IRC channel logs

2021-10-11.log

back to list of logs

<hackeryarn>The other day, I sumitted my first patch ever to guix or any gnu project. Was I supposed directly include the maintainers of the specific file I modified, or is just submitting the patch enough?
<hackeryarn>I am just asking because it's a totally new process to me, I've always used github or gitlab in the past. I wanted to make sure I was following the process properly.
<civodul>hackeryarn: hi! just email guix-patches@gnu.org and eventually a committer or seasoned hacker will reply
<nckx>Hi hackeryarn! Guix doesn't have the concept of file/subsystem maintainers like some other projects. Of course some people have expertise, but that's not formalised. If you e-mailed guix-patches@ you did just right ☺
<civodul>ah right :-)
<civodul>see also the check list at https://guix.gnu.org/manual/devel/en/html_node/Submitting-Patches.html
<nckx>☝👍
<singpolyma>Is there a policy about harassing to get patches reviewed? ;) I know everyone is busy, but as a maintainer if other things myself I also know sometimes I just forget/miss things
<nckx>Polite pingage is allowed around the 2-week mark 😉 This can also be pinging someone specific you know might be more familiar with the code, or eager to get it merged.
<nckx>Apart from the 2-weekish number, common sense applies. Don't CC 10 people. Don't, of course, to ruin your joke, harass.
<hackeryarn>Thanks! Glad I didn't miss anything in the manual. The porcess overall is very well described, just wanted to make sure I was getting it in fron of the right people.
<singpolyma>👍
<hackeryarn>Gotta say, I am really liking the patch process compared to the Pull/Merge Request process.
<nckx>(Actually, common sense applies to the ~2 week rule as well: if that would cause an important deadline like a c-u merge to be missed, ping friendly sooner. Just to cover my, er, bases.)
<nckx>hackeryarn: Thanks! It's an acquired taste (so do I). Unfortunately patches don't get reviewed as fast as we'd like. There are people who claim that a GH/GL-like Web 3.0.2 WASM workflow would improve this. Others are sceptical. I'm not sure.
<singpolyma>Changing flow won't give up person hours
<singpolyma>...gin up
***sneek_ is now known as sneek
<nckx>If only.
<KE0VVT>A graphical live system with GNOME would be nice.
<drakonis>sure.
<attila_lendvai>i'd be happy with gnome 4.0. it wasn't fun to return to all the bugs of the v3 line when i installed guix...
<KE0VVT>attila_lendvai: I have enjoyed GNOME 40 so far, and Connections in GNOME 41 looks like it is going to be good.
<attila_lendvai>KE0VVT, you mean on guix? is there some not-yet-merged branch with the gnome 40 packages?
<KE0VVT>attila_lendvai: Oh no, I am on another distro.
<KE0VVT>I’ve become impatient with small distros.
<drakonis>well, gnome 40 is already on core-packages-frozen
<KE0VVT>drakonis: Oh, so it can be used in Guix?
<drakonis>yes
<KE0VVT>Nice!
<drakonis>it is currently up to date with master
<drakonis>so it should be as easy as switching branches right now
<KE0VVT>Oh, okay.
<attila_lendvai>that's good to know!
<drakonis>core-updates-frozen includes a rather large amount of improvements, to boot.
<drakonis>hmm
<drakonis>well okay
<singpolyma>How is it decided what goes in cuf vs master?
<drakonis>its documented
<drakonis>master, staging, core-updates
<drakonis>cuf is supposed to be a frozen version of core-updates right now
<drakonis>in preparation for 1.4.0
<drakonis> https://guix.gnu.org/en/manual/en/guix.html#Submitting-Patches singpolyma
<drakonis>its based on how many rebuilds are done
***tyawp` is now known as tyawp
<raghavgururajan>Hello Guix!
<brendyn>hello
<brendyn>Icecat doesn't seem to support firefox's sync feature for logins etc. anyone know the reason for that?
<civodul>Hello Guix!
<mothacehe>hey civodul!
<xd1le>hi guix hi civodul
<PurpleSym[m]>I pulled today (commit 33de3fe296e63cebe5280f7433853ccf8a4ba790) and I’m getting weird error messages a la “unexpected url from substituter /gnu/store/…”. Anyone else?
*civodul tries "guix time-machine -- describe"
<civodul>PurpleSym[m]:
<civodul>PurpleSym[m]: you mean you upgraded the daemon?
<PurpleSym[m]>civodul: Yes, I ran `guix pull` both as user and root (same commit) and restarted guix-daemon and guix-publish via `systemctl`. Can’t reproduce the issue though, it seems to be ephermal.
<civodul>could you paste the message?
<civodul>(time-machine worked but i didn't upgarde the daemon; i'm at guix-1.3.0-9.f743f20
<civodul>)
<PurpleSym[m]>The exact message got lost in scrollback unfortunately, civodul. Sorry :(
<civodul>np
<ennoausberlin>Hi. I want to build a vm, that includes not only my config.scm, but also my current channels.scm and a given profile build from an existing manifest. Is this possible in an automated way?
<lilyp>guix system purposefully stays away from user configuration, but it shouldn't be too hard to at least get the profile to build inside a time machine
<ennoausberlin>lilyp: I have to read the docs for time machine, but I think that's not the way to go. I want to deploy an application in a reproducible way as vm with the same state as my developer environment. If I run the vm on qemu/boxes it should be in the same state as my dev environment at the time of deploying. The target machine host might not have internet. Sounds weird for an web app, but anyway. So config.scm describes the os plus some always
<ennoausberlin>needed packages like git, vim, emacs. The manifest includes some django packages and python itself and is installed in a certain profile and the channel file includes the repository for the application. Maybe I do not understand fully the purpose of channels, manifests and profiles. How would you handle this use case.
<lilyp>ennoausberlin: I think `guix pack' ought to cover this case. It doesn't copy the OS underneath, but you still get the closure of your environment
<lilyp>most importantly, it can build a docker container, so while not a full-blown vm it should satisfy some of your virtualization requirements
<ennoausberlin>lilyp: Thank you. There is so much to discover. guix pack helps me on machines, where guix is not the host.
<ennoausberlin>lilyp: I am not a big fan of docker, but it is still nice to have this option
<lilyp>tbf tarball is the superior option imo
<wigust>hi guix
***sam__ is now known as Guest5349
<ruffni>wigust: hello
***Dhruvin[m] is now known as dhruvin
<roptat>hi guix!
<nckx>Hi Guix feat. roptat, wigust.
<wigust>nckx: hi
<the_tubular>"(Quit: You have been kicked for being idle)" what ?
<qyliss>I think that happens if they haven't logged into Matrix for a month
<qyliss>so they don't have to maintain the TCP connection
<nckx>Idle TCP is the devil's plaything.
***sam__ is now known as Guest9265
<mekeor>hi guix. how can i inherit a package and add a patches-field to its source-field? O:)
<mekeor>maybe i could write something like `(define-public child (package (inherit parent) (source (origin (origin-of parent)) (patches ...))))' but i don't know if such a `origin-of' function exists.
<podiki[m]>I'm not sure, you can try with just the patches line in source to see if it works, but probably it needs the whole source part copied (manually)
<podiki[m]>at least that's the case in modifying other parts, as I think it just overrides rather than combines if you have the same section in the definition
<podiki[m]>alternatively, if you just want to build/install, you can use a package transformation --with-patch https://guix.gnu.org/manual/devel/en/html_node/Package-Transformation-Options.html#Package-Transformation-Options
<nckx>mekeor: Origins are records, so you can (source (origin (inherit (package-origin parent)) (patches …))
<nckx>Such accessor functions exists for all record types & fields.
<nckx>-s
<nckx>Eh, package-source.
<nckx>The pattern is <record>-<field>, of course. ☺
<podiki[m]>ah nice, good to know!
<podiki[m]>is there anything like that if adding/changing the arguments part?
<pinoaffe>podiki: you can use the function (package-arguments), which should return a list that you can modify
<podiki[m]>ah nice, that makes sense
<podiki[m]>lisp etc
<podiki[m]>I guess I've always been just rushing through a package def to see if it builds and runs....time to learn
<nckx>podiki[m]: Also look at modify-keyword-arguments if you don't want to replace all argument keywords wholesale. It can look a bit intimidating, but c'est la vie.
<nckx>Learning you will do ☺
<detrout>Hi I was trying to get exim running on a guix system, and it's was dying until I created the log directory in /var/spool/exim/. I was wondering if exim's default logs should instead go into /var/log/exim?
<podiki[m]>I still find all things arguments related a bit much :) always copying pasting and adjusting
<podiki[m]>but at this point I think I've done a lot of the main things I see there, so should be getting easier!
<mekeor>nckx: are you sure that (package (source (inherit (package-source)))) should work? i get `inherit: unbound variable`
<nckx>(origin (inherit …
<mekeor>nckx: cool, thank you!
<nckx>YAVW.
*civodul just sent "guix shell v2": https://issues.guix.gnu.org/50960#72
<ngz>Hello. The <https://guix.gnu.org/packages/> page hasn't been updated since last friday.
<singpolyma>civodul: these patches are making me rethink my workflow. I've been using a module to package my dev stuff to make it easier to build something other that the "top" package, but built in support like this may make a non-module version worth the trade-off
<singpolyma>Though maybe I can get both by using a module and importing it into a manifest file