| 7 Nov 2025 |
Mic92 | Sure you didn't build meson in release mode or something by accident? | 19:32:04 |
Mic92 | This was with gcc now, didn't test clang etc | 19:33:51 |
John Ericson | Mic92: I see the PR | 19:51:01 |
John Ericson | so wow I thought that TBB thing had something to do with blake3 | 19:51:16 |
John Ericson | but turns out it is unrelated? | 19:51:20 |
John Ericson | is libstc++ doing some has_include trickier by anychance? | 19:51:35 |
John Ericson | * is libstc++ doing some has_include trickier by any chance? | 19:51:37 |
Mic92 | John Ericson: it was triggered by us adding blake3, seems like gcc has a trigger. | 19:51:37 |
Mic92 | it does | 19:51:48 |
Mic92 | https://github.com/gcc-mirror/gcc/blob/d14d24b26126ec9c6c8ba5b2e573ed8e26347e6b/libstdc%2B%2B-v3/include/bits/c%2B%2Bconfig#L943 | 19:51:54 |
John Ericson | ah yeah | 19:53:55 |
John Ericson | libstdc++-v3/include/bits/c++config:# define _GLIBCXX_USE_TBB_PAR_BACKEND __has_include(<tbb/tbb.h>) | 19:54:00 |
John Ericson | I would mention that + the blake3 in the commit message | 19:54:16 |
John Ericson | IMO it boils down to, we want tbb to act like an (optional) private dependency of blake3, but it is public and leaking out like this | 19:54:58 |
Mic92 | added | 19:58:53 |
Mic92 | * added https://github.com/NixOS/nix/pull/14509/files#diff-cad591cdc5170341a06019235c8734d55351bb1683223e0ddccabc1446452712R49 | 19:59:20 |
John Ericson | approvd | 20:09:16 |
Sergei Zimmerman (xokdvium) | Nixpkgs bump: https://github.com/NixOS/nixpkgs/pull/459549. If anybody wants to throw some non x86_64-linux compute on it that’d be great | 21:36:34 |
| @emma:rory.gay left the room. | 22:40:57 |
| 8 Nov 2025 |
| cassianistefano joined the room. | 15:55:40 |
| Mindavi joined the room. | 18:00:46 |
| 9 Nov 2025 |
| ghpzin (moved to @ghpzin:envs.net) changed their display name from ghpzin to ghpzin (moved to @ghpzin:envs.net). | 15:03:44 |
| 10 Nov 2025 |
tomberek | John Ericson: https://hydra.nixos-cuda.org/project/nix | 03:21:46 |
John Ericson | tomberek: nice! | 03:23:50 |
tomberek | CUDA team's Hydra. Also added the Oxide builders to it. | 03:24:42 |
John Ericson | tomberek: very good! I guess we'll need an i686 builder heh (or it is time to drop that entirely?) | 03:28:45 |
John Ericson | it's good we can actually look at the eval errors nwo | 03:28:53 |
John Ericson | per PR | 03:28:58 |
John Ericson | and think about these things some more | 03:29:03 |
tomberek | We can either hook up the GitHub PR integration or just do it manually for a bit. (Or make a CLI tool and API it....... hrm...) | 03:30:36 |