IRC channel logs
back to list of logs
<civodul>bricewge: oh; would you mind emailing bug-guix? <civodul>with that commit the same session should be reused <bricewge>That's what I thought too when reading the commit, I didn't understood the meomnize part <bricewge>At least one other person reported the same issue on IRC a month or so ago <apis_and_ipas>does anyone here have a Guix Home configuration with Exwm I can see? I'm having trouble getting things configured similarly to how I do on Arch. I see in the Pkg definition I'll need an .exwm file in $HOME, but I'm wondering if I'll be able to access my other packages from there.
***rekado_ is now known as rekado
<Ashy>how do i search for which package contains the pdflatex command? <Ashy>hmm actually, i think the real issue is that `guix shell` doesn't seem to pick up changes in the guix.scm file <Ashy>but `guix environment -m guix.scm` does, is that known behaviour? <Ox151>would anyone know how I can not have the prompt "Scheme implementation:" in emacs everytime I open a .scm file. over at #emacs they say its related to guile, but I have nothing setup with guile. <Ox151>MysteriousSilve geiser is installed in my packages but I havent done any configuration to it. <Ox151>i honestly dont know the usefulness of geiser, i just installed it when switching to Guix <Ox151>MysteriousSilve: yes that seemed to solve it, thanks so much. <Ox151>wait spoke too soon.... now its giving (error "Cannot find guile implementation") <Ox151>it opened the files fine for a second then after I sent that message it started giving the error. <Ox151>it seems I have to open them twice. once gives the error, the second time opens normally. <Ox151>just installed the guile package from guix, going to restart emacs daemon and see if that fixes it? <Ox151>same error with guile package in guix and geiser in emacs <Ox151>that `seemed` to work this time. thanks again. maybe one day ill figure out what geiser actually does for those packages to be needed to installed just to edit scm files <char[m]>How can I modify files in /etc/skel? I'm getting read-only file system. <char[m]>I solved by removing the symbolic link to /etc/static/skel and copying the files to a /etc/skel directory. <crodges>How do I reconfigure guix system from another distro? I hahve manjaro with guix as well, where I can mount guixsd partition <char[m]>messing with the skel directory causes a boot error <lilyp>char[m]: see the skeletons field of operating-system <lilyp>Obviously trying to mess with Guix-owned directories is a bad idea. <lilyp>crodges: you could try to get a chroot working, but honestly it's often more pain than it's worth <char[m]>Of course, I just couldn't find what was in control of it. <lilyp>One would guess the symlink could clue you in... <vivien>Dear guix, did someone try to package nextcloud? <vivien>Were there some hidden deal breakers? <lilyp>vivien: we do have the client, but for the server it's probably understanding php <lilyp>might want to check if the client needs an update too <vivien>I want to install nextcloud on my server; if it’s possible I will definitely do it but it would be a relief if I could contribute it to guix (for security updates…) <lilyp>You would need to package it and add a service in pretty much one series <lilyp>Don't forget about the mysql dependency <lilyp>Further, owncloud has this very weird self-update scheme, which I'm not sure if it works well with the spirit of Guix <lilyp>(which nextcloud probably inherited) <lilyp>you might look towards Nix if they already have a service for it and adapt that <yarl>I am trying to package distcc, guix build --round=2 works. Still testing. I would appreciate if someone could take a look, that the first time I define a package. Just to check if I am not doing something completely stupid. https://paste.debian.net/1238262/ <lilyp>"free" is superfluous, "easy to install" subjective marketing buzz <attila_lendvai>is there a way to have a guile script that can both be started from the shell, and also loaded into Geiser? IOW, how do i detect in a toplevel form that the scm file is launched as a scipt from the shell? <attila_lendvai>i could do 'guile -e main -s' in the shebang line, except on guix i must use '#!/usr/bin/env guile', where it cannot have arguments... what am i missing? *attila_lendvai has found the paragraph beginning with 'For maximum portability' in the guile manual <yarl>lilyp Yes, the tests fails, I disabled it just temporarily and I will investigate why it failed. As for the synopsis and description, I took what's on the github page but I surely can remove "free" and "easy to install". Thank you for your suggestions! <ZhuAisi[m]>attila_lendvai: Though not portable, but it's valid to use `#!/usr/bin/env guile -s` for `env` provided by coreutils <attila_lendvai>ZhuAisi[m], env must get a single binary. /usr/bin/env -S guile -s may work, though <ZhuAisi[m]>It's quite convenient because Guile support `#! ... !#` block comment 😄 <ZhuAisi[m]>If you want to play this trick with emacs lisp, it's quite difficult <lilyp>attila_lendvai: use the good ol #!/bin/bash exec guile workaround <attila_lendvai>lilyp, isn't it the same i pasted? (modulo the extra jump to avoid a reference to /bin/bash) *attila_lendvai needs to run... :/ <yarl>In 18.2 Running Guix Before It Is Installed there is a note about ".. source newer than compiled .." and I just got https://paste.debian.net/1238275/ (notice the .cache) and make (as is the documentation) does not help, what should I do? <jpoiret>yarl: you should always run make when .scm files change in a local checkout <yarl>./pre-inst-env guix build -K distcc <jpoiret>oh, but distcc.scm is a new file, right? <jpoiret>you should add it to gnu/local.mk, in the GNU_SYSTEM_MODULES part, and maybe regenerate the Makefile using ./bootstrap and/or ./configure <jpoiret>maybe ./bootstrap and ./configure are not even needed, but it wouldn't hurt :p <yarl>jpoiret oh, thanks you. I was greping every file to see where "hello" is defined, I was thinking "distcc should probably be declared somewhere...". Thanks a lot! <jpoiret>although i'd suggest not creating a new file just for distcc, maybe you could add it to an already existing file <yarl>jpoiret, ah. Where do you suggest? <jpoiret>the funny thing is that the only example i could think of, ninja, is actually in its own file :( <jpoiret>but it does seem wasteful, maybe there could be a build-systems.scm file <lilyp>sneek: later tell attila_lendvai: sorry, you're right, I forgot to read your paste <jpoiret>but since it's made to work with GCC exclusively from what i read, you could consider adding it to gcc.scm <jpoiret>in any case, it should be pretty straightforward to move it to another file if someone else in patch review has a better idea, so you could just add it to gcc.scm for now <lilyp>there is already a build-tools.scm <jpoiret>oh, right, that seems like a good place <yarl>Well, IIRC, I already got it working with clang. Not sure. <lilyp>people tend to write more single-package files than necessary <jpoiret>i'd get discouraged writing all these #:use-modules <lilyp>the thing is, those tend to not get used very much at all <tom[m]1234>Hello . I would appreciate any help or advice! Please recommend single-board computer so that I can install Guix Trisquel, Parabola , Hyperbola and everything works for me! I know that today there is not a libre single-board computer , as FSF writes. I would like to understand what options are possible to use today. I found these options > freedombox,banana Pi,rock64. Is there a similar information page for free distros that goes into more <acrow>Yes, hello all as well as Mysterious Silver.
***yewscion97 is now known as yewscion
<EMax`0Mancer[m]>Where's the best place to ask for help with a package build? (Trouble getting documentation to build.) guix-help? guix-devel? <acrow>For a guix package I think you are in the right place. <EMax`0Mancer[m]>I'm trying to build awesome with luajit, and it works, except for lua-ldoc, it still seems to want to use lua 5.3 instead. (if I just comment out lua-ldoc, it builds fine.) <acrow>Emax: Can you put your package where we can see it? eg ix.org -- paste.debian.org or whatnot?
***arjan` is now known as arjan
<acrow>Emax: Maybe you just need to do a transformation on lua-ldoc to get to the same lua version that awesome is using? Though this is just a guess. <acrow>Congratulations Cividul for acting on and realizing some beautiful ambitions. <civodul>congrats everyone! it's not a single person's work :-) <arjan>the SSH timeout seemed to be caused by the target machine being sluggish, not the build environment <acrow>What does everyone want for their birthday? Guix-wise... <wonko-the-sane>10 years old?, hmm i think guix transcends spacetime, its from the future! <EMax`0Mancer[m]>(I probably also should be "inheriting" from the regular awesome package, I suppose) <acrow>Emax: I'm trying to build awesome-luajit right now.... it may take a bit. <acrow>EMax`OMancer: It got off to a good start but you, apparently, have a better hardware budget than I do. Somewhere in there "-j 8" was too much to ask for... <acrow>EMax`OMancer: The build fails but is it is in ldoc. Is this what you saw? <EMax`0Mancer[m]>yes, that what I saw. and it seems to refer lua5.3, which I think is the issue. <EMax`0Mancer[m]>(if I comment out `lua-ldoc`, then it builds and runs without issue) <acrow>EMax: Yes, I see the reference to 5.3 also and agree that this is suspicious bc your intention is to use 5.1, right? <acrow>EMax: well, maybe leaving out lua-ldoc is the right thing to do. <EMax`0Mancer[m]>yes - and I'm guessing that could be the issue. though it complains about "module 'pl.class' not found", which I'm not sure whether that's actually a lua version issue or something else. <acrow>EMax: I'm able to reproduce it. It builds wo lua-ldoc as a native-input but fusses if you include it. Do you know a way to check for the availability of the documentation from inside the installed awesome-luajit? <EMax`0Mancer[m]>(a different packaging question - are there any examples of package builds using `dpkg` to extract+install binaries from a .deb ?) <EMax`0Mancer[m]>acrow: I have no idea about checking documentation from inside of awesome(-luajit) unfortunately. <acrow>EMax: I do not know of such an example; though I suspect it could be done. <acrow>EMax: I think we are on two different tracks now. 1. awesome-luajit builds! (celebrate) -- 2. You want to build it from the source in a .deb pkg? (I don't understand the motivation aside from the hacking glory you may achieve.) <acrow>EMax: 3. perhaps verify that the docs got into the built packages. I see only the minimal standards in the package I built. And that may be ok. <acrow>EMax: Alternatively, add a phase to install the missing docs... <acrow>EMax: maybe you want to discuss the (package (origin (deb-src "whatever"))) notion with others here. I imagine that others have already given it some thought. Maybe such guile has already been written. <lilyp>after hosing /boot/efi, is there anything I can do that's quicker than reinstalling guix? <apteryx>lilyp: chroot into your old system using a bootable guix/GNU+linux iso and reconfigure <lilyp>apteryx: segmentation fault, so chroot is out of option <heyarne[m]>Hi! I'm trying to set up guix on an foreign armhf system. `./guix-install.sh` ran fine, but trying to run anything else fails. In the output of the failed derivations I can repeatedly find the line "In procedure getaddrinfo: Servname not supported for ai_socktype" <heyarne[m]>Can somebody help me figure out why the error is raised? <arjan>@lilyp you could probably build an image from your system config and just get the efi partition contents from it <lilyp>apteryx: nope, still segfault <apteryx>lilyp: eh, sorry, I'm out of ideas then. I'd try a couple various ISOs to see if chroot always bail out like that <lilyp>hmm I think it might be bash that segfaults <lilyp>yup, and gash gives me no such file or directory <lilyp>image also appears to do nothing <heyarne[m]>apteryx: was the nscd advice meant for me? i tried installing nscd but I still get the same error (it's running as described in the manual), but still no dice… I've uploaded the shell logs here http://0x0.st/ockr.txt and the build logs mentioned in the previous file here http://0x0.st/ocki.txt <lilyp> okay, it appears as though the guix system command itself does nothing <apteryx>heyarne[m]: yes; I thought it had to do with glibc plugins not working correcly <apteryx>heyarne[m]: networking works fine on the host? <apteryx>perhaps there are things missing in your /etc/services file? <heyarne[m]>yes, networking is fine. i can curl any of the addresses in the log <apteryx>you could try backing up yours and using that one instead, to see if it might help <apteryx>I'm running out of clues, though, so hopefully someone more knoweldeable w.r.t. libc will tip in <heyarne[m]>Ah thanks! I didn't have /etc/services and just created an empty one. The substitutes seem to be updated fine after using the file you catted <lilyp>oh, just for the record, it's one of those dreaded 1.3.0 installation disks <lilyp>very dreaded 1.3.0 installation disks <apteryx>lilyp: can you strace the failure of chroot? it may point to something else being borked <lilyp>ld.so.preload appears to be the culprit <lilyp>either that or /proc/self/exe <char[m]>lilyp: thank you, the skeletons option looks really good for me. Do you happen to know what controls /etc/bashrc? My changes there keep getting reset. <lilyp>no one replies in that speed it's cool <apteryx>PotentialUser-97: are you attempting a graphic install from the 1.3.0 installation disk? <lilyp>the good thing about Guix is that OS reinstalls are super clean :) <lilyp>except that the store doesn't appear cleaned up at all <lilyp>does “guix gc” detect stuff in /gnu/store that's not in the db? <acrow>hmmm... isn't there a guix gc command for that.... gotta go read. <lilyp>you can probably hack something with comm <acrow>lilyp: What about --verify=repair ? <lilyp>that's for stuff that exists but is broken <lilyp>i.e. exists according to the db <lilyp>okay, maybe the store really has been cleaned, but i have enough other trash <acrow>lilyp: I'm wondering if it would help to do "guix gc --verify=content,repair" The output might then be filterable into a long "guix gc -D <corrupted_files>", no? <lilyp>guix gc did collect all those untraced files <lilyp>just noticed the difference in output between 2553 MiB and 97.599,34089 <lilyp>weird way of formatting an integer, but I shan't complain <lilyp>in other words, guix reinstall is dirty, but gc is clean <acrow>guix delivers again and on its birthday! <zacchae[m]>Today is Guix's birthday? Better mark it on my calendar :) *the_tubular Sings happy Birthday <the_tubular>I haven't listen to it yet, but apparently Stallman did a speach on the state of a few package manager, anyone got it ? <akonai>are only python libraries supposed to be prefixed by python- and go in the python-* files? <apteryx>akonai: yes, or packages that include both Python command(s) and a library <apteryx>if it's purely designed to be used as a command, then you can drop the 'python-' prefix <ss2>such a cool gif in the blog post. Congrats for ten years! <arjan>udevd is using a lot of CPU and memory continually from boot on my armv7 BeagleBone Black, any idea what could cause that?