| 12 May 2025 |
 fzakaria | Whose familiar with SourceAccessor in CppNix ?
We were seeing divergent failures for missing files; I was seeing "FileNotFoundError" and Shahar "Dawn" Or (mightyiam) was seeing "SysError" (although not in unit tests).
We've rebased and it's gone away but it's bugging me not knowing the difference.
| 15:58:50 |
| fzakaria | My best guess -- is his Nix build was picking a different SourceAccessor. Mine seemed to be picking PosixSourceAccessor. | 15:59:27 |
|  @michael.zeagler:matrix.org joined the room. | 20:25:48 |
| 13 May 2025 |
 John Ericson | Eelco: if (settings.buildHook.get().empty() || !worker.tryBuildHook || !useDerivation) return rpDecline; | 19:31:48 |
| John Ericson | do you know what !useDerivation is needed? | 19:31:56 |
| John Ericson | this comes from 1511aa9f488ba0762c2da0bf8ab61b5fde47305d | 19:32:25 |
| John Ericson | what confuses me is that I thought we would convert to using in-memory derivations with the build hook | 19:33:06 |
| John Ericson | i.e. we computed the inputs closure, and then we send that to the build hook | 19:33:17 |
| John Ericson | and thereafter we're working in-memory | 19:33:25 |
| John Ericson | so I don't get what this condition accomplishes, and yet if I remove it, tests hang | 19:33:53 |
| John Ericson | does remote building even work with an eval store? | 23:00:49 |
| 14 May 2025 |
 jade_ | i just released https://jade.fyi/blog/misadventures-in-dtrace/ and it occurs to me that it might be prudent to give y'all a heads-up of our plight with MonitorFdHup and to suggest potentially picking our macOS platform-specific implementation of it | 02:23:25 |
| 15 May 2025 |
|  James Wilmot joined the room. | 01:27:33 |
