IRC channel logs
2023-03-21.log
back to list of logs
<vagrantc>so diffoscope 238 required poppler-next, but 239 can work with either poppler or poppler-next ... i switched to poppler-next in 238, but should I switch back for 239 ? <oliver`>This is my first post ever in ERC! Glad to be here. <vagrantc>re diffoscope, i opted to keep poppler-next, but mentioned it in the bug <lrabbt>hello, I'm having trouble running "guix build git" to check the store folders <lrabbt>it spits a traceback saying the variable "pre-commit" is not bound <lrabbt>weird thing is it builds ok when using ./pre-inst-env on the most recent master <lrabbt>and as far as I know, it's only on "git" package <mirai>there has been some changes surrounding pre-commit <mirai>it was a duplicated package and one of its copies was purged <mirai>so maybe it has to do with that <mirai>I'm not too sure about it, maybe someone else can weigh in on the matter <lrabbt>bumble[m]: thanks, I'm using git on guix home, it's working fine, I don't really NEED to run the build, I was just trying to figure out why "git send-email" wasn't working, then stumbled on the build problem <lrabbt>I solved the "git send-email" problem, I was using two different profiles for git and git:send-email, but the "guix build git" problem persists <apteryx>lrabbt: what does 'guix describe' say? <podiki[m]>stumpwm needs newer autoconf for current commits (last release is fine as is), I don't think it breaks anything in current version <lrabbt>the "guix-my-channel" was just me messing around, it has only 1 package <podiki[m]>what would be the commit message for a patch there, something like gnu: stumpwm: Fix future builds. ? not sure (using a transformation locally is okay, but wants to rebuild a lot since it is an input replacment on autoconf) <apteryx>podiki[m]: perhaps hold the change until the next version is released <podiki[m]>apteryx: stumpwm did release twice last year after a 2 year gap, so that is promising... <lrabbt>(sorry, it's in portuguese, but the las error is the "variable unbound" one) <podiki[m]>maybe I'll submit the patch for anyone that runs into trying to use git stumpwm (it should fix some bugs I see for plug/unplug monitors which is why I'm interested) <podiki[m]>(to be clear the rebuilds are just for inputs in the stumpwm tree when using input replacement, actual package def change should just be the ~20 packages that depend on stumpwm) <apteryx>what is autoconf at on core-updates? <podiki[m]>stumpwm currently uses autoconf, but latest commits to stump require autoconf-2.71 <podiki[m]>"soon (tm)" :-) (props to Andreas, one person army) <podiki[m]>apteryx: looks to still be autoconf-2.69 on core-updates <lrabbt>apteryx: if we have the same guix main channel, shouldn't your "guix build git" fail too? <apteryx>lrabbt: depends of the commit, I'm taking a look at your paste above now <apteryx>hm. I was expected you may have been on an old commit <apteryx>lrabbt: I guess it's an issue with one of your channels <lrabbt>yeah, I had the problem yesterday, but I'm running guix pull everytime, just to be sure <apteryx>try isolating from which channel the error comes from, then report it to them <apteryx>it'd be nice if guix had better errors in this regard <lrabbt>ok, I'll do that, is it normal that the trace points to gnu/packages/python-xyz.scm, though? <lrabbt>either way, thank you for the help <onemorepeter[m]>I will teach you how to earn $1k with $200 in just 2hours from Blockchain … if you’re interested message me on Telegram on how to start earning… <lrabbt>it's kinda late for me, I'll drop <lrabbt>I'll try to figure it out tomorrow <lechner>sneek / later tell ArneBab_ / Bitte entschuldige, dass ich Deinen Namen im Email etwas verkuerzt habe. Es war ein Versehen, und aus Eile <sneek>lechner, you have 1 message! <sneek>lechner, jlicht says: you pointed out guix-patches@debbugs.gnu.org, but it should probably be guix-patches@gnu.org. The wrong address used to be in the manual, but should have been fixed :-) <lechner>sneek / later tell jlicht / thanks for the pointer about guix-patches@gnu.org. i get tripped up because the address for bugs is XXX@debbugs.gnu.org, and I still use the old interface from time to time <bumble[m]>I'm trying to use the guix emacs packages rather than emacs' package manager. "emacs-use-package" is installed by guix home, but when starting emacs with an init.el that uses use-package this error occurs: Warning (initialization): An error occurred while loading '/home/bumble/.config/emacs/init.el': Symbol's function definition is void: use-package <bumble[m]>I hope someone here will tell me what I am doing wrong <bumble[m]>I am using the current "emacs" package and not the "emacs-next-pgtk" <bumble[m]>my ./guix-profile/share/emacs/site-lisp is empty... do the packages need to be installed in some other way, rather than simply using "guix home reconfigure" on my home config file? <bumble[m]>... there exists a ~/.guix-home/profile/share/emacs/site-lisp on this system perhaps that should be used instead... <bumble[m]>this channel is so quiet tonight, did I say something wrong and everyone is ignoring me? <unwox>bumble[m]: don't overthink it, it's just a quiet night :) <civodul>no regression compared to the previous successul one apparently <Janio>Anyone use hpaio on Guix? My HP scanner non present in HP backends. Help me preas. <sarg>hey guix, when I use the recommended `git send-email` command from `22.6.2 Sending a Patch Series`, the patches end up deeply threaded (in git format-patch wording). I think the preferred way is `shallow threading`, so the info pages should be updated with appropriate `format-patch` and `send-email` flags. <TristanCottam[m]>Hi guys, quick question: Do you keep your Guix Home configuration files in a separate repo? <jpoiret>sarg: if you're sending a new issue, make sure to send only one mail at first, then once you get an issue number you should reply to that new mail adress and not guix-patches, otherwise a new issue number will get opened for each mail <sarg>jpoiret that I know. But when I `git send-email --to=nnn@debbugs *.patch`, they end up deeply threaded. <TristanCottam[m]>In the case of Emacs, what are the benefits of storing configuration in a custom package definition? <jpoiret>yeah, i think it's the default unfortunately <sarg>jpoiret time to fix the defaults. I'll send a patch for the info pages later <sarg>also I assume that guixers are also emacsers, so maybe we could add a section about preparing and submitting patches with magit? <jpoiret>i think it's a git default though :p you can change it with a global .gitconfig <sarg>it's just two flags: `--[no-]thread` and `--[no-]chain-reply-to`. I need to figure out which one to use though <sarg>and the defaults should be put in `guix/.git/config`, `[sendemail]` section <sarg>`issues.guix.gnu.org` displays the same diff correctly <sarg>jpoiret: may I ask, what is `core-updates` for? Sorry, don't follow `guix-devel` closely. <jpoiret>up until now, it's been the branch where changes that cause a lot of rebuilds went, to be merged periodically. We're working on merging it into master now. Hopefully we'll stop using it after this (at least with that meaning) <rekado>sarg: it’s odd, but the list interface also processes text in order to remove what it thinks are email addresses (even when they are not) <jpoiret>the main problem is that it bunches up a lot of completely unrelated changes, and so when the time for a merge comes, a lot of work has to be done. The switch to feature branches (if it does happen) should make those efforts more focused <sarg>jpoiret: got it, thanks. Branches would be nice for bulky patchsets. E.g. this one: `[PATCH 00/70] gnu: Add mautrix-whatsapp.` <Guest59>After uninstalling guix with apt, pre-inst-env finally doesn't pick up the wrong repository anymore and I can hack on packages <Guest59>Weird that it wouldnt work otherwise, and now I have to manually add the daemon user and group <evilsetg[m]>Has anybody here had problems with nm-applet and eduroam connections? For when trying to configure the connection nm-applet crashes with Settings schema 'org.gnome.nm-applet.eap' is not installed <nutcase>I managed to add the connection via nmcli in the end <civodul>evilsetg[m]: hi! i don't use nm-applet (only nmtui), but Eduroam works well for me via NetworkManager <evilsetg[m]>I managed to set up the connection with networkmanager-qt. I remembered from some time ago that nmcli did not support wpa2 enterprise connections, but maybe that is outdated. <nutcase>evilsetg[m]: this worked for me: nmcli con add type wifi con-name "eduroam" ifname "wlp61s0" ssid "eduroam" wifi-sec.key-mgmt "wpa-eap" 802-1x.identity "myAccount@myUniversityDomain.tld" 802-1x.password "myAccountPassword" 802-1x.system-ca-certs "yes" 802-1x.domain-suffix-match "myUniversityDomain.tld" 802-1x.eap "peap" 802-1x.phase2-auth "mschapv2" <nutcase>but still, if I want to edit the connection via nm-applet, the nm-connection-editor crashes <evilsetg[m]>nutcase: Nice to know that nmcli works now for eduroam. <nutcase>evilsetg[m]: yes, both are useful I think. I wasn't aware of nmtui. However I now archived the nmcli command in my notes and next time I have to set up the eduroam connection it's quite straight forward (hopefully) :-) <evilsetg[m]>I think I will write down the command into a small script for future use. <nutcase>or you put your decredentialized /etc/NetworkManager/system-connections/eduroam.nmconnection into some git repo. I didn't figure out, if NetworkManager can pick up connection definition files from my home directory. If I just symlink /etc/NetworkManager to somewhere inside $HOME, the target of the link gets owned by root. <evilsetg[m]>When I send a patch revision after a first patch the manual tells me to use git-send -vREVISION. Should I start counting at 1 or at 2? I am unsure because maybe the first patch should count as v1. <user_____>guix pull is not recognizing my installed packaged for some reason <rekado>user_____: can you provide some more details please? <user_____>I have 243 packages installed, but whenever I issue guix pull command it end up "compiling profile with 1 package" <bjc>user_____: ‘pull’ doesn't upgrade packages, it just updates their definitions. use ‘upgrade’ to change what's actually installed <user_____>I tried guix upgrade. and got "nothing to do" <minima>sarg: hey thanks for thinking of me and sending the update, very kind; and great news that abcdw has been working on this <user_____>hmm something is really wrong here, after I did guix pull I tried guix search linux-libre just to check <ngz>user_____: What is the output of `which guix'? <ngz>This is not right ; it should be /home/user/.config/guix/current/bin/guix <ngz>Did you install `guix' as a package, e.g. guix install guix ? <ngz>Or something is wrong with your $PATH <ngz>Try guix remove guix and `which guix' again <user_____>guix 1.3.0-32.682639c out /gnu/store/2sfyaxf13ckajiwvsrqv667n583a47qx-guix-1.3.0-32.682639c <ngz>For the rest of humankind, I sincerely hope I'm not. <ngz>user_____: you're welcome! <rovan>Not strictly guix: Why does tail -F /var/log/some.log > zstd -o some.log.zstd not work? <sarg>rovan in what sense it doesn't work? `-F` is for follow, you'll need to interrupt the process <rovan>sarg: Work as in nothing is ever written to some.log.zstd even though /var/log/some.log is filled with data. <sarg>oh wait, why `>`? it should be a pipe <rovan>Sorry, that is a typo. It is a | in my terminal. <fislysandi>can you help me out fixing broken upstream? by that i mean i have this 'aborting reconfiguration because commit d5fece6bfe6b2eaf93f936a4a6dea8fbfe118140 of channel 'guix' is not a descendant of 01fd830f2fdd388f56e6e00df747f052bbde3906' <fislysandi>i tried everything and i cant fix it it feels my config is cursed. <fislysandi>im trying to make something called guix kickstart <sarg>rovan when you Ctrl-C you kill `zstd`, but you need to kill `tail` instead <sarg>did an experiment with `echo "brrr" >> test` and `tail -F test | zstd -o test.zstd` in another console. If I `pkill -f tail` it works as expected <rovan>Right, zstd waits for an EOF before starting to work? <sarg>fislysandi: try with `--allow-downgrades` <sarg>you shouldn't have to. Do you use the default guix channel? <fislysandi>its a self imposed issue because i tried to replace the upstream repo of guix with a mirror to it to make it faster <sarg>you can checkout guix repo locally and use it as a channel, though not sure it'll boost the speed <sarg>use the default channel and be happy. No regrets <abcdw>libgtk is still missing debugging information. <sarg>hmm, you need `--allow-downgrades` just once. What does `guix system describe` show? <fislysandi> commit: d5fece6bfe6b2eaf93f936a4a6dea8fbfe118140 <fislysandi>when i try to guix home reconfigure guix-kickstart i get this 'guix home: error: aborting reconfiguration because commit d5fece6bfe6b2eaf93f936a4a6dea8fbfe118140 of channel 'guix' is not a descendant of 01fd830f2fdd388f56e6e00df747f052bbde3906' <fislysandi>the problem is i need to cure my commit has to be equal to 01fd830f2fdd388f56e6e00df747f052bbde3906 so that it will not complain <cbaines>fislysandi, does the error message make sense to you? <fislysandi>yea i need to cure my channels.scm, putting channels.scm with the correct hash doesnt make this issue go <cbaines>that's not an explaination of the error message <cbaines>why is guix preventing you from using commit d5fece6bfe6b2eaf93f936a4a6dea8fbfe118140 ? <fislysandi>it wants to use 01fd830f2fdd388f56e6e00df747f052bbde3906 <cbaines>fislysandi, try reading the error message, the information is there <sarg>01fd a later one, it can't be an ancestor of d5fe <mirai>nutcase: you're going to have some issues with symlinking nmconnection files <mirai>NetworkManager is fussy about their permissions <mirai>and they can't symlinks iirc <fislysandi>ok i guix pull --commit=01fd830f2fdd388f56e6e00df747f052bbde3906 with the hash now ill try to guix home reconfigure again. <fislysandi>'guix home: error: aborting reconfiguration because commit d5fece6bfe6b2eaf93f936a4a6dea8fbfe118140 of channel 'guix' is not a descendant of 01fd830f2fdd388f56e6e00df747f052bbde3906' nope <sarg>fislysandi aren't you trying to move back in time? It seems that your system has newer commit already <cbaines>trying things without continuing to try and understand the problem is, well, trial and error, which can be quite slow <cbaines>fislysandi, I think the thing you need to understand is the error messages you've posted <apteryx>civodul: Hello! I think I'll have to break my promise to send a new bug report against 'guix style' for each one fixed soon, as they seem to be getting harder to come by! I've opened https://issues.guix.gnu.org/62303 to not forget about that one. <minima>all works fine up to `make` and i also go through `make check` mostly unscathed (with just a handful of fails) <minima>when i eventually try `make authenticate` though, i get an unforgiving `guix: command not found` <minima>(handful of fails = 2 fails as a matter of fact) <civodul>apteryx: heh, thanks for the new bug report :-) <civodul>TBH, i'm not satisfied with the code of the pretty printer <jpoiret>minima: if you don't have guix already available, you won't be able to guix authenticate :) <jpoiret>you wouldn't trust an unauthenticated guix to authenticate itself, right? <civodul>i feel like i'm missing a pattern to make it nicer, more accurate, and more flexible <civodul>but it does the job and there's good test coverage, so we can keep tweaking it <minima>jpoiret: i should have added that i'm on a foreign distro (but i do have guix as a package manager, if that's what you mean) <jpoiret>so you do have `guix` available in your path? <minima>yeah, definitely, but let me double-check i haven't messed things up with my `.envrc` envs <minima>jpoiret: ok, so i may have found the culprit, if i work from within a shell (`guix shell -D guix --pure`) then i'm not able to `make authenticate`; things are fine if i leave the shell (as in `guix shell ...`) though <minima>it's the -D and --pure together i suppose? <apteryx>civodul: OK! There are a lot of code formatters out there (for other languages) that we could perhaps review to get hints from (python-black, gofmt, rustfmt, clang-format, etc.). <apteryx>it'd be great for the Guile/Scheme community if we could eventually move the Guix pretty-printers bit into Guile itself, or to a separate project <abcdw>I would appreciate any help on loading debug info for shared libraries to gdb in guix! <jpoiret>minima: just the --pure means you don't have access to the installed guix <jpoiret>abcdw: inside the `guix shell` environment, is GDB_DEBUG_FILE_DIRECTORY set? <minima>jpoiret: right, i see; i'm a bit confused though on what commands on the manual page above are supposed to be run from within the guix pure shell and what directly from the system <minima>from what i understand, one has to leave the pure env before running `make authenticate` <abcdw>jpoiret: Yes, it's set and outside of the shell it doesn't <jpoiret>I personally don't use --pure for the guix checkout <minima>might it be worth to add a note to the docs that mentions that <jpoiret>abcdw: no idea, maybe you need to split GDB_DEBUG_FILE_DIRECTORY on : as well? <jpoiret>since you're filtering by file-exists? <nutcase>mirai: I can move /etc/NetworkManager to $HOME/something (e.g. ~/.local/etc/) and symlink to that location from /etc, but the owner of the directory is changed to root:root. That's what I observed. Though it still seems to work. <abcdw>jpoiret: it contains only one directory /gnu/store/hhkwqiwsq7xxi3zrq1h4hzh5vzza88wk-profile/lib/debug: http://ix.io/4rqQ <jpoiret>well, I'm out of my depth here then, just throwing out the usual gotchas <abcdw>jpoiret: ok, thank you anyway!) <mirai>nutcase: does it still work after a reboot? <mirai>note that NetworkManager only reads the .nmconnection files at a specific point in its startup sequence <mirai>you have to restart it or issue some nmcli command to make it "reload" the directory with the .nmconn files <nutcase>mirai: I think it works. My link (/etc/NetworkManager -> /home/nutcase/.local/etc/NetworkManager) is from Feb 20 and my uptime is 8 days, 4 hours. <nutcase>When nm reloaded (I think 'herd restart NetworkManager' is sufficient) the connection files' ownerships changed to root:root, I think <mirai>does it complain about anything? <nutcase>mirai: No, there is nothing NetworkManager complains about. Should it? It just works <nutcase>or do you mean in the moment the files' ownerships changed? <lechner>sneek / later tell bumble[m] / it was not anything you said. mondays are usually quieter than other days. also nckx, who was the heart and soul of this channel, somehow disengaged. it probably did not help that i got into an argument here yesterday. people avoid mean places and instead spend time with friends <mirai>nutcase: does it not demand that the nmconnection files be regular files <mirai>in any case, do try a reboot <mirai>my memory is a bit faded in what the troubles were with nmconn <mirai>lechner: lets bury the hatchet <civodul>abcdw: is GDB_DEBUG_FILE_DIRECTORY set? <civodul>at worst, you can explicitly do "set debug-file-directory /gnu/store/.../gtk+-3.1.2-debug/lib/debug" to see if it makes a difference <TristanCottam[m]>Hi everyone, quick question: How reproducible are runtime errors with Guix? <apteryx>abcdw: I'd build without grafts, as there's a bug: #48907 <apteryx>abcdw: also if you use the sh trick I think it won't load the symbols <apteryx>so instead I use a 'run-gdb' home made script <TristanCottam[m]>Also, as a more general question, how can I debug runtime errors on Guix System? <abcdw>apteryx: I'll try to use binary directly. Will read a thread about grafts and gdb <apteryx>abcdw: good luck! and in some tricky cases, the debug symbols are not complete enough (they're made from -O2 optimized binaries often), so in these situations I end up building from source manually with the right flags <minima>abcdw: (i've been experimenting with `guix-home-service-type`, which works like a charm for a couple of tasks i wanted to crack, thanks for that!!) <nutcase>Is anyone able to run eclipse (Java IDE) on a Guix system? After downloading the installer and trying to run the installer, I get: zsh: no such file or directory: ./eclipse-inst. The file is in . and has 755 rights. This is kind of a show stopper in my case :-( <abcdw>apteryx: --no-grafts helped! Thank you very much <abcdw>civodul: The problem was related to grafting <abcdw>ACTION is going for a walk to fresh up the mind <sarg>nutcase: it's an ELF binary? ldd it, some dep is missing <podiki[m]>nutcase: is it just a binary? try the fhs option in a guix shell container; the "no file" error usually indicates it can't find the ld loader since guix does not have e.g. /lib, /bin <sarg>^ yeah, with fhs it became much easier to run foreign binaries <podiki[m]>I see guix does have a lot of eclipse libraries, but I don't know how it works (haven't used java in many many years and never on guix) <nutcase>podiki[m]: that brings me one step further: In a `guix shell --container --emulate-fhs openjdk@17:jdk openjdk@17:out zlib` I get a problem with SWT "SWT OS.java Error: Failed to load swt-pi3, loading swt-pi4 as fallback" and finally "Could not load SWT library." in the log after the application dies (nearly immediately) <jpoiret>you probably need java-swt in the shell as well <jpoiret>i don't know anything about java though :p <nutcase>jpoiret: tried that as well and it didn't make a difference <spiderbit>Hi I try to get equake going in stumpwm I added the code in init.lisp from that side that uses split-sequence function but when I reload the config it does not find that function/library. Yet I have explicitly installed it systemwide with: (specification->package "sbcl-split-sequence") <spiderbit>that is the code I added to my stumpwm config. is there something wrong with the way I installed the library so that stumpwm / sbcl can't find the library? <apteryx>hm, 'guix shell' hangs on 'substitute:' <apteryx>seems to be caused by my own substitute server ^^' <gnucode>unmatched-paren: hello pal! What's been going on? <gnucode>sweet frozen chocolate icecream! that sounds awesome! <gnucode>that will be helpful. Those suckers are really hard for me to understand! <gnucode>unmatched-paren: how would you feel about setting up a guix subscription for the guix.gnu.org/blog ? <gnucode>paid people get to see new posts immediately. everyone else gets to see the new post inside a week... <gnucode>I wish guix had somekind of business aspect to it....just my two cents. <apteryx>gnucode: you're free to start a co-op providing guix-related services :-) <jpoiret>unmatched-paren: will we get the answer to the age-old question "why is there #~#$ in code?" <apteryx>jpoiret: I think that's a hack to lower the object in-place <gnucode>apteryx: what services do you think I should offer? <apteryx>I don't know! It depends of the customers you'd be trying to attract, or that have come to you with needs <unmatched-paren>apteryx: what places would it be useful to pass a <GEXP> created like that rather than just, say, a <LOCAL-FILE>? <apteryx>unmatched-paren: I'm not sure, I'd have to see more concrete examples and scratch my head for 30 minutes <gnucode>unmatched-paren: I get that...but man I wish guix had more funding. <unmatched-paren>apteryx: in LIBAIO there's this expression #$(if (%current-target-system) #~#$output #~%output)... please tell me someone has created this abomination by mistake rather than it being something you'd actually have to do :/ <jpoiret>apteryx: I think so too! I was just wondering if this little question we sometimes get would be answered there :) <jpoiret>because lowering the object in place and then inserting it in a g-exp _should_ be a no-op <unmatched-paren>i think, except for #~#$output, the uses of #~#$ are just silly mistakes? not certain though <jpoiret>I think there are corner cases where it made sense last time I looked, but I'm not too sure anymore <jpoiret>apteryx: disregard my previous comment, I was mistaking it with #$#~ :p <apteryx>unmatched-paren: ah, it's to quote the #$output and %output special values <apteryx>otherwise they'd be looked in the global scope when the module is compiled (I think) and fail <unmatched-paren>apteryx: "<unmatched-paren> apteryx: ah, there's the case of #~#$output" :) <jpoiret>cross-building has some peculiarities compared to non-cross-building <jpoiret>it doesn't have the legacy %inputs, %native-inputs, %output and so on I think <jpoiret>that was before the phases had them as arguments iirc <apteryx>mirai: is there a use case to have mcron log to another place than /var/log/mcron.log ? <mirai>apteryx: I have a peculiar case <mirai>but in general it's good to have it "adjustable" <mirai>I haven't finished writing it yet <mirai>but having the log adjustable is necessary for me to make a clone of it <mirai>&& home-mcron-service can be deduplicated as well <apteryx>I was wondering if this was for guix-home <mirai>someone™ should give it a sweep, it's basically 98% copy-paste from mcron.scm :) <apteryx>I'm still haven't migrated to guix home myself; I'm not too keen on having yet another profile to manage, but it's increasingly attractive, thanks to all the contributions it gets :-) <mirai>caveat: anyone thinking on undertaking it, read #61982 & #61809 <sarg>gnucode: idk, but one business opportunity could be consulting for reproducible scientific projects. Reproducibility-as-a-Service <mirai>apteryx: I also haven't given it a try hence me trying to pass that to someone else 😅 <mirai>gnucode: patreon = no-go but perhaps liberapay? <mirai>you could consider “requests” as opportunity <sarg>may I ask to review/merge #62223 ? <jpoiret>sarg: I'll have a look at it, but I'm no committer <sarg>hmm, `src/gopkg.in/yaml.v3/decode_test.go:180:26: constant -9223372036854775808 overflows int`. Fails on i686, but succeeds on x86_64. Why would that be so? <sarg>oh, i686 is 32-bit. It explains <jpoiret>yes, i386, i586, i686, etc... are all the 32-bit versions of x86 <mirai>jpoiret: did you give a spin at the fstrim thing? <mirai>(subtly nudging you for a review) <mirai>it should make your system perform slightly better? :) <Guest3>Hello. Suppose one would want to turn Guix System into a user-friendly, mainstream distribution that even your grandma could use. Do you think that would be possible? What would be potential problems, except that there is no graphical software center for Guix yet? <pkal>Guest3: Why Guix and not some other system like Trisquel? <msavoritias>I would just have gnome or xfce or something with ability to remotely login to the system <msavoritias>pkal: makes it easier for me to manage the issues personally <gnucode>Guest3: I believe the goal is to eventually make guix easier to use. We want to make a graphical frontend to the guix command line for example... <Guest3>pkal: Guix seems like a joy to make a distro out of. Trisquel doesn't have that flexibility of lisp <pkal>No, but that is what makes me wonder what advantage Guix would have to someone who won't notice the difference. <rekado>(hint: Gen X grandmas exist. There is no reason to assume that grandmas have a harder time with Guix than Gen Z tik tok users.) <rekado>general usability problems that stem from the forceful integration of immutability with software designed around mutability need to be addressed. <rekado>the gnome experience, for example, is rougher on Guix System than on any other distro <Guest3>pkal: If I have a clear vision of a distro that people will want to use, then I have a better chance implementing that vision with Guix. The user won't notice the difference, that's true. <rekado>one common problem is due to store file names in cache files in $HOME <msavoritias>Would the distro you are thinking also have home environments managed by guix? <bjc>i'm just grousing here, but man, having to submit patches in the gnu changelog format really makes me not want to submit patches =() <pkal>bjc: are you using Emacs? <rekado>bjc: it’s such a small requirement. <gnucode>I actually just found out how to use the i3 window manager with xfce. :) I think it would be awesome if the xfce-service-type would let you specify alternative window managers. <rekado>bjc: we have two yasnippets to help, and there’s etc/committer.scm too <bjc>it's arduous for someone who doesn't do it all the time. i have to look up the minutae every time, and it's always confusing <gnucode>bjc: there'a an awesome solution for that problem. Just submit more patches more often. :) <bjc>i've never used the snippets. is there a guide somewhere? i know it's mentioned in the handbook, but i don't remember seeing a how-to <Guest3>msavoritia: Well, not completely, but some new defaults should me set, yes <rekado>bjc: (require 'yasnippet) (add-to-list 'yas-snippet-dirs "~/path/to/guix/etc/snippets") (yas-global-mode) <apteryx>bjc: basically just install emacs-yasnippet and accept the .dir-locals.el file at the root of the repo <apteryx>rekado: already covered by .dir-locals.el <rekado>ACTION is older than .dir-locals.el <bjc>ah, i don't have yasnippet installed. i'll do that then <Guest3>rekado: that's a good hint, thank you <apteryx>bjc: then, at the commit message buffer, type 'add' followed by TAB for example <apteryx>or invoke etc/committer.scm as rekado mentioned, to automate it fully for simple changes <rekado>commit message buffer ==> use magit <bjc>the change isn't that simple, but i'll try it <bjc>i am using magit, does it already have hooks for gnu changelog? <apteryx>uh, the progress bar goes backward sometimes while registering items <jpoiret>omg, this is the first time I'm hearing of that magit C thing <jpoiret>You're telling my muscle memory of typing changelog commits is useless? <jpoiret>i've been typing all of them by hand <agnem>I find that I can't remove the gdb-file-system service from %deskop-services by using (modify-services %desktop-services (delete file-system-service-type)), at least as reported by guix system extension-graph. Does the delete form not work when there are multiple instance of a service type (gdb, fontconfig)? <sarg>hmm, `sudo ./pre-inst-env guix system reconfigure ...` fails with `no code for module (gcrypt hash)`. At the same time the same command w/o sudo runs just fine (obviously failing to install symlinks in the end). <jpoiret>file system services are added afterwards using entries in (file-systems ...) i think <jpoiret>sarg: yes, that's because sudo (depending on its configuration) doesn't pass all env variables to the subprocess <jpoiret>you can use --preserve-env=ENV1,ENV2,... to preserve the right environment variables <sarg>let me try -E as I don't know which are the right ones <bjc>have the docs been updated for teams.scm? i haven't submitted anything since that change went in <bjc>and thanks for the ‘C’ magit key, btw. that was very helpful! <bjc>that teams exist at all. =) <bjc>i haven't used the script yet to send a patch, and wanted to read up on it first <bjc>when sending email to a team, should i prefix the subject with the team name? <pkal>bjc: vc-mode can generate the template with a single command (C-c C-w) if you weren't already familiar. Of course it doesn't describe the changes :) <bjc>pkal: do i do that from the diff buffer or the commit buffer? <pkal>It inserts the template at point <bjc>so rarely does my random complaining produce so much useful help. thanks everyone =) <jpoiret>oh my god the C-c commands in the commit buffer are incredible <gnucode>unmatched-paren: sweet! I'll take a look! <f3n1x>i'm having issues with the audio management tool, pavucontrol. Everything is fine... but when in the 'Playback' tab, which is not offering a choice for (loudspeakers, headphones, etc... ) as it use to do. I'm puzzled. What am i missing ? thanks, thanks, thanks <jpoiret>f3n1x: do you have those choices in the output devices tab? <f3n1x>jpoiret: i used to... Not anymore. <jpoiret>what's under the configuration tab for your sound card? <f3n1x>jpoiret: changing the sound card doesn't seem to fix the issue. <f3n1x>i mean changing not the sound card itself but the sound configuration card doesn't seem to fix the issue. <jpoiret>so you have absolutely no output device available? are you using pulseaudio or pipewire? if you don't know then it has to be pulseaudio <jpoiret>can you try doing `pulseaudio -k` first and restarting pavucontrol <agnem>jpoiret: Thanks for the answer. I found the answer to be having to specify (service-kind gdm-file-system-service) instead of file-system-service-type. <jpoiret>what does `pactl list sinks` say? (please use paste.debian.net) <mirai>agnem: sure you want to remove it? <jpoiret>f3n1x: not seeing anything weird, have you tried rebooting? <mirai>what's generating all these headers of the kind: X via Guix-patches via <guix-patches....> <jpoiret>i think it's mailman, it's quite annoying <mirai>it doesn't always happen though <mirai>it seems specific to only some <f3n1x> i'll try to reboot jpoiret ! thanks, thanks thanks for your help <agnem>mirai: I just want to launch my window manager from a simple login shell and don't want gdm launching just to occupy an unused seat. I've removed gdm-service-type as well. <mirai>removing gdm-service-type should remove the filesystem too <mirai>it definitely makes sense for it to be an extension instead <agnem>The gdm-file-system-service is also exported by the desktop module, so I assumed someone was using it without the gdm service <mirai>well, gdm resides in xorg.scm