| 18 Sep 2025 |
K900 | https://wiki.qt.io/Qt_6.10_Release | 19:36:16 |
K900 |  Download image.png | 19:36:24 |
K900 | OK make that late October | 19:36:29 |
Lun | I can't easily get parmetis to build with the metis/gklib bump for cmake 4 compat. is it ok to mark it broken? since parmetis is unfree it seems like things that depend on it have it as an optional dep and off by default.
I guess the alternative would be to try not bumping any of them and pass -DCMAKE_…=3.5 for metis, keeping it on the 2013 version 🙃
| 19:59:27 |
emily | maybe ping the maintainer | 20:00:47 |
emily | how load-bearing is metis? do we need a fix for it soon? | 20:00:52 |
emily | we can always land https://github.com/nim65s/nixpkgs/commit/64cb7e5f0c13b2dd36f9634540132e9e11933a81 now and then bump it later | 20:01:03 |
emily | but 2013 doesn't sound good | 20:01:08 |
K900 | /nix/store/knm0ara2f6i6q3fs6m5rpabhj034f4f1-metis-5.1.0.drv x86_64-linux 321
/nix/store/z348wv1m9gj9pw73xqycr5p7qzmg6saz-suitesparse-5.13.0.drv x86_64-linux 130
/nix/store/a98vnmd39c4dfp2drcpdifyi2bnqlxqn-ceres-solver-2.1.0.drv x86_64-linux 122
/nix/store/dxar08a762mc0zwyyrsf4ngpwkcm0i7q-mumps.drv x86_64-linux 40
/nix/store/c9j4vrlq0c8h4mls95f6xmm6m78d9mkz-petsc-3.23.5.drv x86_64-linux 6
/nix/store/i1nrib0sbp2c1nhawkjzkzslm1rxcdcx-petsc-3.23.5.drv x86_64-linux 4
/nix/store/5p0ixslzqgh2sdi0gw3gh64c7z9v0wjf-petsc-3.23.5.drv x86_64-linux 4
/nix/store/41806f9js9243rpk8v3f36vn644n9mn3-iEDA-0-unstable-2025-06-30.drv x86_64-linux 1
/nix/store/p8sabys3gcmzi1b18gkl3vj1c6b2wa37-kahip-3.18.drv x86_64-linux 1
/nix/store/9hgi75j4lbsxdk25q9q5fyyh08n2dp93-mumps.drv x86_64-linux 1
/nix/store/md4rw62wxla87z1h7fklkf81dkngvmdc-netgen-6.2.2504.drv x86_64-linux 1
/nix/store/khl9m9sqd5zx54cz8igpm8xbnizv8bx6-octopus-16.0.drv x86_64-linux 1
/nix/store/krcx5j35impy3h5j35ck1a4g6d0j635n-p4est-2.8.7.drv x86_64-linux 1
/nix/store/vwkjs5dsxwbgcgw556l66xr7s46n3krc-p4est-2.8.7.drv x86_64-linux 1
/nix/store/4gvikjqxxb6h169yzby0vf8jssjmh4cj-parmmg-1.4.0-unstable-2024-04-22.drv x86_64-linux 1
/nix/store/vzrsnaihh9yizvx3ggcg0p5jxzjc2lk6-kahip-3.18.drv x86_64-linux 1
/nix/store/7hdcya8zvsa59h4204zk3ffhyyzwrxkc-netgen-6.2.2504.drv x86_64-linux 1
/nix/store/qivpp6yls0i3grza69lbadw0aqyqxq9m-python3.12-opensfm-unstable-2023-12-09.drv x86_64-linux 1
/nix/store/885hcackcfmgls8ca9rf3ng82pfanaxa-kahip-3.18.drv x86_64-linux 1
/nix/store/v5c6b8pjg6znx4q8fqdsri680v1a5d58-python3.13-opensfm-unstable-2023-12-09.drv x86_64-linux 1
/nix/store/8ywj5a4n1yjxim0vnmxjhdhnmshf4xbf-sofa-24.12.00.drv x86_64-linux 1
/nix/store/n9lnrh1w2w6cl1m46593183f6nn3pngz-spral-2025.05.20.drv x86_64-linux 1
| 20:01:38 |
K900 | All of this looks extremely niche tbh | 20:01:43 |
emily | well a lot of the closure is non-free I think | 20:01:55 |
emily | but yes | 20:01:59 |
Lun | i guess nim65's commit for now would be fine
torchWithRocm depends on metis indirectly right now via suitesparse | 20:02:09 |
K900 | I think you'll just have to find the two people that care about this | 20:02:11 |
emily | looks like we found one of them | 20:02:19 |
Lun | 😭 | 20:02:25 |
Lun | i'm sorry next time i'll buy nvidia | 20:02:33 |
Lun | i think parmetis is super niche due to the bad license but metis is a dep of a lot more indirectly | 20:03:23 |
K900 | It is a dep of exactly 321 things | 20:03:41 |
K900 | On like three day old master at least | 20:03:45 |
Lun | whereas parmetis is … 1 thing that isn't behind a withParmetis flag that defaults off? | 20:04:08 |
K900 | Man this clang build is taking a minute | 20:05:08 |
K900 | I really want to see it work and go the fuck to sleep tbh | 20:05:17 |
Lun | in nixpkgs or the original software? | 20:05:27 |
Lun | no maintainer in nixpkgs | 20:05:33 |
Lun | there's an issue about compile issues that has had no response on upstream already bleh | 20:06:28 |
K900 | Just break it and see who screams | 20:06:32 |
Lun | * no maintainer in nixpkgs I looked at the wrong one we have one for parmetis! | 20:08:16 |
emily | for ParMETIS? | 20:08:18 |
emily | I reviewed a PR by someone for it a while back | 20:08:25 |