| 19 Apr 2025 |
 maralorn | Soo imo we are ready. | 10:56:06 |
| maralorn | alexfmpe: Do you understand the eval error for miso-action-logger? https://hydra.nixos.org/jobset/nixpkgs/haskell-updates#tabs-errors | 11:08:12 |
| maralorn | I fixed the other eval error … | 11:09:28 |
 Alyssa Ross | In reply to @sternenseemann:systemli.org
maralorn: looks like HLS/hlint are the last big things from the report. Musl/Static stuff can be ignored and needs to be reevaluated in staging(-next). thielema package failures look weird, maybe worth a bug report upstream. large-records is not fixable for the moment. Got a link to musl/static stuff? If it's something that I need I'd like to get on it proactively. :) | 11:31:46 |
| maralorn | You can have a look at all the build errors here which mention static or musl: https://github.com/cdepillabout/nix-haskell-updates-status | 11:36:10 |
| Alyssa Ross | Oh pkgsMusl.haskell.compiler sounds important. | 11:37:04 |
| Alyssa Ross | Although maybe I don't understand what that is, because it doesn't seem to be required to build pkgsMusl.pandoc… | 11:41:24 |
| Alyssa Ross | Can't reproduce the re2c build failure that causes it anyway so I guess it's transient. | 11:43:44 |
 sterni (he/him) | wolfgang walther claims it’s fixed on staging | 11:57:26 |
 alexfmpe | nope, and I can't reproduce either | 16:14:56 |
| alexfmpe | package is unchanged on hackage since 2019 | 16:15:07 |
| alexfmpe | maybe something else was borked? or bit flip I dunno | 16:15:25 |
| maralorn | My intuition was that we added some new hydra job which creates an invalid combination. | 16:19:22 |
| alexfmpe | oh wait a sec | 16:20:50 |
| alexfmpe | that's the native build | 16:20:57 |
| alexfmpe | hostPlatform.config = "x86_64-unknown-linux-gnu"
package.meta.platforms = [
"javascript-ghcjs"
]
package.meta.badPlatforms = [ ]
| 16:21:58 |
| alexfmpe | shouldn't it, like, not try to eval? | 16:22:30 |
| maralorn | Well the problem is that it has a dependency which doesn't eval. | 16:26:10 |
| maralorn | So the correct solution seems to be to restrict the plattforms of miso-action-logger if that's factually correct. | 16:27:09 |
| alexfmpe | but they are? | 16:27:32 |
| alexfmpe | the error message says so | 16:27:43 |
| alexfmpe | building locally I also get
a) To temporarily allow packages that are unsupported for this system, you can use an environment variable
for a single invocation of the nix tools.
$ export NIXPKGS_ALLOW_UNSUPPORTED_SYSTEM=1
| 16:28:04 |
| maralorn | In reply to @alexfmpe:matrix.org
but they are? No, the error message is showing a part of the ghcjs-base derivation. The miso-action-logger derivation has no restrictions. | 16:31:32 |
| alexfmpe | oooh | 16:31:44 |
| maralorn | I wonder if we have to null ghcjs-base or something... | 16:32:22 |
| alexfmpe | I'll try a few things | 16:32:50 |
| alexfmpe | IIRC miso itself didn't build for native in nixpkgs until recently, because cabal would find no libraries or anything after going through the conditionals | 16:33:33 |
| alexfmpe | it just built for me now though, technically was cached | 16:33:53 |
| alexfmpe | think that was changed by the 1.8.7 release | 16:34:13 |
| alexfmpe | not enough because the package always unconditionally declares the dependency and imports from it | 16:37:43 |
