gchristensen changed the topic of #nixos-borg to: https://www.patreon.com/ofborg https://monitoring.nix.ci/dashboard/db/ofborg?refresh=10s&orgId=1&from=now-1h&to=now "I get to skip reviewing the PHP code and just wait until it is rewritten in something sane, like POSIX shell. || https://logs.nix.samueldr.com/nixos-borg
cransom has quit [Quit: WeeChat 2.1]
<gchristensen> what if
<gchristensen> to get the maintainers of changed packages
<gchristensen> I get the list of attributes changed
<gchristensen> use unsafeGetAttrPos or whatever to get the file it came from
<gchristensen> and see if that file is in the diff
<gchristensen> I'm a few beers in so maybe that isn't a good idea
<samueldr> that's because mapping changedFile to changedAttribute is harder, right?
<gchristensen> yeah
<gchristensen> I think?
<gchristensen> I have no idea how to do that
<samueldr> you don't want to ping the whole world on a coreutils change
<gchristensen> exactly
<gchristensen> !!!
<samueldr> ⁉️
<gchristensen> I should write nix code a few beers in more often apparently
<samueldr> your liver and bosses may not approve ;)
<gchristensen> ah ha!
<{^_^}> #45195 (by Ralith, 2 weeks ago, merged): vulkan: 1.1.77.0 -> 1.1.82.1
<samueldr> madness
<gchristensen> guh I think it could work
<samueldr> it probably can!
<samueldr> try it with fun PRs :)
<gchristensen> links to examples?
<samueldr> ideally, you want to have false negatives, and not false positives (miss maintainers than ping wrong ones)
<gchristensen> yeah definitely
<samueldr> none! I thought you would know
<gchristensen> hehe
<gchristensen> I'll try it on some.
<gchristensen> maybe "dark"-launch it
<gchristensen> this is too powerful.
<samueldr> I'm not sure I like your scheming tone... "dark", "too powerful"
<samueldr> are you turning into a villain?
<gchristensen> turning?
<samueldr> you know what happens to all(?) villains?
<samueldr> they do a dumb mistake that an bumbling idiot "hero" profits of to turn the tides ;)
<gchristensen> :)
<gchristensen> samueldr: want to try it?
<samueldr> defeating you?
<gchristensen> lol
<gchristensen> samueldr: want to try the script?
<samueldr> yeah, if you have something ready
<gchristensen> save this at ./nixpkgs/pr-maintainers.sh
<gchristensen> chmod +x ...
<samueldr> that .sh is... sneaky
<gchristensen> ./pr-maintainers.sh 45912
<gchristensen> it really needs to be run in a checkout matching the PR so it'll sometimes fail, but ... :)
<samueldr> I may integrate something like that in my review script :)
<samueldr> (I already diff attributes, so I don't even need to check github)
<gchristensen> well
<gchristensen> sure
<gchristensen> I was thinking I might add it to ofborg this weekend :)
<samueldr> https://github.com/NixOS/nixpkgs/pull/43680 repeats maintainers because of the way mono is multiple attributes
<{^_^}> #43680 (by leo60228, 6 weeks ago, open): msbuild: init at 15.6
<samueldr> (just in case you wanted to know)
<gchristensen> I'm getting a lot of assertion errors :( + nix-instantiate --eval ./data.nix
<gchristensen> error: Linux v4.18.2018.08.31 is not yet supported by zfsonlinux v0.7.9.
<gchristensen> Try zfsUnstable or set the NixOS option boot.zfs.enableUnstable.
<samueldr> (I'm not on their head, I'm on their head merged into nixpkgs-unstable though)
<gchristensen> I think this + outpaths.nix in ofborg/ofborg/src could work
<gchristensen> anyway, long past bed time.
<gchristensen> g'night :)
<{^_^}> Night!
<gchristensen> lol <3
cransom has joined #nixos-borg
orivej has joined #nixos-borg
<LnL> gchristensen: there are a few edgecases that wouldn't be detected
<LnL> nix edit uses meta.destription with a fallback to name to find stuff, but if there's a generic.nix with variants there might only be changes in ./x.y.nix
jtojnar has quit [Quit: jtojnar]
jtojnar has joined #nixos-borg
<gchristensen> LnL: true. is there a way to "fix" those?
<LnL> not that I can think of
<LnL> but those are not false positives, so might be ok
<gchristensen> maybe, unfortunately, duplicating the maintainer fields out to the generic
<LnL> yeah, duplicating meta is a relatively sane workaround
<LnL> and that also fixe nix edit
<gchristensen> yeah
<gchristensen> when trying to determine maintainers for https://github.com/NixOS/nixpkgs/pull/45468a
<gchristensen> when trying to determine maintainers for https://github.com/NixOS/nixpkgs/pull/45468
<{^_^}> #45468 (by dtzWill, 1 week ago, merged): boost168: init
<{^_^}> #45468 (by dtzWill, 1 week ago, merged): boost168: init
<LnL> yeah, but that also changed generic.nix
<gchristensen> before my diff it didn't find any maintainers
<gchristensen> hmm
<gchristensen> oh, it didn't modify generic.nix
<gchristensen> pkgs/development/libraries/boost/1.68.nix and pkgs/top-level/all-packages.nix
<LnL> oh, that's not the diff from the pr
<gchristensen> yeah, if we changed boost's definition to be like that, then the PR would have been correctly detected
<LnL> does your thing follow maintainers, or description like nix?
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos-borg
<gchristensen> maintainers
<LnL> I think I might have simplified this problem a bit in my head
<LnL> doesn't moving meta break the reverse?
<LnL> changing only generic should also ping maintainers
orivej has quit [Ping timeout: 245 seconds]
jtojnar has quit [Remote host closed the connection]
jtojnar has joined #nixos-borg
WilliButz has quit [Ping timeout: 272 seconds]
WilliButz has joined #nixos-borg
<gchristensen> LnL: heh, true
<gchristensen> LnL: maybe I should look at many fields, version, pname, src, meta
<LnL> oh, hmm
<LnL> yeah, that would work for this usecase
<gchristensen> ok VERY weird thing
<{^_^}> #45635 (by matthewbauer, 1 week ago, open): mesa: remove darwin-specific derivations
<gchristensen> oh it was deleted a few days ago
<LnL> cbdcf6b4db2e94438fcb54a7394fa31e31502faf
<LnL> git log -G is pretty useful for this kind of stuff
<gchristensen> -G?
<gchristensen> nice
<gchristensen> I use -S sually
<LnL> ah, it's basically the same
<gchristensen> I really really really want an indexed database of all of a git log
<gchristensen> like Hound, but history-aware
<LnL> that would be __amazing__
<LnL> the fact that hound only indexes on branch has also bitten me a few times
<samueldr> even more so if it's in a workable format to somehow link it with more data (e.g. local cache of github PRs)
<gchristensen> is github ... down?
<gchristensen> like I can't even connect, and when I do I get 500's
<samueldr> seems up here
<LnL> fine here
<gchristensen> ok. weird.
<gchristensen> I guess it is time to go to the store then :)
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 240 seconds]
cransom has quit [Ping timeout: 252 seconds]
cransom has joined #nixos-borg
cransom has quit [Quit: WeeChat 2.1]
cransom has joined #nixos-borg
timokau[m] has joined #nixos-borg
<gchristensen> damn
<gchristensen> I can't look at the `name` attribute because it is always in /home/grahamc/projects/nixpkgs/pkgs/stdenv/generic/make-derivation.nix
<gchristensen> # A hack to make `nix-env -qa` and `nix search` ignore broken packages.
<gchristensen> # TODO(@oxij): remove this assert when something like NixOS/nix#1771 gets merged into nix.
<{^_^}> https://github.com/NixOS/nix/pull/1771 (by oxij, 35 weeks ago, open): nix-env.cc: make `--query` evaluate `.outPath`s
<gchristensen> name = assert validity.handled; name + lib.optionalString
<LnL> yeah, that's why edit tries the description first
<gchristensen> ah
<gchristensen> I think it works really well
<gchristensen> it is super ugly as-is but only because I wanted it to be easy to run and share as a shell script while I hacked on it :)
<gchristensen> shockingly it correctly detected a mass-rebuild (bison) had no maintainer
<gchristensen> so ... I think that is good ... :D
<LnL> yeah, overlaying meta.maintainers with the diff should be a pretty good heuristic
<gchristensen> I'm scared of some scary condition where I ping 200 people
<LnL> btw. I wonder if wouldn't be better to put this and the rebuild expression in nixpkgs
<LnL> what ofborg uses now is kind of hidden
<gchristensen> yeah
<gchristensen> the reason I didn't is I don't trust the PR's nixpkgs... but maybe I could just point to a trusted revision.
<LnL> oh yeah, we can keep it duplicated in the repo
<LnL> or use the one from the deploy nixpkgs or something
<gchristensen> yeah
<gchristensen> fetchurl :)