Haskell in Nixpkgs/NixOS | 721 Members | |
| For discussions and questions about Haskell with Nix, cabal2nix and haskellPackages in nixpkgs | Current Docs: https://haskell4nix.readthedocs.io/ | More Nix: #community:nixos.org | More Haskell: #haskell-space:matrix.org | 144 Servers |
| Sender | Message | Time |
|---|---|---|
| 28 Jan 2025 | ||
| Anyone ever seen this sort of behavior? I added a nix override to use a certain version of beam and now an upstream package which uses beam (rhyolite-account-types) is saying that it exports it, but it most definitely does not | 23:10:39 | |
| Or im completely misunderstanding this error | 23:10:50 | |
| 29 Jan 2025 | ||
hmm might be Id was pulled at some point and you're on an older rhyolite that still has it? | 01:44:36 | |
| ah no I'm misreading | 01:45:12 | |
do you have a local hoogle setup? what comes up when searching for PrimaryKey ? | 01:47:14 | |
| hmmm looking at the "originally defined" bits, it kind of looks like you're getting one beam-core from an unpacked thunk of the repo and another from cabal/hackage or something? | 01:50:40 | |
huuuuh are you using ob ? | 01:55:49 | |
I'd try cabal clean and re-enter the nix-shell after making sure you're throwing the thunk beam-core properly into the nix bitsmaybe add https://github.com/obsidiansystems/obelisk/pull/1102/files for good measure | 01:57:23 | |
| hmm I believe rhyolite's nix brings
| 02:06:08 | |
need to re-enter the shell to have the beam-core override take full effect | 02:07:23 | |
(yes, ob is a pile of hacks, we're replacing it in near future with the proper cabal multi-repl now that it's out there) | 02:07:51 | |
| * hmm I believe rhyolite's nix brings
| 02:08:33 | |
| also, for obsidian-verse questions, #reflex-frp:matrix.org ends up being the default place to ask, due to heavy user/maintainer overlap | 02:09:47 | |
There is no real chance to override mtl with mtl_2_3_1 in GHC <= 9.4, right? Because it's a boot library? | 13:34:51 | |
When I do it, it seems to rebuild shake, which I definitely don't want | 13:35:14 | |
| Overriding boot libraries is fine in principle with a few exceptions which mtl is not afaik. | 13:47:35 | |
| It triggering a rebuild of shake is to be expected. If you don't want that you need to pick shake from an unoverriden packageset | 13:48:42 | |
| Fine in the sense of "It will just rebuild GHC from scratch and then everything should work"? 😅 | 13:49:06 | |
| Ah, I don't really know why exactly it builds shake, I don't specify it anywhere explicitly | 13:49:34 | |
| I took care to take e.g. cabal-install from an unoverridden package set | 13:50:47 | |
| No, unless you are doing something very explicit to ghc it won't rebuild it. | 13:50:50 | |
In reply to @manuelbaerenz:matrix.orgWell maybe it's a dep of hls? | 13:51:23 | |
| Maybe, I'll look into that, but don't I need to take HLS from the overridden set? After all, I want it to load the right mtl | 13:51:55 | |
| This is getting hairy. But no, the hls just needs to be from a packageset built with the same ghc. | 13:52:41 | |
| I changed HLS to come from the vanilla pkgs and still it rebuilds shake 😅 I think I need to dive deeper into where this is coming from | 13:54:00 | |
| I have no 100% proof of this but I think the only libs that need to match are unoverridable anyway and will thus agree between different packagesets. | 13:54:09 | |
In reply to @manuelbaerenz:matrix.orgnix why-depends might help | 13:54:56 | |
| The trace from my build command looks like this:
To me this looks like it tries to bootstrap the build of GHC 9.6 from a binary GHC 9.2, and fails while trying to build shake, which is needed for hadrian | 13:57:44 | |
| * The trace from my build command looks like this:
To me this looks like it tries to bootstrap the build of GHC 9.6 from a binary GHC 9.2, and fails while trying to build shake, which is needed for hadrian | 13:58:07 | |
| * The trace from my build command looks like this:
To me this looks like it tries to bootstrap the build of GHC 9.6 from a binary GHC 9.2, and fails while trying to build shake, which is needed for hadrian | 13:58:28 | |