| 21 Sep 2025 |
Grimmauld (any/all) | i don't want to carry more than necessary | 12:07:04 |
Vladimír Čunát | Too new version can also create work, as you see, but the security maintenance work is hard to predict. Though recently they did have lots of CVEs. | 12:10:06 |
Vladimír Čunát | Anyway, if you think 2.15 will be better, I certainly don't oppose that. | 12:10:49 |
K900 | So who's getting sniped into writing multilib stdenv for 26.05 | 12:22:22 |
K900 | (please don't be me) | 12:22:29 |
K900 | (this message is brought to you by pandoc-i686-linux) | 12:24:03 |
emily | I trust containers org | 13:11:42 |
emily | so switching seems good | 13:11:46 |
K900 | If you want to snipe me to it, feel free | 13:12:04 |
K900 | I'm just waiting for my big pile of i686 to complete | 13:12:16 |
K900 | So I can then triage whatever fell out | 13:12:34 |
K900 | (34 errors so far) | 13:12:39 |
emily | this confuses me - I'm pretty sure I built toml11? 🤔 | 13:12:47 |
emily | wait I thought we didn't depend on GHC on i686 | 13:15:05 |
K900 | Who's we | 13:15:16 |
emily | IIRC it wasn't cached when I checked | 13:15:17 |
emily | Nixpkgs | 13:15:22 |
K900 | Well I'm building a pandoc | 13:15:32 |
K900 | So that's not true | 13:15:38 |
K900 | Maybe on Darwin | 13:15:44 |
Wolfgang Walther | Which version of GHC did you check, emily? | 13:15:58 |
Wolfgang Walther | It could only be the default version that is cached. | 13:16:05 |
emily | no this was Linux. (i686-darwin is not a thing) | 13:16:07 |
emily | pkgsi686Linux{,.buildPackages}.ghc IIRX | 13:16:50 |
emily | perhaps I do not recall xorrectly | 13:17:14 |
Wolfgang Walther | That seems cached for me (I don't think I ever built this on master) | 13:17:14 |
emily | okay well that's useful/depressing information | 13:17:34 |
Wolfgang Walther | % nix-build -A pkgsi686Linux.ghc
/nix/store/0bd49qs1cy7qgdhsg24gbj4y2v4zdg7l-ghc-9.6.7
The PR we were discussing was about 9.8.4 and 9.10.x
| 13:17:41 |
Wolfgang Walther | Still, I think pandoc is mostly set up as "if available to bootstrap from, then do it" | 13:18:21 |
Wolfgang Walther | So it could very well be, that we could still remove i686-linux GHC without actually breaking stuff, except not building some docs anymore for that platform. | 13:18:50 |