be shipped as torbrowser when I get there. modern browsers are difficult, painful. but I like pain and the challenge. it's just not moving very fast because I need time, and I've settled on Thunderbird first (to patch it to a state where it can be included,
<ng0>if at all, recent developments etc etc, long story)
<ng0>I do not have a working firefox, which is why I can talk about this ;)
<mlin>ng0: so you have a working recipe for 57?
<ng0>I guess you'll see a functional chromium first
<ng0>read the very long sentence again
<mlin>Oh missed it sorry
<ng0>and if I had, I would not mention it here
<mlin>okay, thanks for the info
<pkill9>oh, does Guix distinguish between build-time dependencies and run-time dependencies?
<castilma>native-input is built time
<pkill9>so you could change the run-time dependency of a recipe and the built package would be exactly the same as before (same hash, and subsequently able to use the build-servers substitute)?
<nckx>pkill9: Kind of.
<nckx>pkill9: ^ to your previous question.
<nckx>ACTION slow.
<pkill9>oh ok
<nckx>Now I'm thinking: no, since that last sentence makes no sense to me.
<pkill9>which one?
<nckx>Dependencies are simply references: if package A points to (a specific build of) package B, B is considered a dependency of A.
<nckx>pkill9: ‘change the run-time dependency of a recipe’; what do you mean by this?
<pkill9>like a package required in order to build the package
<nckx>Anyway, you can't change A to point to package C or a different build of B without changing the hash of A.
<pkill9>i'm probably applying traditional concept of dependencies where it doesn't apply
<nckx>I think grafts come closest to what you have in mind.
<nckx>They avoid rebuilding packages, but still change the hashes.
<nckx>It's very mindfucky stuff when you're coming straight from the Old World and I'm doing a horrible job at explaining it, sorry.
<nckx>ACTION is studying for an exam tomorrow and though ‘oh good a simple question I can answer to procrastrinate’. Bad choice.
<pkill9>yeah it looks like grafts are what i'm interested in
<civodul>heheh :-)
<sneek>Welcome back civodul, you have 1 message.
<sneek>civodul, Apteryx says: I like your new man-db profile hook: 6611 entries processed in 12.2 s :)
<pkill9>hehe well i understand a little better so ty for the answers
<civodul>Apteryx: heh, glad you like it!
<civodul>it does feel faster
<Apteryx>it at least 3 times faster I think.
<nckx>pkill9: If you've already read the manual and still don't understand, that means the manual needs improvement. Please don't hesitate to suggest improvements. New perspectives are invaluable and hard to fake.
<Apteryx>nckx: I know the feeling. You can write patches too, even better.
<nckx>Patches are machine-readable suggestions :-)
<nckx>I've noticed that ‘patches welcome’ is too often use to dismiss any other kind, so I tend to avoid it, but I certainly agree that's ideal.
<nckx>^ in other projects; not Guix.
<Apteryx>hehe, true. I meant it as a mean of getting entertained amidst your exam studies ;)
<nckx>Apteryx: Be very careful what you wish for now.
<civodul>LO fails to build on core-updates, i wonder if we should update to 6.0.0
<Apteryx>nckx: ahah!
<Apteryx>nckx: by the way, most of your emails are flagged by gmail as spam :/.
<Apteryx>Big G's interface unhelpfully says: This message has a from address in but has failed's required tests for authentication.
<nckx>Apteryx: Thanks. So yeeeah... I'm aware of that. Short answer: DMARC. Long answer: it's a deliberate decision to disallow forwarding the way (amongst others) does it. I might revisit that decision; I've mellowed with age.
<nckx>The only reason I haven't is that it disproportionately affects Gmail, which I despise, which is nice.
<nckx>OK maybe ‘mellowed’ is not the right word.
<Apteryx>I see. I once started reading about how to manage my own SMTP server... and then found out about all the modern anti-spam technologies and their requirements... I think reverse DNS was one... which didn't seem like it'd do well with my host being resolved through dyndns.
<Apteryx>if you have a GuixSD config somewhere setting a SMTP server up I'd be interested in taking a peek.
<nckx>Thinking about it some more: the only way to stop that would be publishing a policy of ‘forging e-mails from my domain is A-OK’, which I'm not willing to do... :-/
<nckx>Apteryx: my entire mail set-up is GuixSD. Ask me again in February.
<Apteryx>nckx: Cool :)
<Apteryx>I will.
<Apteryx>good luck with the exams!
<nckx>^ well, almost. There's a hilarious bit where e-mails get sent from Paris to a FreeBSD box in Miami to be signed and straight back to Paris before they're sent because I'm too lazy^Wbusy to finish packaging dkimproxy. But otherwise it's not that horrible. Thanks!
<nckx>ACTION afk zzz