| 29 Jul 2025 |
 Mic92 | It did turn it on for nix-eval-jobs when I was testing | 21:59:56 |
 John Ericson | ah gotcha | 22:00:07 |
| John Ericson | once we stop using the silly drv hashes in realisation keys | 22:00:29 |
| John Ericson | you just look up other table with drv path | 22:00:36 |
| John Ericson | there is no reason to read in derivation to compute "static output paths" or anything like that | 22:01:14 |
| Mic92 | https://github.com/nix-community/srvos/blob/cc76247e77c04cf2ad0bf04be10a5d47bc3da594/nixos/mixins/nix-experimental.nix#L29 | 22:03:03 |
| John Ericson | ah, OK | 22:12:25 |
| 30 Jul 2025 |
| John Ericson | OK based on today's saga, would like to discuss https://github.com/NixOS/nix/pull/13263#issuecomment-3134777234 tomorrow and hopefully merge it | 03:59:06 |
|  Kuritsu joined the room. | 04:33:42 |
 fzakaria | so no new meeting for Monday ? | 14:54:45 |
| fzakaria | I hope to come today a bit late. | 14:54:48 |
| fzakaria | although I didn't hack much this week. | 14:54:53 |
| fzakaria | mostly working with Yannick on some fetchGit cache bug | 14:55:10 |
| John Ericson | there was one later in the day | 14:57:36 |
| John Ericson | I guess the calendar is not updated still | 14:57:44 |
| John Ericson | https://github.com/nlohmann/json/discussions/2612 | 14:57:48 |
| John Ericson | so maybe we should use that for the new derivation format | 14:57:55 |
| John Ericson | canonical and easy to read | 14:58:01 |
| John Ericson | frankly, I don't think non-canonical would cause issues, because it is fine to have fewer cache hits | 14:58:24 |
| John Ericson | but it do feel better with it | 14:58:28 |
| John Ericson | * frankly, I don't think non-canonicity would cause issues, because it is fine to have fewer cache hits | 14:58:49 |
| John Ericson | oh the RFC exists, but nlohmann doesn't yet implement it | 14:59:16 |
 emily | (from the last time this came up, including my strong warning against canonicalized JSON formats and suggestions for alternatives) | 15:00:26 |
| emily | the good thing about JSON canonicalization schemes is that there are so many to pick from! | 15:00:54 |
| John Ericson | emily: how about CBOR? | 15:09:51 |
| John Ericson | Robert Hensing (roberth) was once concerned that making it binary would be a humans debugging drawback | 15:10:24 |
| emily | mentioned here (there's also three versions of that spec unfortunately) | 15:10:39 |
| John Ericson | to be clear, we need a parser that validates the canonical format | 15:11:19 |
| John Ericson | it is not good enough to make just the serializer do the thing | 15:11:31 |
| emily | IMO textual formats are good but you need to keep them separate from the non-malleable canonical forms since the two are pretty in tension. Preserves does this. (though again I doubt Preserves is appealing for direct adoption given lack of ecosystem, it's just the measuring stick for these things in my view) | 15:11:38 |
