| 10 Oct 2025 |
 lovesegfault | addressed and pushed 🥳 | 17:33:14 |
| lovesegfault | Another little bit of the curl-s3 work: https://github.com/NixOS/nix/pull/14209 | 19:41:22 |
 nulleric | i'm new to matrix so apologize if this is the wrong place, i saw someone say they discussed this on matrix but i cant find it searching - cmake3 has been removed from unstable as it is EOL - but no one has shared a link to any announcement of cmake3 being EOL. just trying to understand this - i have 0 care if nixpkgs migrates to cmake4 by default - but in the embedded world there's still things out there with cmake2 in use today | 20:19:48 |
| nulleric | case in point - a submodule in the latest picotool uses cmake2.18 i believe (though need more direction on how this works) https://github.com/NixOS/nixpkgs/pull/450680 | 20:23:52 |
 John Ericson | @nulleric:matrix.org: you'll want to take that to #dev:nixos.org | 20:28:48 |
| John Ericson | This channel is for Nix the tool, not for Nixpkgs | 20:29:12 |
| John Ericson | @xokdvium:matrix.org: I have a suspicion that the assert bug and the bug that needed the revert are two symptoms of the same bug | 20:29:43 |
 Sergei Zimmerman (xokdvium) | What's the clue? | 20:30:05 |
| John Ericson | My suspicion is because they both seem to go away the second time | 20:30:20 |
| John Ericson | I think as I make more goals, we're seeing some underlying issue in the work infra | 20:31:03 |
| John Ericson | Still have no idea what the hidden state is, though | 20:31:16 |
| nulleric | thanks! | 20:31:24 |
| John Ericson | No problem, people make this sort of mistake all the time :) | 20:31:48 |
| nulleric | (i dont understand how matrix works yet, that channel you linked seems to not even be part of this "server") anyways i'll figure it out i guess ha | 20:35:02 |
| John Ericson | nulleric: oh oops that was my bad | 21:05:14 |
| John Ericson | I did do it wrong | 21:05:17 |
| John Ericson | #dev:nixos.org this channel | 21:05:28 |
| John Ericson | sorry! | 21:05:34 |
| John Ericson | Sergei Zimmerman (xokdvium): your suggestion didn't quite work so I just std::ranges:find harder until it did :) | 21:14:54 |
| John Ericson | if (auto it = std::ranges::find_if(
externalBuilders.get().begin(),
externalBuilders.get().end(),
[&](const auto & handler) {
return std::ranges::find(handler.systems, drv.platform) != handler.systems.end();
});
it != externalBuilders.get().end())
return &*it;
return nullptr;
| 21:15:48 |
| John Ericson | :) | 21:15:54 |
| Sergei Zimmerman (xokdvium) | Oh no why | 21:18:54 |
| John Ericson | Sergei Zimmerman (xokdvium): added a commit to make it a set | 21:19:15 |
| John Ericson | the inner thing | 21:19:21 |
| lovesegfault | Sergei Zimmerman (xokdvium): RE https://github.com/NixOS/nix/pull/14209#discussion_r2422146488
I can make the change to include the s3 params in the cacheKey
| 21:56:29 |
| lovesegfault | shouldn't be hard | 21:56:30 |
| Sergei Zimmerman (xokdvium) | I think it should be done somewhat centralized and with a common abstraction for the stores. This is a pre-existing problem already for the s3 store thoughh | 21:57:18 |
| Sergei Zimmerman (xokdvium) | * I think it should be done somewhat centralized and with a common abstraction for the stores. This is a pre-existing problem already for the s3 store though | 21:57:21 |
| lovesegfault | ACK | 22:04:18 |
| lovesegfault | pushed a new rev of https://github.com/NixOS/nix/pull/14209 if you could take another look :) | 22:06:30 |
