| 7 Sep 2025 |
 bglgwyng | For some reasone, cabal behaves differently from my local cabal(not in Nix sandbox). | 13:43:12 |
 toonn | Builds are sandboxed by default on Linux, so network access outside of FODs is prohibited. | 13:54:05 |
 chreekat | yes, but that's an unrelated warning. @bglgwyng:matrix.org the error is that cabal is looking for a file 'root.json' next to the tarball. Still I think you'll run into other problems eventually... That root.json is part of a partial TUFT implementation that is gonna look at timestamps and isn't reproducible | 14:00:30 |
| chreekat | * yes, but that's an unrelated warning. @bglgwyng:matrix.org the error is that cabal is looking for a file 'root.json' next to the tarball. Still I think you'll run into other problems eventually... That root.json is part of a partial TUF implementation that is gonna look at timestamps and isn't reproducible | 14:00:38 |
| bglgwyng | I expected network not gonna happen as I configure cabal to use only local repository. But it still happens and it's another problem. | 14:08:59 |
| bglgwyng | As chreekat said, what confuses me is that cabal in Nix tries to read root.json while local cabal doesn't do so and successfully read the package index. | 14:09:40 |
| bglgwyng | * I expected network access not gonna happen as I configured cabal to use only local repository. But it still happens and it's another problem. | 14:15:11 |
| bglgwyng | To provide context, I made deterministic cabal build from plan.json in Nix, and now am trying to generate plan.json deterministic in Nix. | 14:26:21 |
| bglgwyng | https://github.com/haskell/cabal/issues/11185
seems relevant.. | 15:14:19 |
 emily | everything good for me to hit the button on https://github.com/NixOS/nixpkgs/pull/440271 now that the unpins are moved out? | 16:04:13 |
 sterni (he/him) | can't believe we're on track to get this done in under a year | 19:36:22 |
| emily | all I want for Christmas is someone to make Hadrian cross work | 20:20:27 |
| emily | except not so quickly that I feel like I wasted all that time | 20:20:40 |
 Alex | At this point, would it be less effort to compile GHC with MicroHs?
(I haven't been trying in a while, need to revisit it when I get the time.) | 20:21:21 |
| emily | I doubt it, since I sorted the LLVM patches in a few days but AFAIK MicroHs is lacking several major type system extensions to even think about compiling GHC? | 20:22:23 |
| Alex | Last I checked, the main limitation was DataKinds being missing + a lot of GHC-specific code like the fast strings. | 20:24:09 |
| Alex | I'm willing to rewrite the type checker if necessary, and the GHC problems can probably be fixed with a lot of tedious CPP. | 20:25:06 |
|  Ari Ade Supriyatna joined the room. | 20:26:56 |
 Artem | often times, GHC HEAD can't be compiled with a GHC just out of the support window (last three releases or so). I'd expect it to be a non-trivial projecct to make MHS compile GHC.. | 20:32:24 |
| Artem | * often times, GHC HEAD can't be compiled with a GHC just out of the support window (last three releases or so). I'd expect it to be a non-trivial project to make MHS compile GHC... | 20:32:37 |
| emily | does GHC not use type families or generics? | 21:02:18 |
| emily | or implicit parameters or magic hash? | 21:02:31 |
| emily | https://discourse.haskell.org/t/microhs-and-hackage/12916/6 | 21:02:36 |
| emily | sounds unlikely to me | 21:02:43 |
| Alex | Can't know how much effort it is until someone actually tries. | 21:16:57 |
| emily | well | 21:22:32 |
 maralorn | Looking at all the hype that MicroHS is getting I can’t imagine that no one is thinking about it. 😄 | 21:22:44 |
| emily | people tried to get even ancient GHC building with Hugs without success | 21:22:45 |
| emily | so my prior is at least not totally uninformed | 21:23:03 |
| emily | the sad thing about MicroHs is that AFAICT it won't offer a true from-source bootstrap | 21:23:21 |
