IRC channel logs
back to list of logs
<graywolf>So I have this weird problem: The outputs in Xorg are not hot-pluggable. When I connect the USB dock station, the devices are registered (I see them in /dev/dri), but xrandr does *not* show the outputs. If I restart Xorg (back to tty and startx again), the output are visible and usable. Any ideas how to get the outputs to hot-plug correctly? I have to admit, this is slightly annoying. (It works <radio>A firend sent me a link from lists.gnu.org and xinit, and it does seem to make sense, but I am still unable to start X as a normal user <graywolf>radio: Can you share more details, for example how yours startx script is looking? I can start X as normal user, so it is possible. <radio>graywolf: I am now in the tty, so I can't sent you the link I am refering to, but I saw someone saying it was better to use xinit directly, because sx and startx don't work well with the filesystem hierarchy of guix <radio>I'll sent u my script to start X <graywolf>Are you using guix home? How did you install X? <radio>I am using guix home, but I have to admit I don't understand well yet how it works <graywolf>Then I can guess what the problem is; X is install into ~/.guix-home, so the module path is wrong <graywolf>I'm in process of rewriting it into guile, but I was using it for past ~2 months and it works fine. So give it a try. <anemofilia>I just have to change the wm I want it to open, because xmonad is acting strange <dcunit3d>i can't seem to get shepherd 0.10 to work. it just won't load modules and fails with errors about continuation bounaries. <dcunit3d>i've tried everything i can think of. i was in the middle of writing an email to the email@example.com list. i was capturing strace logs, etc, but i wasn't able to send them. since either shepherd or xsession starts my xdg-desktop-portal, i can't send attachments with thunderbird or a browser. <dcunit3d>i thought i'd just wait on that, but for some reason i can't get xsession to start either. i was probably going to ask elsewhere for that, but i'm supposed to go to a small local conference tomorrow and i've already been up like 36 hours. it's at a community college where every few months, i have certificate problems on their wifi, so that's probably going to be a small hiccup. i'm just a bit stressed out because i was hoping to <dcunit3d>network to look for a job and now that would just backfire. i'm on my desktop at the moment. <dcunit3d>and the cert problems are like def. something weird on the college network. i'm just venting a bit here, sorry. i'm glad that shepherd has added a few features, but i'm in over my head. <bjc>i'd say just rollback if you can, until you have more time to debug it <dcunit3d>it's in my dotfiles though. i wish i had moved to guix home already, but i just don't know very many people in my local area to interface with in person. it's hard for me to get around. it's definitely a microsoft town, but there are a few people. i just don't get out much or meet tech people bc there's so few meetups. <dcunit3d>i feel like shepherd is a great way to learn about IPC. reading through the tests in that project really opened my mind up. i printed some out with enscript -> *.ps :) <dcunit3d>and ive learned so much through guix and the project that i simply never encountered elsewhere. i think i'll move to guix home soon, but i donno. i'll be alright, i'm just stressed i guess. <anemofilia>how can I check the value of a variable like %base-services or %setuid-programs? <bjc>you can do it from ‘guix repl’ with ‘(@ (gnu services base) %base-services)’ and ‘(@ (gnu services system) %setuid-programs)’ <superkamiguru>So I have a simple service that I am trying to run on boot, and it doesn't seem to be doing anything to the directory permissions. Anyone have an idea why? paste.debian.net/1280715 <apteryx>mirai: is it because GUIX_LOCPATH is not preserved and glibc-locales missing from the environment? <mirai>hmm, GUIX_LOCPATH isn't set in the non pure environment as well <mirai>yet it still works correctly <mirai>adding glibc-locales to the guix shell environment doesn't help <mirai>I'll file a report on it in a few <mirai>unrelated to the utf8 thing, I think the “simple enterprise” of split a <HOST>:<PORT> (or endpoint as you call it in 2/5 #63402) is a big can of worms <mirai>did some more research on the topic and I think not even the (ice-9 peg) based snippet I pasted would cut it <mirai>the correct way of things is to follow what RFC3986 says, even though <HOST>:<IP> is not a URI, it is a “portion” of a URI so the same rules should apply <mirai>of course cooking up a partial RFC39866 impl. is a significant undertaking for the wireguard purposes and these kind of efforts should go into Guile (web uri) <mirai>I'd like to rework the (web uri) in guile to expose some partial or more exotic uri manipulations in the future when I get the opportunity to <mirai>but in the meantime, for #63402 I'd say the best course of action is to craft a dummy uri, by concatenating "dummy://" with the “endpoint” field <mirai>and retrieve the host & port from the returned #<uri> record <mirai>you get validation and don't have to worry about the sharp corners this way (make it Guile's problems) <altomcat[m]>I need advice. Is it better to provide a patch attached to the bug request or to create my own patch and mention the relative bug in it? What is the most effective way to do so? <abrenon>well if you already have a patch, might as well send the patch while describing the problem <abrenon>you usually write to firstname.lastname@example.org, with your patch attached (or send it directly from git but I've never used that) <abrenon>there'll be a short delay for manual moderation if that's your first patch <abrenon>and then an issue will be opened and attributed a fresh number <futurile>Q: is there a way to specify a development package in a manifest - so the equivalent of guix shell --development vim <abrenon>altomcat[m]: oh sorry I hadn't understood you correctly <abrenon>there is already a patch, then of course send your patch as an answer to the same <abrenon>except if its scope is really much wider and it only "happens" to fix the reported bug <abrenon>see, in the last message there already is an attached patch <abrenon>futurile: not that I know of but you can still mix -m and -D -f on the same command line <abrenon>I've doing this these days on a particularly messy set of packages, on I was still developping and which is more or less destined to be integrated in another repos whic as of now is only a heap of dependencies as a manifest and a set of scripts using it <abrenon>so I'm opening my dev environment with : guix shell -m ../that-other-project/manifest.scm -D -f guix.scm <some-one>Hi! Can anyone check if ci.guix.gnu.org is reachable? <ngz>some-one: It doesn't seem to be. <some-one>Ah, wonder how long it takes me to update without substitutes =[ <ngz>I still can fetch substitutes, tho. <some-one>making '/gnu/store/g113g4k50lmzqyy9gmah2bran3nvzx1x-system' the current system... <some-one>setting up setuid programs in '/run/setuid-programs'... <some-one>populating /etc from /gnu/store/3sh5a1qg2alpbd0vfwi9i5kx7944yn7f-etc... <degauss>Hi! Is anybody here familiar with the setup of Guix' debbugs instance? <cbaines>I've started nginx on berlin now so ci.guix.gnu.org should be back <cbaines>degauss, I don't believe Guix has a debbugs instance, we use the Gnu one <degauss>Thanks cbaines, I was wondering if there may be a way for particular projects that have been configured in "exclusive" mode (according to debbugs.gnu.org/Using.html) to manage how bug replies are processed, but there probably isn't. It feels weird that the original poster of a bug doesn't get copies of replies sent to that address. I thought if was the case in the Debian instance, but I may be wrong… <Kabouik_>I tried to add a package to my private channel and it produced an error I was not able to fix when pulling, so I removed it from the git repo and guix pulled again, but the issue still occurs: https://0x0.st/Hqs_.txt. How can I fix that? <alethkit>Does anyone happen to have a staging channel for a KDE service? I don't want to reinvent the wheel by packaging my own service. <Kabouik_>I fixed my issue mentioned above, the package was in fact still in my private channel. Would be interested in fixing the package though, if someone understands the error message. <sneek>macrocreation, you have 1 message! <macrocreation>Wondering when you do a inherit - how do you do things like change the inputs list etc. For example if you want to take an existing package def and remove some inputs and add new ones? <oriansj>anemofilia: then don't include sudo in your package list <lambdanil>you can remove sudo from %base-packages if that's what you're asking <lambdanil>(define my-packages (remove '(lambda (x) (string=? (package-name x) "sudo")) %base-packages)) <lambdanil>I think you don't need the quote before the lambda actually <anemofilia>how can I see all the packages a module provides? <janneke>are you looking for the output of: "guix show sudo"? <civodul>anemofilia: "guix package --list-available | grep gnu/packages/admin" for instance <macrocreation>I am using inherit to modify a package definition. In the original definition I modify standard phases. Is there a special thing I need to other then use modify-phases %standard-phase to change the original definition <anemofilia>using a package module doesn't mean it imports all the packages a module provides, right? <anemofilia>because uwufetch, for example, is in (gnu packages admin), and I don't have it <oriansj>anemofilia: it just means they are available to install in your configuration <lambdanil>importing a module makes the definitions available <lambdanil>so you can then add the packages from the module into your package list <lambdanil>you'll have to remove it from %base-packages if you don't want it <anemofilia>iufhshfiuhd okay then, I came here thinking it was, but maybe I misread someone's message as if it wasn't <andreas-e>There is a new wget release. It has between 900 and 1000 dependents to be rebuilt. Will this be picked up by QA? <andreas-e>Is this little enough to update it directly on master? <macrocreation>Hi Everyone, is there a docs system for the common methods like (subtitute-keyword-arguments) <ngz>macrocreation: I don't think so. There the docstring attached to the macro, but it isn't exactly clear either. <ngz>Then, there are examples across the code base. <ngz>Such information, if it is ever written, could go into the Cookbook. <macrocreation>does anyone what these symboles mean e.g. ~(modify-phases vs #~(modify-phases <macrocreation>I suspect these are scheme macro annotations but I can't find them anywhere <ngz>#~ are called G-expressions. There are explanations in the manual, and, I think blog entries about them. <civodul>ACTION is amazed that 'git send-email' automatically adds X-Debbugs-Cc <macrocreation>'(modify-phases ,phase -> #~(modify-phases #$phases fixes things <civodul>you can use "guix style -S arguments" to make this transformation automatically <guixfren>also isn't package already a macro? so then you'd be trying to put a macro inside a macro <macrocreation>Still would appreciate pointers to where this is explained. I just hate having vodoo edits that work. For example I know what #: is a keyword argument <guixfren>I'm not sure I was right with what I said, those were guesses more than anything so don't get the wrong ideas <macrocreation>guixfren - you were right. That command line snippet would have fixed it - but would not have extended my knowledge. <ngz>macrocreation: ` and , is basically the same as #~ and #$, but the latter happens specifically at build time. <macrocreation>Does anyone have an example of substitute-keyword-arguments which doesn't use gexpression <lilyp>If you don't do G-Expressions, you need quasiquotes, i.e. `(modify-phases ,phases ...) <lilyp>Note that you can't quasiquote your way through if ,phases already has G-Expressions <macrocreation>lilyp thanks - that explains the issue I was getting. It's actually working with suggestions from ngz and guixfren - but now I am trying to build an understanding by using alternative code. <oriansj>macrocreation: the best I can give you is Detroit or Lansing <davidl>Hi, I have an issue when compiling saying a recompilation is needed for a specific file - how can I do that? I want to avoid running a full make clean and restart. <rekado>I wasn’t aware of this and only just found out that the VST SDK has had its license changed. <rekado>since 2017 or so it can be either distributed under the terms of the proprietary Steinberg license or the GPLv3 <rekado>only for VST3, not VST2, but that’s great <rekado>we no longer need to patch out VST support then <anemofilia>how can I, without using the %base-packages, put guile in the packages system declaration? <guixfren>you'll have to append it to the package list <guixfren>(gnu packages guile) only imports the package definition <anemofilia>/etc/config.scm:80:25: error: guile: unbound variable <anemofilia>the line 80 is inside the packages definition, and only contains guile guile-readline and guile-colorized <efraim>the variable may be named guile-latest or guile-3.0, I can never remember the names of the guile packages <anemofilia>is (use-modules (gnu packages something)) equivalent to (use-package-modules something)? <anemofilia>(use-package-modules linux) should give a value to iproute2, no? <efraim>IIRC iproute2 is named iproute in (gnu packages linux) <unmatched-paren>any sourcehut user know how to change your payment details? i can't seem to find any options for that <anemofilia>/etc/config.scm:80:25: error: guile: unbound variable <anemofilia>(define (map-setuid-programs package package-programs) <anemofilia> (program (file-append package (string-append "/bin/" package-program))))) <anemofilia>Is this correct? It's intended to work like this: <anemofilia> '("passwd" "chfn" "sg" "su" "newgrp" "newuidmap" "newgidmap")) <anemofilia>/etc/config.scm:117:2: warning: representing setuid programs with file-like objects is deprecated; use 'setuid-program' instead <guixfren> (program (file-append nfs-utils "/sbin/mount.nfs")))) <guixfren>I see this as an example in the documentation <anemofilia>I've read the documentation, I want to understand the warning <anemofilia>For as long as I know, it should be fine to do this <sughosha>Hi, is it possible to add multiple keyboard layouts in `config.scm`? <sughosha>I would like to switch between "en_US.utf8" and "de_DE.utf8" in `sway`. <anemofilia>and the portuguese keys only work when holding the menu key <anemofilia>You can also make a key change the keyboard layout and keep it until you press it again, but then you have to read some wiki on setxkbmap syntax and stuff <anemofilia>Also, you can use other key than the menu key if you want to <anemofilia>The first element of %setuid-programs is like this, in the default: <anemofilia>+scheme@(guix-user)> (@ (gnu system) %setuid-programs) <anemofilia>$4 = (#<<setuid-program> program: #<file-append #<package email@example.com gnu/packages/admin.scm:949 7f80a90d2c60> "/bin/passwd"> setuid?: #t setgid?: #f user: 0 group: 0> <anemofilia>guix system: error: #<<setuid-program> program: #<file-append #<package firstname.lastname@example.org gnu/packages/admin.scm:949 7f06e24138f0> "/bin/passwd"> setuid?: #t setgid?: #f user: 0 group: 0>: invalid G-expression input <anemofilia>They are literally the same thing, (if you don't consider the thing that comes after the package module) <davidl>Hi, I sent in a new patch for an old issue, bug #51512 add bash-bcu, but I don't see it showing up in the email archives or the debbugs issues tracker. Can someone check what's going on? <davidl>...and ofcourse now it shows up.