pmy_ has quit [Read error: Connection reset by peer]
<Rovanion>
Not sure if this qualifies as development of NixOS, but: I'm trying to package a Go module that is failing during the check phase. Using `nix-build $NIXPKGS --run-env -A prometheus-slurm-exporter` I can get an environment populated with the functions that belong to a generic build. Is it possible to get shell/environment populated with the functions/phases that are used during a buildGoModule
<Rovanion>
build?
<symphorien[m]>
replace nix-build by nix-shell and run `genericBuild`
<symphorien[m]>
you should do that in a temporary directory
<symphorien[m]>
it will fail from installPhase on, because the store is read only
pmy_ has joined #nixos-dev
ili has joined #nixos-dev
pmy_ has quit [Read error: Connection reset by peer]
ili has left #nixos-dev ["User left"]
ili has joined #nixos-dev
kini has quit [Remote host closed the connection]
pmy_ has joined #nixos-dev
kini has joined #nixos-dev
pmy_ has quit [Read error: Connection reset by peer]
pmy_ has joined #nixos-dev
pmy_ has quit [Read error: Connection reset by peer]
pmy_ has joined #nixos-dev
pmy_ has quit [Read error: Connection reset by peer]
pmy_ has joined #nixos-dev
betawaffle has quit [Ping timeout: 240 seconds]
feepo has quit [Ping timeout: 265 seconds]
mdlayher has quit [Read error: Connection reset by peer]
rajivr has quit [Ping timeout: 260 seconds]
sdier has quit [Read error: Connection reset by peer]
prusnak has quit [Read error: Connection reset by peer]
zhaofeng has quit [Read error: Connection reset by peer]
s1341 has quit [Read error: Connection reset by peer]
alunduil has quit [Read error: Connection reset by peer]
georgyo has quit [Read error: Connection reset by peer]
grfn has quit [Read error: Connection reset by peer]
sorear has quit [Read error: Connection reset by peer]
nh2 has quit [Read error: Connection reset by peer]
mdlayher has joined #nixos-dev
vdemeester has joined #nixos-dev
cstrahan has joined #nixos-dev
sorear has joined #nixos-dev
betawaffle has joined #nixos-dev
chrisaw has joined #nixos-dev
s1341 has joined #nixos-dev
grfn has joined #nixos-dev
dmj` has joined #nixos-dev
prusnak has joined #nixos-dev
alunduil has joined #nixos-dev
davidtwco_ has joined #nixos-dev
nh2 has joined #nixos-dev
emilazy has joined #nixos-dev
sdier has joined #nixos-dev
georgyo has joined #nixos-dev
rajivr has joined #nixos-dev
zhaofeng has joined #nixos-dev
thoughtpolice has joined #nixos-dev
feepo has joined #nixos-dev
<yorick>
(funfact: store is not read-only when running nix-shell as root)
<qyliss>
oh fun
midchildan has joined #nixos-dev
raboof has joined #nixos-dev
<Rovanion>
Thanks. Tried it and both `nix-build --run-env` and `nix-shell --run-env` seem to result in the same environment/shell. A thing I missed is that calling genericBuild works (up until the test failure) but calling checkPhase does not work, complaining that there is no makefile or custom checkPhase.
<Rovanion>
In reality I just want to know the output of the test phase :/
<gchristensen>
when it goes down like that it is usually because someone changed the primary branch or renamed a repo or something, and I just have to re-sync
<yorick>
adisbladis: I am
<adisbladis>
I had houn set up on my desktop before
<adisbladis>
It was pretty sweet
<adisbladis>
But now I'd probably use sourcegraph
kini has quit [Remote host closed the connection]
kini has joined #nixos-dev
Cale has joined #nixos-dev
jonringer has joined #nixos-dev
<lukegb>
sourcegraph is alright but it seems to dislike indexing all-packages.nix
* lukegb
can't possibly imagine why 👀
<gchristensen>
big files are so often the most important ones
<adisbladis>
At some point I'd really like to get rid of all-packages.nix
<gchristensen>
I like that there is an explicit map of package names to imports, however that is
<gchristensen>
and readdir doesn't even promise to return all the dirents eyeroll.avi
<adisbladis>
gchristensen: Sure, I think reorganising it into subtrees is preferable
<adisbladis>
And then have some merge mechanics for the subtrees
<adisbladis>
Recursive readdirs sounds silly and expensive
<gchristensen>
yeah
<gchristensen>
I wouldn't mind an automatically generated all-packages.nix
<adisbladis>
Except you'd still have merge conflicts if it was auto-generated
<gchristensen>
potentially yea
<adisbladis>
This is one of the bits I'd like to solve if we were to split it
<__monty__>
Rovanion: You can only run checkPhase if you go through all the phases before it first. Otherwise the environment isn't set up like checkPhase expects.
rajivr has quit [Quit: Connection closed for inactivity]
cole-h has joined #nixos-dev
saschagrunert has quit [Remote host closed the connection]
<infinisil>
Damnit, adding event notifications to hydra for started and finished evaluations isn't as easy as I'd hoped
<infinisil>
This checkJobsetWrapped function is weird. It sometimes `die`'s on errors, but sometimes it doesn't. And in some cases it calls `notify eval_*` but in others it doesn't (that's the events I'd want)
<infinisil>
And there's different events which are documented nowhere
<infinisil>
And I'm not sure if all branches of that function are covered with such a notification (either _failed, _cached or _added should indicate it being finished)
<worldofpeace>
Is `hash` and sri hashes good for backport to 20.09?
<worldofpeace>
atm my position is just to keep `hash`, and tbh I would want to go as far is to merge fetchgit with SRI hash support... that was august and if no one responds that's just their loss? an unpleasant truth but yeah.
<jtojnar>
yeah, I would do that as well, but I do not want to go against Eelco’s wishes
<worldofpeace>
same, but if eelco doesn't have a time and a place to present them then... idk have u pm'd about it before?
<jtojnar>
no, did not want to bother him
<gchristensen>
I heard from him w.r.t. the hydra work I was doing that he's extra committed these next few weeks on a client
<piegames>
supersandro2000: Some improvements for your review thingy:
<piegames>
Don't post a review comment if somebody else already has (for the same architecture) and there have been no changes inbetween. (Had two examples for this today)
<piegames>
Maybe only post a comment when things fail if there has been a review for *any* architecture? Most of the time, if one works, we expect all to work. Thus, there's only need to mention it when it doesn't.
orivej has quit [Ping timeout: 265 seconds]
feepo has quit [Ping timeout: 246 seconds]
grfn has quit [Read error: Connection reset by peer]
grfn has joined #nixos-dev
feepo has joined #nixos-dev
<eyJhb>
Not sure if this is a dev thing, but anyone (especially etu), is it possible to get a more minimal install of PHP, currently I have this (did the withExtensions to ensure nothing was there) - https://termbin.com/3ak8 and that will end up with a little less than 200 MB image. I can see all this getting pulled in - https://termbin.com/zzbw (systemd, etc.)
<eyJhb>
Is that a PHP or a Docker thing? Because without PHP I get a very very small image
<gchristensen>
you might use nix why-depends /nix/store/dd9kbm3f59mdrcmlsvrb423yicf8z65r-php-7.4.14 /nix/store/nwpwnz9ghydmqiic0psjc8gl4jlw9fmc-systemd-247.2
<eyJhb>
I think I found out, that I can disable a lot of flags in the php/generic.php
<eyJhb>
I think I found out, that I can disable a lot of flags in the php/generic.nix**
<eyJhb>
But eh. I am not sure how to override such flags. I so rarely use it
<eyJhb>
I rarely use all these things, I forget how easy some things are in Nix...
<supersandro2000>
piegames: you are good with graphql by chances?
<supersandro2000>
I actually check if ofborg build the packages but incase of python it does not run all 3 variants
<gchristensen>
oh?
<supersandro2000>
🤔 we could easiyl do that if the commit was included in all nixpkg-reviews as a comment
<supersandro2000>
I wanted to do that at some point but forgot doing it
<gchristensen>
what do you mean about the python bit?
<supersandro2000>
but until then nothing I can do because the comments can be from whenever and posted whenever and posting the results from a self run is not that helpful because it won't check the hashes
<supersandro2000>
python packages have one variant for each python version and I currently do not check that
<gchristensen>
ah
<supersandro2000>
just when the result is a single package
<supersandro2000>
if github provided private notes I could post to I would never post them except if there is something to do
<supersandro2000>
but they don't
<qyliss>
can't you just... keep your own notes?
<qyliss>
why would GitHub have to provide them?
<supersandro2000>
that would be notes for a few thousand PRs and I am just lazy
<qyliss>
... wow
<eyJhb>
Why not just use sqlite?
<eyJhb>
Simple effective
<eyJhb>
( gchristensen without systemd it went from 189 MB to 85 MB )
<supersandro2000>
I mean I could search for other nixpkgs-review like comments but I have a different syntax than rmcgibbo and the standard one
<gchristensen>
nice, eyJhb
<supersandro2000>
eyJhb: because if I don't have it in one place it is just inconvenient
<gchristensen>
I feel it is important to value the impact on other contributors higher than the inconenience to me when I'm doing something
<supersandro2000>
if we had the data in machine parsable format than I could easily do it
<samueldr>
<details><summary></summary><pre></pre></details> <- you could embed json data here
<qyliss>
gchristensen++
<{^_^}>
gchristensen's karma got increased to 422
<samueldr>
(but that wouldn't help if it's not posted)
<qyliss>
could edit the PR body
<samueldr>
that would, annoyingly, make the tool less useful for people that are not committers yet :/
<supersandro2000>
gchristensen: I really know some people who should think about that way more often than me
<gchristensen>
PMing.
<supersandro2000>
well I think about not posting by default
__monty__ has quit [Quit: leaving]
<jtojnar>
some out of band channel that could store review metadata without clobbering the comments would be useful