| 17 May 2023 |
 hexa | 😄 | 21:45:38 |
 Lily Foster | Oh. The prePatch | 21:46:05 |
| hexa | npmRoot fix when 😄 | 21:46:21 |
| hexa | fixes the build, breaks the update workflow 😫 | 21:46:39 |
| Lily Foster | Well when npmRoot is working, it should fix the update script, right? | 21:47:03 |
| hexa | I would think so | 21:47:09 |
| Lily Foster | Okay, wasn't sure if I was misunderstanding | 21:47:17 |
| hexa | I have evcc which is basically doing the same, but without the need for npmRoot | 21:47:25 |
| hexa | would be great to sort this out before the release, because I did in fact tell people about buildNpmPackage, when I asked them to migrate off of node 14/16 🙂 | 21:48:51 |
| Lily Foster | I do fully intend to have these sorts of things fixed asap, since we're coming up so close on release and with the node 14/16 stuff | 21:49:45 |
| Lily Foster | So I'm thinking npm is ignoring the cache because it's may ending up in this code path, which throws an error without logging and that error gets silently caught (thanks npm, real cool): https://github.com/npm/pacote/blob/3307ad9c1600b6a60b2250c2239240ac41fc7b07/lib/fetcher.js#L317 | 21:53:06 |
| Lily Foster | * So I'm thinking npm is ignoring the cache because it's maybe ending up in this code path, which throws an error without logging and that error gets silently caught (thanks npm, real cool): https://github.com/npm/pacote/blob/3307ad9c1600b6a60b2250c2239240ac41fc7b07/lib/fetcher.js#L317 | 21:53:13 |
| Lily Foster | Time to add some debug logging and figure out what that cacache function is getting called with... hopefully it's not a bug, but npm having a ton of silently broken things because of their silently-bail-out-and-refetch logic all over the place is not news | 21:55:58 |
| Lily Foster | (like their hosted tarball url thing for gitlab is just straight up wrong and they have no intentions to fix it -- so when you do gitlab: deps they're unnecessarily way slower than any other hosted git provider because it tries the tarball, fails, and falls back to checkouts through complicated spaghetti code paths) | 21:57:00 |
| Lily Foster | (so it's "not a problem" because it handles it I guess) | 21:57:21 |
| hexa | yikes 😄 | 22:02:00 |
| hexa | https://github.com/bokeh/bokeh/blob/branch-3.2/bokehjs/package.json#L29-L32 | 22:10:50 |
| hexa | I'm not quite sure which of these workspaces is the interesting one | 22:11:04 |
| Lily Foster | Okay. So. This isn't actually an issue with workspaces. I looked in the lockfile to make sure it existed with more than nothing in it. But I must be tired because while it's present, it's actually just missing integrity and resolved fields | 22:11:19 |
| Lily Foster | * Okay. So. This isn't actually an issue with workspaces. I looked in the lockfile to make sure the p-map dep existed with more than nothing in it. But I must be tired because while it's present, it's actually just missing integrity and resolved fields | 22:11:31 |
| Lily Foster | So this lockfile needs to be deleted and regenerated | 22:12:24 |
| Lily Foster | https://github.com/bokeh/bokeh/blob/branch-3.2/bokehjs/package-lock.json#L101-L114 | 22:12:31 |
| hexa | (django-hijack is finished btw :)) | 22:12:38 |
| hexa | rough, not going to commit a 143 kB lock file into nixpkgs ;D | 22:12:56 |
| hexa | * rough, not going to commit a 143 kB lock file into nixpkgs 😄 | 22:12:58 |
| Lily Foster | npm still will not do automatic fixup of this issue: https://github.com/npm/cli/issues/4460 | 22:13:13 |
| Lily Foster | Which is dumb | 22:13:20 |
| Lily Foster | In reply to @hexa:lossy.network
rough, not going to commit a 143 kB lock file into nixpkgs 😄 Yeah the lockfile committing to nixpkgs across multiple ecosystems has been a bit unfortunate (the node2nix node-packages.nix files included)... You could submit a PR to fix the lockfile and fetchpatch the PR if you are so inclined, to avoid that | 22:16:28 |
| hexa | alright | 22:16:41 |
| Lily Foster | But there's no reproducible way to handle that as-is. Except maybe doing something like how importCargoLock handles git deps with the extra hashes and hope and pray the version exists in the registry | 22:17:29 |
