| 10 Jul 2024 |
 TyberiusPrime (smilodon inopinatus) | and of course that does have the drawback that nix has to download the source before it knows how to build the package | 10:19:15 |
 truh | I guess if you don't need the info earlier than that it should work out. | 10:19:37 |
| TyberiusPrime (smilodon inopinatus) | it's the same place that overrides/build-systems.json comes into play | 10:19:52 |
| truh | Oh really, never made it that far into the abyss myself | 10:20:41 |
| TyberiusPrime (smilodon inopinatus) | and now I'm thinking that there's terrible little difference between 'fileExists ${src}/pyproject.toml', and actually importTOML | 10:20:59 |
| TyberiusPrime (smilodon inopinatus) | and yes, obviously it would be worlds nicer if poetry.lock actually had version-locked build-system information. | 10:21:15 |
| TyberiusPrime (smilodon inopinatus) | but it don't | 10:21:18 |
| TyberiusPrime (smilodon inopinatus) | and I think that we're pulling the build-system packages from nixpkgs anyway. | 10:21:32 |
| truh | I was wondering how the build systems were bootstrapped in p2n | 10:23:48 |
| TyberiusPrime (smilodon inopinatus) | I haven't dug into it, but at least I couldn't get anything using 'whey' to actually be build easily. | 10:24:08 |
| TyberiusPrime (smilodon inopinatus) | (that's a (minor) python build system) | 10:24:17 |
| truh | I saw that, I asummed it should just work if you tell it which build system to add to the buildInputs as long as the p2n is actually able to build the build system. But there is something about known build systems in poetry2nix, maybe that's required | 10:27:21 |
| TyberiusPrime (smilodon inopinatus) | tried that, and well, I have added things to build-system-overrides that ain't in that list, and whey won't be found even if in that list, so I don't know. | 10:29:02 |
| TyberiusPrime (smilodon inopinatus) | It's a series of overlays, all teh way down | 10:30:29 |
| TyberiusPrime (smilodon inopinatus) | if only I wasn't fighting the CI so hard. | 10:34:46 |
| TyberiusPrime (smilodon inopinatus) | (one turn around on github takes 3-ish hours) | 10:34:54 |
| TyberiusPrime (smilodon inopinatus) | and, well the way it's setup, you can somewhat run it locally with act | 10:35:07 |
| TyberiusPrime (smilodon inopinatus) | after you've commented out all the tests you don't care about | 10:35:18 |
| TyberiusPrime (smilodon inopinatus) | and if you haven't commited | 10:35:20 |
| TyberiusPrime (smilodon inopinatus) | also things weirdly fail with 'you have a /homeless-shelter', so somehow the env is not quite as it is in the actions | 10:36:05 |
| TyberiusPrime (smilodon inopinatus) | (you can run the tests individually using nix, but again, it's not quite the same env) | 10:36:20 |
| TyberiusPrime (smilodon inopinatus) | and... the aarch64 test failed once again. | 10:37:11 |
| TyberiusPrime (smilodon inopinatus) | I think it's also some kind of aggregate test-lots-of-things thing. Why you even have a matrix of os * tests... | 10:37:29 |
| truh | Yesterday I waited for a CI build for over an hour just to realize I had forgotten -L after nix build and gained basically no insight at all. | 10:37:35 |
| TyberiusPrime (smilodon inopinatus) | we've all been there. | 10:38:01 |
| TyberiusPrime (smilodon inopinatus) | and you can't download the logs before all tests have finished | 10:38:11 |
| TyberiusPrime (smilodon inopinatus) | and the github gui dosn't really work well with those megabytes of output^^ | 10:38:24 |
| TyberiusPrime (smilodon inopinatus) | so I'll wait for the next one to fail and then we'll look agai.n | 10:38:33 |
| TyberiusPrime (smilodon inopinatus) | * so I'll wait for the next one to fail and then we'll look again. | 10:38:35 |
| truh | I wish there was a way to cache the stuff that didn't fail to build. | 10:39:54 |
