| 19 Feb 2026 |
 alexfmpe | haskell-updates started branching from staging as of last year or so | 23:41:47 |
| alexfmpe | so if you pull at a bad time, you can eat nasty rebuilds | 23:42:12 |
| alexfmpe | e.g. after staging was merged into it but before hydra CI cached that | 23:42:31 |
 woobilicious | I guess I can just checkout master and the cherry-pick the patch? or maybe just let the scripts be called by the next merge? | 23:43:41 |
| alexfmpe | that said, haskell-updates is mostly for bumping LTS and dealing with the fallout | 23:43:46 |
| woobilicious | I think I'm meant to call ./maintainers/scripts/haskell/regenerate-hackage-packages.sh instead shrug | 23:44:02 |
| alexfmpe | so if your sbv patches fix it on master, you can probably target that | 23:44:18 |
| alexfmpe | * so if your sbv patches fixes aren't specific to latest versions you can probably target that | 23:44:40 |
| alexfmpe | * so if your sbv patches fixes aren't specific to latest versions you can probably target master | 23:44:45 |
| woobilicious | ahh okay, I'm just disabling tests. | 23:45:12 |
| alexfmpe | huh? | 23:45:13 |
| alexfmpe | you shouldn't need to call that unless you mess with the .yaml files or so | 23:45:52 |
| woobilicious | yeah I unmarked it as broken | 23:46:11 |
| alexfmpe | ah ok relaxing brokenness to tests | 23:46:34 |
| alexfmpe | * ah ok relaxing broken to broken tests | 23:46:44 |
| woobilicious | upstream says it's probably unrealistic for us to test it, they also take like 45mins to run | 23:47:55 |
| woobilicious | i.e. https://hydra.nixos.org/log/5hsi41vkklihxa78a0lmj49za0d9mifn-sbv-10.2.drv | 23:48:12 |
| alexfmpe | wth | 23:51:23 |
| woobilicious | yeah actually we don't even package mathsat, so tests can't pass without us doing that. | 23:54:17 |
| alexfmpe | can't they...like...have two test suites that both call the same thing but with different quickcheck iterations | 23:54:41 |
| alexfmpe | * can't they...like...have two test suites that both call the same thing but with different number of quickcheck iterations | 23:55:02 |
| alexfmpe | then their CI can run the fancy 45min thing, but distributors have a better choice than 45min or no tests | 23:55:29 |
| alexfmpe | * then their CI can run the fancy 45min thing, but distributors have more than the choice than 45min or no tests | 23:55:44 |
| alexfmpe | * then their CI can run the fancy 45min thing, but distributors have more than the choice between 45min or no tests | 23:55:50 |
| woobilicious | wtf why did regenerate-hackage-packages.sh change quotation in a script? | 23:56:17 |
| alexfmpe | quotation? | 23:56:33 |
| alexfmpe | there's something funny going on with regenerate-hackage-packages.sh where you're supposed to call it inside the top-level nix-shell so it will find treefmt and lint away | 23:57:03 |
| woobilicious | - preBuild = "export LD_LIBRARY_PATH=`pwd`/dist/build\${LD_LIBRARY_PATH:+:}$LD_LIBRARY_PATH";
+ preBuild = ''export LD_LIBRARY_PATH=`pwd`/dist/build''${LD_LIBRARY_PATH:+:}$LD_LIBRARY_PATH'';
| 23:57:06 |
| alexfmpe | ah there you go, it's the auto-linter | 23:57:18 |
| alexfmpe | nixpkgs CI has lint checks for months now | 23:57:35 |
