| 21 Sep 2025 |
 Lun | wait we're on the latest sha already | 18:16:24 |
 K900 | Huh | 18:16:32 |
| K900 | OK | 18:16:32 |
| K900 | I thought that date was weird | 18:16:37 |
| Lun | Patch vendored / swapped to substitute | 18:22:26 |
| K900 | Thanks, sending it | 18:24:04 |
| Lun | i feel like i've been given opposite advice to turn vendored patches into fetchpatching prs before 😨
hope i haven't filled rocm-modules with timebombs | 18:25:25 |
| K900 | Basically github will sometimes just decide to GC merged PRs | 18:26:21 |
| K900 | And the commits disappear | 18:26:28 |
| K900 | We don't entirely know why | 18:26:32 |
| K900 | Or when | 18:26:36 |
 Alyssa Ross | This is documented (somewhat) in pkgs/README.md BTW | 18:42:16 |
| K900 | OK I'm getting close I think | 18:51:04 |
| K900 | Close to building my whole thing that is | 18:51:12 |
| K900 | Still need to fix dotnet | 18:51:28 |
| Alyssa Ross | In reply to @k900:0upti.me
Basically github will sometimes just decide to GC merged PRs Not merged PRs | 18:52:02 |
| Alyssa Ross | They never GC the current PR state, whether open, merged or closed. | 18:52:21 |
| Alyssa Ross | The problem is that an open PR can later be rebased and then they'll GC the older state we had pinned. | 18:52:54 |
| K900 | OK so I now need to build four webkits lol | 19:19:10 |
| K900 | So that's going to be an overnight kind of deal | 19:19:52 |
| K900 | Would appreciate a rubber stamp on https://github.com/NixOS/nixpkgs/pull/445006 | 19:19:59 |
 emily | fetchpatch2 for Jellyfin isn't okay | 19:20:39 |
| emily | that URL has truncated hashes | 19:20:43 |
| emily | it also looks unnecessary? | 19:20:52 |
| emily | no rename | 19:20:54 |
| K900 | I stole it from ffmpeg/generic | 19:22:22 |
| K900 | I don't know why they fetchpatch2 there | 19:22:31 |
| emily | we can't do it without a ?full_index=1 equivalent | 19:22:46 |
| emily | unless you're saying we already have it applied there without that | 19:22:57 |
| emily | in which case eurgh | 19:22:58 |
