!djTaTBQyWEPRQxrPTb:nixos.org

Nixpkgs Architecture Team

232 Members
https://github.com/nixpkgs-architecture, weekly public meetings on Wednesday 15:00-16:00 UTC at https://meet.jit.si/nixpkgs-architecture53 Servers

Load older messages


SenderMessageTime
30 Jan 2023
@rick:matrix.ciphernetics.nlRick (Mindavi)But maybe that could be explained somewhere if it isn't yet17:11:03
@infinisil:matrix.orginfinisil nbp: It's there, first drawback item :) 17:13:14
@nbp:mozilla.orgnbpWell, I found it useful, to look for the number of games or browser packages.17:14:41
@nbp:mozilla.orgnbp * Well, I found it useful, to look for the number of games or browser packaged.17:14:43
@nbp:mozilla.orgnbphow does repology helps finding packages from similar categories?17:16:29
@piegames:matrix.org@piegames:matrix.orgBtw I think you might need an exception to the path rule for update scripts. Or at least have a closer look at common patterns there and somehow deal with them17:18:23
@infinisil:matrix.orginfinisil piegames: Update scripts can be specified in passthru.updateScript in the package itself, the stable boundary requirement doesn't conflict with that I'm pretty sure 17:20:26
@piegames:matrix.org@piegames:matrix.orgNo the thing is you usually do cd ../../../.. to get to the root for then running nix commands.17:21:33
@infinisil:matrix.orginfinisil piegames: Oh I see 17:23:27
@piegames:matrix.org@piegames:matrix.orgDid you do any test runs already?17:24:40
@sandro:supersandro.deSandro 🐧there is a test PR, that should also be linked as an example17:25:13
@infinisil:matrix.orginfinisil piegames: So that file reference requirement is mainly meant for Nix files, so that evaluation doesn't fail. I shouldn't apply to e.g. update scripts. But yes those paths should definitely be fixed somehow 17:25:20
@infinisil:matrix.orginfinisil * piegames: So that file reference requirement is mainly meant for Nix files, so that evaluation doesn't fail. I shouldn't apply to e.g. update scripts. But yes those paths should definitely be updated somehow 17:25:30
@sandro:supersandro.deSandro 🐧
In reply to @piegames:matrix.org
No the thing is you usually do cd ../../../.. to get to the root for then running nix commands.
TBH we should abstract most of those workarounds away and get away from writing those scripts from zero.
17:25:46
@sandro:supersandro.deSandro 🐧so a bit in the direction what update-source-version is doing17:26:19
@qyliss:fairydust.spaceAlyssa Ross unstableGitUpdater <3 17:26:26
@infinisil:matrix.orginfinisilYeah while working on this I also discovered that most of these file references are actually just anti-patterns and it's often better to get rid of them than to work around them17:27:14
@sandro:supersandro.deSandro 🐧but that also blows up the scope massively 17:29:59
@infinisil:matrix.orginfinisilYes, so the tooling detects such cases where moving a packages files around would break eval, and just doesn't move those packages (it could give a warning though)17:31:45
@sandro:supersandro.deSandro 🐧moving update scripts does not break eval but breaks at execution17:36:21
@sandro:supersandro.deSandro 🐧most of those use either cd in code or nix-shell with relative paths in the shebang17:36:33
@infinisil:matrix.orginfinisilNot sure yet what to do about that, it seems really tricky to accurately detect this in shell scripts17:37:18
@infinisil:matrix.orginfinisil And there's no clear API for these update scripts, they might not even be in passthru.updateScript, they might just be random files that expect to be executed 🤔 17:39:18
@infinisil:matrix.orginfinisil After a quick grep for ../, it looks like there's relatively few instances, these could probably be fixed semi-manually 17:40:04
@sandro:supersandro.deSandro 🐧take https://github.com/NixOS/nixpkgs/blob/master/pkgs/applications/misc/ArchiSteamFarm/update.sh as a bad example17:40:05
@sandro:supersandro.deSandro 🐧
In reply to @infinisil:matrix.org
And there's no clear API for these update scripts, they might not even be in passthru.updateScript, they might just be random files that expect to be executed 🤔
we should enforce putting them there otherwise CWD can be anything and you usually have some lines of glue code around that
17:41:07
@infinisil:matrix.orginfinisilThat sounds good, but probably better to keep out of this RFC to not blow up the scope17:42:32
@whentze:matrix.orgWanja Hentze
In reply to @sandro:supersandro.de
TBH we should abstract most of those workarounds away and get away from writing those scripts from zero.
I'm not going to say the F word but this looks like a problem that might be solvable using a certain experimental nix feature, maybe
17:46:03
@whentze:matrix.orgWanja Hentzeor maybe that certain feature needs to accommodate arbitrary updater scripts17:46:41
@whentze:matrix.orgWanja Hentze
In reply to @infinisil:matrix.org
We just opened the RFC https://github.com/NixOS/rfcs/pull/140 🎉
also, wtf, now I got the animation when switching to this tab
18:00:58

Show newer messages


Back to Room ListRoom Version: 9