| 10 Aug 2023 |
 @ambroisie:belanyi.fr | Because I don't want to find out that either point is wrong :') | 17:01:15 |
 dotlambda | Haha | 17:01:41 |
 Lily Foster | It does lockfile v1 output so npm can still get spooked. It should be deterministic but may not be reproducible across versions so doing it in updateScript is probably a better idea | 17:05:24 |
| dotlambda | What does "reproducible across versions" mean? | 17:13:35 |
| dotlambda | Using it in an updateScript has the big disadvantage of having to commit thousands of lines of code to nixpkgs. | 17:13:35 |
| dotlambda | In reply to @robert:funklause.de
What does "reproducible across versions" mean? Oh, across versions of pnpm-lock-export | 17:16:33 |
| Lily Foster | In reply to @robert:funklause.de
What does "reproducible across versions" mean? Like if the output is one way in version 1.0 but then later if the package is updated to 2.0, the output could be subtly different (ask me how I've become paranoid about this 🙃) | 17:16:53 |
| Lily Foster | In reply to @robert:funklause.de
Oh, across versions of pnpm-lock-export Yeah | 17:16:56 |
| Lily Foster | * In reply to @robert:funklause.de
What does "reproducible across versions" mean?
Like if the pnpm-lock-export output is one way in version 1.0 but then later if the package is updated to 2.0, the output could be subtly different (ask me how I've become paranoid about this 🙃) | 17:17:08 |
| dotlambda | That could be remediated by using different attributes in all-packages.nix for different versions | 17:18:23 |
| Lily Foster | It could maybe. I've seen it change across minor versions for, e.g., tomlq though, so unless the project specifically either allows specifying a reproducible format or promises output reproducibility I've become distrustful of it :) | 17:20:14 |
| Lily Foster | I'd rather we just made our own pnpm-lockfile-equivalent because then we would be in control of the output reproducibility | 17:20:38 |
| Lily Foster | (And not just because npm still does Bad Stuff™️ with v1 lockfiles sometimes) | 17:21:04 |
| Lily Foster | In reply to @countoren:matrix.org
Lily Foster: Thank you for your help Okay apparently it was RUST_LOG not RUSTLOG, but don't worry about grabbing the log. I fiddled with it locally and found the problem. Can you try updating your flake input from that branch, invalidate the fetcher hash, and try again? | 17:49:34 |
| Lily Foster | In reply to @countoren:matrix.org
Lily Foster: Thank you for your help * Okay apparently it was RUST_LOG not RUSTLOG, but don't worry about grabbing the log anymore. I fiddled with it locally and found the problem. Can you try updating your flake input from that branch, invalidate the fetcher hash, and try again? | 17:49:44 |
| Lily Foster | * oren: Okay apparently it was
RUST_LOG not RUSTLOG, but don't worry about grabbing the log anymore. I fiddled with it locally and found the problem. Can you try updating your flake input from that branch, invalidate the fetcher hash, and try again? | 18:15:24 |
 countoren | sure | 18:43:16 |
| dotlambda | In reply to @lily:lily.flowers
I'd rather we just made our own pnpm-lockfile-equivalent because then we would be in control of the output reproducibility We can fork the project cause it looks kinda dead anyway. | 18:44:59 |
| Lily Foster | In reply to @robert:funklause.de
We can fork the project cause it looks kinda dead anyway. I haven't looked into the code enough, but if it's not cancer, sure if you want go ahead | 18:45:56 |
| countoren | In reply to @lily:lily.flowers
oren: Okay apparently it was
RUST_LOG not RUSTLOG, but don't worry about grabbing the log anymore. I fiddled with it locally and found the problem. Can you try updating your flake input from that branch, invalidate the fetcher hash, and try again? same branch right? | 18:46:14 |
| Lily Foster | In reply to @countoren:matrix.org
same branch right? Yep! | 18:47:25 |
| countoren | I think we downloaded them, I am getting missing build script tho | 18:49:49 |
| Lily Foster | Does the package have a build script in package.json? | 18:50:17 |
| countoren | the package.json I have here is just deps | 18:50:18 |
| countoren | no | 18:50:22 |
| Lily Foster | Or is it possibly called something else? | 18:50:28 |
| countoren | was using "outside" npx command | 18:50:32 |
| countoren | npx ng build | 18:50:41 |
| countoren | does buildNpm give you ability to define it in nix? | 18:51:06 |
| Lily Foster | Ah, if it doesn't need to be built then you can just do dontNpmBuild. But if you're needing to run a command to build, it might not be a bad idea to add it to the package.json | 18:51:16 |
