| 2 Dec 2025 |
 raitobezarius | it will be fixed in 2.94.1 | 23:29:10 |
| raitobezarius | thanks for your patience! | 23:29:35 |
 xored | amazing fam | 23:29:50 |
| xored | hmm disabling pasta didn't work though | 23:30:01 |
| xored | same error | 23:30:04 |
| xored | thanks a lot btw | 23:30:08 |
| xored | ohhh I can't read | 23:30:24 |
| xored | warning: Ignoring the client-specified setting 'pasta-path', because it is a restricted setting and you are not a trusted user | 23:30:31 |
| xored | guess we're rebuilding the host | 23:30:55 |
| xored | it's building, so it works without pasta, but you already knew that | 23:35:29 |
 leo60228 | just saw this cppnix pr from the other day, seems interesting? https://github.com/NixOS/nix/pull/14686 | 23:45:04 |
| leo60228 | haven't looked at the implementation but i like the idea of being able to attach metadata to store paths that don't affect the hash | 23:45:21 |
| leo60228 | (the idea is for auditability, i.e. getting more useful information about what software is installed on a machine than just a list of store paths) | 23:46:04 |
| raitobezarius | yup | 23:59:03 |
| raitobezarius | the concept of meta is not to affect the on-disk representation of drv, this PR seems to be going against all these invariants no matter what the new modulo algorithm become | 23:59:51 |
| 3 Dec 2025 |
| raitobezarius | auditability can already be achieved today by recording meta out-of-band like many projects does it, having a nicer way to do is definitely desireable, but i don't think this particular implementation would fly | 00:02:03 |
 Qyriad | we've considered storing meta in xattrs | 08:21:03 |
