IRC channel logs


back to list of logs

<catern>is there a convenient way to enumerate accessible profiles?
<anthk_> gnu/store/w4bdz6y8qs54p4098qgv0ndmfhf0n2cz-xz: (stdin): Compressed data is corrupt
<catern>oh, there's guix-profile-prompt
***kelsoo1 is now known as kelsoo
<ng0>hm... dmidecode with guixsd as user root is giving away very limited information.. specifically bothering "/dev/mem: Operation not permitted."
***Digitteknohippie is now known as Digit
<ng0>yep. seems like dmidecode and flashrom are pretty limited with guixsd
<ng0>looks like I need to boot into some livedisk to update libreboot
<anthk_>gnu/store/w4bdz6y8qs54p4098qgv0ndmfhf0n2cz-xz: (stdin): Compressed data is corrupt
<anthk_>I was doing " guix package --fallback -i lxappearance"
<anthk_>that's from stderr
<efraim>reading nix code reminds me I need to get better with sed
<rekado>I just tried to use the latest Emacs 25 pre-release but found that it won’t find any of the Emacs packages I installed via Guix.
<rekado>is there an easy way to make this work?
<rekado>I think it’s related to the autoloads file which is only generated for 24.5.
<MaliRemorker>i tried to use the advice from to change the resolution, but running guix reconfigure produces the following error:
<MaliRemorker>gnu/services.scm:554:50: In procedure add-edges:
<MaliRemorker>gnu/services.scm:554:50: In procedure struct_vtable: Wrong type argument in position 1 (expecting struct): #<procedure 5526400 at /etc/config2.scm:15:2 (state)>
<alezost>rekado: ouch! how did you try it? do you have a guix package for this emacs?
<MaliRemorker>i have placed (my-slim-service) into 'services' field of 'operating-system ' in the config file: is this the problem?
<MaliRemorker>ok, maybe i ask this on the mailing list
<efraim>phase `check' succeeded after 9126.5 seconds
<efraim>I might need more swap, I was about to close firefox to make sure the tests didn't fail to lack of ram
<efraim>the P4 box I have in the corner, if I get that hooked up again I'm now more seriously thinking of making the original 20GB IDE drive all swap
<alezost>MaliRemorker: this is an outdated example, it don't work anymore because the way how services can be defined changed. Wait a minute, I'll paste something
<alezost>MaliRemorker: try this:
<MaliRemorker>alezost: thanks
<alezost>MaliRemorker: however if X server can't find a proper resolution by default, it's likely that you cannot fix it this way. Do you use radeon? I heard it requires some non-free stuff which does not supported by linux-libre kernel
<alezost>*is not supported
<MaliRemorker>alezost: no, i am running guixsd in a virtual machine, the driver 'qxl' supports many resolutions
<MaliRemorker>alezost: i think i need to change the resolution explicitly on guest in order to increase it
<alezost>ah, ok
<MaliRemorker>alezost: guix system: error: service 'xorg-server' provided more than once
<efraim>julia fix pushed, now to work on c-reduce
<MaliRemorker>i guess i should run it via modify-service ?
<alezost>MaliRemorker: that's because you probably use %desktop-services
<alezost>you need to remove the default slim-service from %desktop-services
<civodul>Hello Guix!
<civodul>ACTION attempts a merge of master in core-updates
<civodul>seems bioinformatics.scm was modified concurrently in both branches
<efraim>i fixed building julia and pushed the wrong change, make sure you have the revert and re-upload if you grabbed somewhere in the middle
<efraim>sounds like fun
<efraim>updates? fixing build problems?
<efraim>i thought they were mostly leaf packages
<civodul>anyone knows why lfam reverted a few python patches in core-updates a week ago or so?
<efraim>it broke matplotlib iirc
<efraim>he moved them to python-updates
<efraim>matplotlib-1.5.2 needs some new python libraries before it can be built
<civodul>ah ok, makes sense
<civodul>bah, i committed stuff unwillingly in b1bf155ffd9b160afdf05aff39d3f0b0f6d11589
<rekado>civodul: huh, bioinformatics.scm was modified in core-updates?
<efraim>sneek: later tell ng0 I found nix's qt module cmake path workaround, where they hardcode %out in each qt module's cmake file, which might help with qtdeclarative
<efraim>sneek: botsnack
<civodul>rekado: maybe not, but git showed conflicts in files where there shouldn't have been conflicts
<anthk_>guix pull fails on tar :|
<anthk_>'guix package -u' sorry
<efraim>can you post the error again? it looked like you had corrupt items in your store
<efraim>anthk_: ^
<efraim>civodul: any ideas with anthk_'s problem?
<efraim>I thought the patch was added back to hydra's store
<efraim>it also looks to me like anthk_'s bootstrap guile is corrupt in the store
<efraim>what filesystem is /gnu mounted on?
<efraim>It looks like you have substitutes enabled since it tries to download the file from hydra
<anthk_>efraim, ext4
<efraim>ok, we've seen problems if the store is mounted on btrfs but ext4 is the default
<anthk_>any fix?
<civodul>anthk_: are you running guix-daemon --disable-chroot?
<anthk_>I am at GuixSD
<civodul>is it on i686 or x86_64? in a VM?
<civodul>also, what command are you running?
<civodul>it seems absolutely no substitutes were found, which is weird
<anthk_>just guix pull
<anthk_>hi adfeno
<adfeno>Hi anthk_ :D
<civodul>anthk_: could you run "guix build hello -n" and paste the command + output?
<adfeno>What's up?
<anthk_>civodul, I ran it under an "script" instance, sorry for the control codes, and beware
<civodul>what do you mean by "script instance"?
<anthk_>script out.txt
<anthk_>sorry, the utility
<adfeno>I think he means a shell script?
<adfeno>Oh... I see.
<civodul>oh i see, i never use 'script'
<adfeno>"script" command
<adfeno>It seems to be used for typsetting terminal sessions.
<anthk_>it's useful
<anthk_>brb, lunch time
<civodul>anthk_: arf, could you try "guix build hello -n --no-grafts"?
<civodul>i think the problem is that you haven't run "guix pull" in a long time and binaries have vanished from the servers
<Akko_Teru>("dance like no-one is watching")
<adfeno>anthk_: Has your issue been solved?
<adfeno>I saw the paste you gave us. I don't really know what's happening because noone told me yet.
<anthk_>adfeno, I am trying
<anthk_>altough is not a hard issue, I still can use other libre systems, but guixsd is useful
<adfeno>I'm just a new Guix "user" (I still have to decide what to install with it), but perhaps I can help...
<adfeno>Let's see...
<adfeno>anthk_: I'm seeing this channel's log now.
<adfeno>I once solved this "networking issues" error with "guix pull", but since you already did it, then I'm out of ideas...
<adfeno>anthk_: Try doing "guix gc", and then retry "guix pull".
<anthk_>adfeno, first I tought on that, but guix gc I think it deletes useful stuff
<adfeno>Well... as far as I know, "guix gc" deletes the remains from build sources, with some exceptions (I think it doesn't do so if the generation or profile you're using makes use of the build results, but we must check the documentation to make sure).
<adfeno>(I'll check that now, actually).
<adfeno>anthk_: Actually, I was a little off in the descript of what "guix gc" deletes automatically...
<adfeno>Accoridng to the documentation: "The garbage collector has a set of known “roots”: any file under
<adfeno>‘/gnu/store’ reachable from a root is considered “live” and cannot be
<adfeno>deleted; any other file is considered “dead” and may be deleted. The
<adfeno>set of garbage collector roots includes default user profiles, and may
<adfeno>be augmented with ‘guix build --root’, for example (*note Invoking guix
<adfeno>I forgot that lines from most texinfo files are broken beyound some character limits.
<adfeno>Also, see this:
<adfeno>"Prior to running ‘guix gc --collect-garbage’ to make space, it is often useful to remove old generations from user profiles; that way, old package builds referenced by those generations can be reclaimed. This is achieved by running ‘guix package --delete-generations’ (*note Invoking guix package::)."
<adfeno>So the contrary seems to hold true.
<adfeno>(that is: it won't remove the packages referenced by some generation).
<civodul>ACTION hacked 'guix publish' to serve content-addressed files
<civodul>that may help in the future for problems of disappearing files like anthk_ reported
<adfeno>"content-addressed files"?
<civodul>you go to and it returns the content of hello-2.10.tar.gz
<civodul>roughly, you give a hash and get the content
<adfeno>I see...
<adfeno>So, in anthk_'s case, it's refusing to take substitutes because some file disappeared from the source?
<adfeno>s/from the/inside the/
<civodul>there are 2 problems: (1) substitutes disappeared, so it tries to build from source, and (2) some source files disappeared as wll
<adfeno>It's strange that he couldn't fix it by doing "guix pull", while I was able to do so.
<adfeno>(I once faced "networking iusses" message, but I just needed to do "guix pull" instead).
<civodul>gitlab, the "old guard", etc.:
<davexunit>civodul: I saw this on reddit last night and got really annoyed
<anthk_>civodul, how isn't an emacs user capable of using emacs+gnus+magit?
<davexunit>you definitely don't need to know how to use gnus or magit to contribute to emacs
<anthk_>I meant what's wrong with pure git and mail protocols?
<adfeno>"pure git"... Hm...
<adfeno>I prefer to use GNU Bazaar instead. :D
<anthk_>adfeno, well, git and a libre git server
<anthk_>you can use notabug
<adfeno>anthk_: Indeed. :D
<adfeno>Git is free/libre software, I'm not against it. :D
<adfeno>Personally, (for my projects only) I prefer GNU Bazaar instead of git because novices and non-developers like me can remember/learn less concepts using it than using git.
<adfeno>And requires less steps to do repository interaction.
<anthk_>adfeno, you don't have to use all the git features
<adfeno>anthk_: Yes, indeed, but there are some actions that require one or another.
<adfeno>And so people will eventually have to learn them.
<adfeno>I don't know if this post can be shared (legally speaking), but here is a *similar* (not equal) issue, comparing git with subversion:
<anthk_>I tried to learn svn
<anthk_>I coudn't
<anthk_>adfeno, well, the current trend is the featuritis on everything, sadly
<anthk_>adfeno too many options and redundant
<anthk_>sometimes I fire up emacs with just "-q" :p
<anthk_>and about javascript... damn, I have to install noscript and the CPU usage went to the minium
<anthk_>emacs -q -nw is awesome for a quick edit :)
<lfam>How do we want to commit the SourceForge URL updates? Each package individually?
<davexunit>civodul: wow, you've done it again! content-addressed mirror with 'guix publish'!
<davexunit>awesome feature
<civodul>lfam: i'd say many at once
<civodul>davexunit: heh, i just had that idea and couldn't resist ;-)
<civodul>guix publish is the hot thing these days :-)
<davexunit>apparently so!
<davexunit>I'm glad to see so many improvements being made to it
<civodul>i'd like to use publish + Cuirass on the new machine, that gives an incentive to improve it
<davexunit>I always enjoy reading the bit where we use the pattern matcher for request routing
<civodul>yes, it's very pleasant
<lfam>mark_weaver: I don't see any non-trivial differences between the SVN and GitHub-tarball muparser source code. The only code change is in 'example1.cpp', which seems irrelevant, although I don't actually know.
<davexunit>civodul: I'm interested in trying out Cuirass as well
<lfam>What should the new version be? 2.2.5-github?
<lfam>We should have a standard for when hashes change but versions do not
<lfam>Ideas, anyone? :)
<davexunit>I need to find a new javascript library for doing FRP-like things so I can take another stab at functional web UIs for guix.
<davexunit>Cuirass will need one.
<civodul>yeah, though i view the web UI for Cuirass as secondary
<lfam>publish + Cuirass: Yes, I'm hungry for dog food :)
<davexunit>civodul: yes, it is secondary
<civodul>lfam: the more or less standard way is to add a revision number, like "2.2.5-2"
<davexunit>filed under "nice to have"
<lfam>civodul: Okay, will do
<lfam>Hm, adding the package revision in (version) complicates (uri (string-append "https..." version ".tar.gz)) since
<adfeno>Perhaps using ":" instead of "-" for the revision separator?
<ng0>version "-" gv ".tar.gz" where gv is defined as the guix-revision
<sneek>ng0, you have 1 message.
<sneek>ng0, efraim says: I found nix's qt module cmake path workaround, where they hardcode %out in each qt module's cmake file, which might help with qtdeclarative
<ng0>and similar
<lfam>adfeno, gv: The issue is that the source URL doesn't include the package revision
<lfam>Or, it can't
<lfam>ng0 ^
<ng0>i know which is why i added similar. the source needs then be different from the version
<ng0>so we should have a way to define version and package version and splitt off the package version from the source
<ng0>efraim: oh? tell me more
<lfam>How about this attempt at separating the upstream version from our internal version:
<adfeno>ng0: So the whole file name would be "[version]-[guix revision].tar.gz"?
<adfeno>lfam: I guess this would work.
<adfeno>(the paste you gave).
<lfam>It's ugly, but so is re-releasing a same version that's actually not the same
<adfeno>I guess we should wait for more opinions on this matter.
<lfam>Yes, I've sent the patch to guix-devel for comments
<adfeno>Another thing that could work out (I'm not experienced with Guix, I'm just trying to help by giving ideas): is to have the package name as a directory, have versions as subdirectories, and have "[revision number]-[hash]" as subdirectories of these.
<adfeno>Sounds weird actually. :D
<lfam>I hope that this problem (new hash, same version) is rare enough that we don't have to invent too much :)
<adfeno>Indeed... I also hope so.
<lfam>Usually, I'd file a bug upstream. But in this case, we were actually packaging a SVN checkout rather than a release tarball, so the situation is more complicated
<anthk_>sometimes I hope something Like LibertyBSD on design but from the GNU side :P
<anthk_>but that's impossible, because GuixSD is truly RMS' envision on GNU/Lisp Mit philosophy
<anthk_>oddly I like both, each approach is valid
<anthk_>but I understand GuixSD was for higher machines as in the 70's :p, because guix gc on my non SSD drive is lethargic
<vlegoll>anyone can help me make sense of that error message : guix build: error: /home/vince/guix-packages/my-own-linux-libre.scm:49:2: package `my-own-linux-libre-4.5.4' has an invalid input: ("kconfig" #<<local-file> file: "./config-vince-86_64-4.6.4" absolute: #<promise #<procedure 421f5d0 at ice-9/eval.scm:416:20 ()>> name: "config-vince-86_64-4.6.4" recursive?: #f>)
<efraim>sneek: later tell ng0
<sneek>Got it.
<efraim>sneek: botsnack
<efraim>vlegoll: can you post a link to your config?
<efraim> interesting log
<bavier>too bad 'guix lint' doesn't catch source uri redirects that lead to "missing interface for module (gnutls)" errors
<vlegoll>sorry went out, i'll pastebin the config
<efraim>I would try ("kconfig" ,(local-file "config-vince-86_64-4.6.4"))
<vlegoll>you mean just remove the ./ from the path ?
<vlegoll>I already tried that, same error
<efraim>how about ("kconfig" ,@("config-vince-86_64-4.6.4"))
<vlegoll>backtrace :
<efraim>i'd try ("kconfig" ,@(local-file "./config-vince-86_64-4.6.4")) or ,@("/full/path/to/config") next
<vlegoll>ERROR: Wrong type to apply: "/home/vince/guix-packages/config-vince-86_64-4.6.4"
<efraim>try going back to what you originally had and add the module (guix gexp) to the top
<rekado>why gexp?
<rekado>isn’t “kernel-config” just returning a string?
<vlegoll>I don't know scheme, it was an example Ludovic gave in
<efraim>local-file is defined in guix/gexp.scm
<efraim>and after grepping the source one of the results is this: /tests/packages.scm: #:use-module ((guix gexp) #:select (local-file local-file-file))
<davexunit>does anyone know which package contains the "hostname" program?
<davexunit>efraim: thanks!
<efraim>my find-foo is getting better :)