| 19 Apr 2025 |
 alexfmpe | someone put up a PR for this on large-records just 2 hours ago, but still needs sanity check so I'm not leaning towards even applying as patch until it's reviewed
I don't think we should block merge to staging on it, can just be PR'd directly to staging if upstream rubber stamps without an immediate release? | 16:45:09 |
| alexfmpe |
https://github.com/well-typed/large-records/pull/171 | 16:45:16 |
| alexfmpe | https://github.com/NixOS/nixpkgs/pull/400121 | 16:54:25 |
| alexfmpe | hey we broke 400k | 16:54:29 |
 sterni (he/him) | agree & agree & agree | 18:40:27 |
| alexfmpe changed their profile picture. | 23:55:28 |
| 20 Apr 2025 |
 maralorn | In reply to @emilazy:matrix.org
the 23rd is the nominal breaking change deadline and I assume there are breaking changes here If no one else does, I will finalize our merge on Tuesday. | 23:03:04 |
| 21 Apr 2025 |
|  @megmug:matrix.org left the room. | 11:25:41 |
| 22 Apr 2025 |
| sterni (he/him) | I'm currently finishing the broken list and am going to merge after that. | 08:52:40 |
| sterni (he/him) | Any further fixes can probably relatively easily be done on staging(-next) | 08:52:56 |
| sterni (he/him) | * I'm currently finishing the broken list and am going to merge after that. Going to take a minute since I manually need to check the queued jobs. | 08:53:20 |
 hellwolf | hurray!
I would invite everyone to have a second look at the diff of 1909d9a. We can still fix packages that are worth fixing on staging (within reason, i.e. another hackage bump on there is rather difficult).
pkgs/development/haskell-modules/configuration-hackage2nix/broken.yaml: "+450-19"
Am I reading right that there are 400+ more borken packages because of the 9.8?
| 10:15:10 |
| hellwolf | https://github.com/cdepillabout/nix-haskell-updates-status
"Top 50 broken packages, sorted by number of reverse dependencies"
Is this the place to start to look for the culprits? | 10:17:02 |
 @malteneuss:matrix.org | Manually applying patches and fixing builds for weeks doesn't seem sustainable. If you have the time and haven't done so, please voice your opinion in Discourse on what has to change in GHC and/or Haskell ecosystem upstream to reduce the churn, e.g.
https://discourse.haskell.org/t/language-library-and-compiler-stability-moved-from-ghc-9-6-migration-guide/5745/92 (GHC stability/backwards compatibility)
https://discourse.haskell.org/t/introducing-the-haskell-foundation-stability-working-group/11743 (GHC stability/backwards compatibility)
https://discourse.haskell.org/t/how-much-effort-does-backwards-compatibility-require-from-library-authors/11584 (Haskell ecosystem stability/backwards compatibility)
https://discourse.haskell.org/t/request-for-comment-cabal-freeze-doesnt-produce-a-lock-file/11374 (Real Cabal lock files like e.g. Rust) | 11:00:23 |
| maralorn | Yeah, but I wouldn’t worry too much about it. e.g. quite a few of them are likely just packages newly released in the last months others might haven been disabled before because of one broken dep. | 11:00:45 |
| maralorn | malteneuss: Would you want us to use a completely different approach or do you just want to nudge the ecosystem so that we need to apply less patches and fix less builds? | 11:12:37 |
| maralorn | * Yeah, but I wouldn’t worry too much about it. e.g. quite a few of them are likely just packages newly released in the last months others might haven been disabled before because of one broken dep which got fixed. | 11:12:45 |
| maralorn | * Yeah, but I wouldn’t worry too much about it. e.g. quite a few of them are likely just packages newly released in the last months others might haven been disabled before because of one broken dep which got fixed. Yet, some others are newly broken, but if they are unmaintained probably no one cares. | 11:13:21 |
| @malteneuss:matrix.org | My gut feeling is that we need both. | 11:13:35 |
| maralorn | My gut feeling is that we are actually delivering an astonishingly large amount of working packages compared to the amount of work we are investing. While we can certainly decrease the friction by removing some small or big paper cuts, I can’t think of a different approach which could improve on that. | 11:16:19 |
| maralorn | Especially anything related to lockfiles is not likely going to help. | 11:16:54 |
| maralorn | What would help would be a) a cultural strengthening that everyone maintains there packages in stackage and b) ecosystem wide sharing of overrides and patches. | 11:17:47 |
| alexfmpe | I still think what nixpkgs does is what stackage should be | 11:22:35 |
| alexfmpe | It's insane to me they don't run test suites | 11:22:56 |
| alexfmpe | Nor ever patch things | 11:23:07 |
| alexfmpe | I don't want to bump my deps if the DB roundtrip tests fail or some encryption library tests fail | 11:24:20 |
| maralorn | I had a cool discussion on zurihac about the fact that there io the concept of hackage overlays. E.g. head.hackage. and it would be awesome if we could maintain our overrides as a hackage overlay which we could import into nixpkgs but als share and maintain together with the rest of the community. | 11:24:24 |
| alexfmpe | Yeah I agree. I don't see the point of there being stackage overlay, nixpkgs overlay, horizon overlay plus iog and obsidian nix jungles | 11:25:44 |
| alexfmpe | We're all tackling the same problem | 11:25:53 |
| alexfmpe | Well stackage isn't an overlay per se | 11:26:34 |
