<ison111>Trying to make a xf86-video-amdgpu package. Seems pretty straightforward by just copying some existing definition in xorg.scm, and it builds successfully. However, when booting I still see "couldn't load module amdgpu module does not exist", so I'm guessing there's more to it? <ison111>lspci reports that I'm using kernel driver "amdgpu", but if I try to configure it (with xorg-configuration-file) then I get the error that there is no amdgpu module. Since guixsd seems to not have a xf86-video-amdgpu package I figure that's what I need to do. <reepca>ison111: could you post the snippet of your system configuration? <ison111>But even without any of that lspci will still report amdgpu as being used, so that seems to be what it wants by default. <reepca>ison111: try putting #:modules (cons xf86-video-amdgpu %default-xorg-modules) in the xorg-start-command and xorg-configuration-file call. <reepca>on an unrelated note, it'd be great if the spinner from "guix package" cleaned itself up when it was done... I was very confused when I saw "-168 packages in profile". <leungbk>Starting download of /tmp/guix-file.lnEEmc <leungbk>X.509 certificate of 'melpa.org' could not be verified: <leungbk>with http instead of https, it downloads properly; how can i resolve the certificate issue? <apteryx>leungbk: Have you set the env var for the nss-certs package as explained in the manual? <apteryx>reepca: there's already a bug about the spinner <reepca>anyone else wanna try to guix download ftp://tug.org/historic/systems/texlive/2018/texlive-20180414-texmf.tar.xz? It keeps getting stuck for me, which is annoying since it's the first step in a guix package -u at this point. <reepca>and it works just fine with wget <apteryx>it's a huge file. is your connection stable? <reepca>as far as I can tell, yes. I've managed to download a bunch of other stuff just fine, including that same file twice now with wget. <reepca>(also it's gotten stuck at 100% about 4 times now) <johnjay>someone told me in another ch annel that guix is very actively developed <johnjay>is that something you can measure like with commits per day or something? <efraim>It looks like openhub is down. Maybe check someone's github mirror for a graph? ***rekado_ is now known as rekado
<rekado>reepca: if wget works fine try “guix download” on the downloaded file ***jonsger1 is now known as jonsger
<reepca>rekado: as in "guix download ftp://tug.org/historic/systems/texlive/2018/texlive-20180414-texmf.tar.xz"? or as in "guix download file:///home/reepca/texlive-20180414-texmf.tar.xz"? <reepca>update successfully continues! Thanks. Although now I wonder whether we have a bug in guix download. guix --version reports 707efe171a4e0e542a7d969c130195fa94b5d615. <rekado>I want to use “guix pull”, but I need to make one change to the configure flags before building the new Guix. <rekado>(I need to override the localstatedir configure flag) <janneke>rekado: can't we have configure be smarter to ddrt wrt localstatedir? <rekado>in my case localestatedir is set to /gnu/var/guix; I don’t think there’s a sane way to detect non-standard locations. <janneke>wouldn't it be safe to assume that the current guix's localstatedir is the best default? <janneke>if so, we'd have to find a way to extract localstatedir from it <rekado>I don’t think we can assume that. Generating a VM, for example, that includes Guix should not use the host system’s localstatedir. <janneke>sure, but wouldn't it do as a better default than /usr/local/var/guix? you'd override that anyway? <rekado>janneke: yeah, the configure default is bad. “guix pull” overrides it. <rekado>the only people affected by this bad default are those who build from source. <nly>i don't know which service/package contains the culprit package, maybe i can remove it? <rekado>guix build: error: build failed: getting attributes of path `/gnu/store/3aigj659vsjxcrhn0r0vmkbjh3kj5pbs-guile-2.0.9.tar.xz': No such file or directory <rekado>I’m still trying to switch from the old pre-inst-env shared Guix installation to one based on “guix pull”, but it’s tricky. <rekado>especially because of this localstatedir issue. <nly>I can show you the log <rekado>are you building this on a machine with little RAM, maybe? <rekado>(not that this would excuse a segfault) <rekado>nly: have you disabled substitutes (on purpose)? You may get a built binary for this package from ci.guix.info. <nly>hi thanks for looking into it <nly>I think I have enabled substitutes <rekado>is this a reproducible segfault? <nly>i ran the command guix build -f config.scm <nly>I was not trying to test reproducibility if that's what you are asking <rekado>no, I’m asking if this happens again when you do this a second time. <nly>I have authorized ci.guix.info following the manual, same error <janneke>aaarghh; a patched source needs xz, which proves severely non-trivial to bootstrap <janneke>now that i have `patch' somewhat working, upstream only provides `.bz2' *janneke is waay too discouraged to go down that rabbit hole right now, repackaging... <mckinley>Wondering if anyone has knowledge on packaging Docker (engine/cli) for guix? Whether it has been undertaken before? <davexunit>I recently saw patches for docker on the mailing list <mckinley>:davexunit :rekado wow, thank you both so much! <janneke>well, binutils also only ships .bz2 -- what a mess <janneke>i guess i'll first add bzip2-mesboot0 too, then :-( <bavier>janneke: could we adjust 'patch-and-repack' to compress to something other than xz? <janneke>bavier: that might be helfpul...but wouldn't it break the build if .xz source is already present? *janneke is debugging a hanging patch-2.5.9 -- apparently it was heavily patched by disto's <bavier>janneke: I guess I was thinking a configurable compression type, which could be used just for early bootstrapping <janneke>hmm, patch-2.6 is from 2010 and needs awk <mbakke>Is it OK if a Guile-Git dependency is fetched using "git-fetch"? I suppose it will lead to bootstrapping issues. ***lostcoffee is now known as atw
<roelj>Is there a way to see the build log while a build is running? <quiliro>hello...is there some newbie tutorial for useing ruby to communicate with an FTDI device? <thomassgn>hehe, I got some kind of notification sound that goes off once in a while. And I can't figure out what it's from or for.... :) <_tibbe>Hi, is there a way to remove a substitute server from the list of trusted servers? <thomassgn>I mean foreign distro with guix package manager? <bavier>_tibbe: you can adjust the default authorized keys in your system configuration, see 'guix-configuration' in 'Base Services' in the manual <bavier>_tibbe: otherwise you can pass a custom --substitute-urls to each invocation, or via the GUIX_BUILD_OPTIONS environment variable <_tibbe>thomassgn: As well on guixsd and guix on foreign distro <_tibbe>bavier: Thats is what I do right now but there must be some place where guix stores the certificates that where once passes to guix archive --authorize and way to remove them <_tibbe>I will try that... But will still not work under Debian (the foreign distro). After all the daemon needs to fetch the urls from somewhere. <bavier>_tibbe: there's an acl file that the deamon uses, I forget where it resides... <bavier>_tibbe: /etc/guix/acl by default, it is human-editable, see 'Invoking guix archive' <bavier>I'm assuming we'd eventually want 'guix archive' to have a '--deauthorize' option at some point <_tibbe>bavier: Yes I was wondering whether there was some hidden option for it which I just could not find <_tibbe>Some kind of interface that can list, add and remove substitute urls would be nice <_tibbe>bavier: And that file is not part of the store and can just be edited in place? <bavier>_tibbe: right, at least on a foreign distro; I'm not sure about guixsd ***amz3 is now known as ali
***ali is now known as amz3
<pkill9>also it would be good to be able to add a note or something to the key, so you know who's it is <bavier>pkill9: spki specifies a comment field; at this point it would need to be added manually <leungbk>when i run make check, i fail the guix-pack and guix-download tests; would someone help me debug this? <leungbk>unshare: unshare failed: Operation not permitted <rekado>leungbk: the “unshare” thing probably means that on your system you cannot use user namespaces (“container” features)