| 16 May 2024 |
 @jade_:matrix.org | fuck yeah | 20:06:04 |
 Lily Foster | In reply to @qyriad:katesiria.org
that's finalAttrs.finalPackage isn't it? yep that's the one | 20:06:11 |
| Lily Foster | isn't it all fun | 20:06:22 |
| Lily Foster | (^ the message above this one is sarcastic in case it was not clear) | 20:06:40 |
 Qyriad | Because technically finalAttrs is only the the arguments to mkDerivation and not the result of that 🙃 | 20:07:07 |
| @jade_:matrix.org | vringar: i hope this makes at least some sense, feel free to ask questions if "girls go read the stdenv source code in real time" wasn't clear enough on what to actually do to fix it XD | 20:07:20 |
| Lily Foster | does mkDerivation still have a TODO comment to change the forcing-doCheck-off-on-all-cross behavior that was added like 5 years ago and not touched since? | 20:07:45 |
| Qyriad | # TODO(@oxij, @Ericson2314): This is here to keep the old semantics, remove when
# no package has `doCheck = true`.
doCheck' = doCheck && stdenv.buildPlatform.canExecute stdenv.hostPlatform;
| 20:08:13 |
| Qyriad | Yep. | 20:08:26 |
| Lily Foster | lovely | 20:08:37 |
| Lily Foster | (oof i've spent too much time in stdenv at this point. it's highly cursed knowledge. i do not recommend) | 20:09:10 |
| Qyriad | on the flip side, cursed stdenv knowledge is basically required to understand Nixpkgs | 20:09:30 |
| Qyriad | (which sucks!) | 20:09:34 |
 vringar | In reply to @jade_:matrix.org
vringar: i hope this makes at least some sense, feel free to ask questions if "girls go read the stdenv source code in real time" wasn't clear enough on what to actually do to fix it XD Currently side tracked by having to entertain cats. Will catch up eventually | 20:09:45 |
| @jade_:matrix.org | strongly approved activity | 20:10:06 |
| @jade_:matrix.org | i should probably write a short blogpost about this kind of broken recursion structure. i wonder if it is detectable... | 20:10:33 |
| @jade_:matrix.org | we should lint on it | 20:10:48 |
| Lily Foster | In reply to @qyriad:katesiria.org
on the flip side, cursed stdenv knowledge is basically required to understand Nixpkgs yeah like when someone unknowingly breaks splicing because they don't know it exists because it's supposed to be "magic" and then I have to fix a ton of call and override points to make it not do that when they finally want cross to work | 20:11:05 |
| Lily Foster | (yes hi, i recently helped someone fix the splicing disaster for dart/flutter) | 20:11:52 |
| @jade_:matrix.org | i think hmmmmmm. technically yes. it should be lintable | 20:12:09 |
| Qyriad | In reply to@jade_:matrix.org
we should lint on it that might not actually be that hard. (like, aside from the effort needed to add a linting system to Lix in general, which we already want) | 20:12:14 |
| @jade_:matrix.org | you would look for anything in passthru that has an arg == to the parent derivation and that == relation is broken if you overrideAttrs the parent | 20:12:52 |
| Qyriad | yeah | 20:13:20 |
| @jade_:matrix.org | In reply to @qyriad:katesiria.org
that might not actually be that hard. (like, aside from the effort needed to add a linting system to Lix in general, which we already want) not convinced it should be in lix, but nixpkgs does have some linty things especially for meta | 20:13:34 |
| @jade_:matrix.org | i would like linting, i am just not sure if the rules should be in lix | 20:13:43 |
| Qyriad | We actually think they should be, or at least some of them should be. Rustc has lints specific to std APIs; I think it's fine for Lix to do so too, if there's a good reason it can't or shouldn't be in Nixpkgs first | 20:15:32 |
| Qyriad | Or in like, a Lix-official plugin | 20:16:01 |
| Qyriad | If it can be in Nixpkgs though that's almost definitely better | 20:21:52 |
| @jade_:matrix.org | yeah the only real issue with this is that it would be possibly expensive eval time wise since it would create more derivations but it could probably be done to avoid that too | 20:23:16 |
| @jade_:matrix.org | (where you would like, just put in an attr and check its still there but not eval the outPath) | 20:23:35 |
