17 Oct 2024 |
Lily Foster | ofborg did fail. it just didn't yell about it. it does the hydra thing where it just silently drops the attrs | 00:02:39 |
Lily Foster | In reply to @emilazy:matrix.org so maybe it is actually just some accursed mystery * config.allowUnsupportedSystem = true ? | 00:02:49 |
Lily Foster | In reply to @lily:lily.flowers ofborg did fail. it just didn't yell about it. it does the hydra thing where it just silently drops the attrs emily: ^ | 00:02:59 |
Lily Foster | check the ofborg outpaths and you'd see | 00:03:10 |
emily | š« | 00:03:13 |
Lily Foster | unsupported system is a silent drop for ofborg | 00:03:20 |
Lily Foster | basically any check-meta failure that isn't unknown-meta or broken-outputs is ignored | 00:03:47 |
Lily Foster | by ofborgj | 00:03:53 |
Lily Foster | * by ofborg | 00:03:59 |
Lily Foster | which in this base sorta makes sense, but we should have some mechanism to tell when some package has erroneously wider supported platforms than it's dependencies perhaps | 00:04:24 |
Lily Foster | * which in this case sorta makes sense, but we should have some mechanism to tell when some package has erroneously wider supported platforms than it's dependencies perhaps | 00:04:33 |
Lily Foster | * which in this case sorta makes sense, but we should perhaps have some mechanism to tell when some package has erroneously wider supported platforms than it's dependencies | 00:04:55 |
Lily Foster | (which would have caught this, but also would have to be a tree-wide fixup because nixpkgs is hell) | 00:05:30 |
Lily Foster | (also also would benefit from more clearly making sure packages have correct platforms to match what is supported upstream and then broken for what doesn't work in nixpkgs) | 00:06:41 |
Lily Foster | * (and also would benefit from more clearly making sure packages have correct platforms to match what is supported upstream and then broken for what doesn't work in nixpkgs) | 00:07:14 |
Lily Foster | (but also maybe nixpkgs entrypoints and nixpkgs meta supported platform checks shouldn't have been this fucked to begin with to require that silently dropping attrs that fail to eval making sense) | 00:08:27 |
emily | well I at least feel reassured that it wasn't neglect on my part that caused this | 00:14:07 |
emily | well. unless you count scouring the outpaths thing for dropped packages when it rebuilds the world as neglect | 00:14:24 |
emily | * well. unless you count not scouring the outpaths thing for dropped packages when it rebuilds the world as neglect | 00:14:28 |
FabiƔn Heredia | https://github.com/NixOS/nixpkgs/pull/343546
libjxl doesn't currently build on staging-next. Will attempt this version bump. | 03:17:48 |
FabiƔn Heredia | * https://github.com/NixOS/nixpkgs/pull/343546
libjxl doesn't currently build on staging-next. Will attempt this version bump. (x86_64-linux) | 03:17:54 |
FabiƔn Heredia | > The following tests FAILED:
> 2788 - CodecTest.TestRoundTrip (Failed)
Currently, a single test failure
| 03:19:56 |
FabiƔn Heredia | version bump fails the same test | 03:24:24 |
K900 | Merged master to next manually again | 06:07:50 |
Alyssa Ross | my libjxl bisect finished: https://github.com/nixos/nixpkgs/pull/346060#issuecomment-2418554407 | 06:43:01 |
Alyssa Ross | The OpenEXR commit that introduced the failure does not look like an intentional behaviour change. Should we just revert the OpenEXR update? It was just an r-ryantm PR so no sign it's in any way urgent. | 06:44:35 |
K900 | Yeah probably | 06:47:43 |
K900 | That change looks spooky | 06:47:46 |
emily | In reply to @qyliss:fairydust.space my libjxl bisect finished: https://github.com/nixos/nixpkgs/pull/346060#issuecomment-2418554407 "so represents a WIP state" š« | 06:48:31 |
Alyssa Ross | it was part of a huge PR tbf | 06:48:54 |