-
Notifications
You must be signed in to change notification settings - Fork 9.1k
fix: Update packageManager to bun@1.3.6 #9618
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
|
The following comment was made by an LLM, it may be inaccurate: Potential Duplicate PRs Found:
Recommendation: Review PR #9597 in particular, as it appears to be handling the same packageManager Bun version issue that your PR (9618) is addressing. |
|
@MBanucu if you're going to update 1 node_modules hash... why not all 4? |
@gigamonster256 Detailed Investigation Report: PR #8865 and Nix Hash UpdatesIssue SummaryPR #8865 updates OpenCode to use Bun 1.3.6 by modifying Testing Results
Key FindingThe
RecommendationMerge PR #8865 to Related PRs/Issues
This ensures the PR fully resolves the issue once integrated into the main branch. |
|
@gigamonster256 do I understand correctly that it is desired to downgrade the rev of nixpkgs in flake.lock instead of upgrading bun to version 1.3.6? |
|
Let me open a new issue then. |
|
putting ^ in the package.json broke the CI and was removed, however the semver check was then ineffectual |
|
@gigamonster256 i've also made a PR to relax the version #9684 . But i guess your one is more appealing |
IMO nix users/committers should aim to provide a nix dev experience as close to CI as possible - which means keeping deps provided by the nixpkgs rev in flake.lock in sync with what is declared in the rest of the project however a maintainer said its ok to relax the bun version requirement, but we should still aim to diverge as little as possible from non-nix environments |
your pr only fixes the package build, when doing local nix environment based dev with a bun version mismatch it will still error |
@gigamonster256 I think so, too. But I still don't get it, why not just upgrade to bun version 1.3.6. If it is failing on some systems because of bun should branch dev of opencode stay behind? There will always be systems where some bugs appear and some get fixed. If dev wants to be a somehow stable branch then why not use the most stable bun 1.2.X version or even earlier 1.0.X? What is the philosophy behind the dev branch? Is there some documentation about this? I think a good philosophy would be to join forces with bun for a while and drive stability development on bun 1.3 until everyone in opencode is satisfied with the stability of bun and then return to development of opencode because a stable base is important. |
|
the upgrade to bun 1.3.6 was applied but then reverted in 5a8a0f6 nixpkgs rev with bun 1.3.6 only made it into dev bevause of bad CI which updated nixpkgs without checking the package still built |
@gigamonster256 so let me file a new issue with the downgraded flake.lock? Would this help? |
|
You can easily revert it. Why not? |
|
I personally like it more if nixos-unstable of opencode is in sync with nixos-unstable of nixpkgs because then I do not need an extra nixpkgs version on my NixOS extra for opencode. But some decisions have to made, I guess. |
|
Some people benefit, other get sacrificed. This is life. |
|
How about an impact analysis of changes and base the decisions on data? But then opencode would have to collect data from users. Oh, no! We can't do that, because people are not open, they are more closed. Why use opencode then? Open people should use opencode. Closed people should use closedcode. I should stop thinking. This is maybe going too far. |
|
(Temporarily) fixed by #9725 |
Fixes #9583
Updates OpenCode to use Bun 1.3.6, including packageManager, dependencies, and Nix hashes.
Problem:
OpenCode build fails because packageManager specifies bun@1.3.5, but Nix provides 1.3.6.
Solution:
Compatibility:
Bun 1.3.6 is backward compatible; this matches the version in Nixpkgs.
Testing:
Verified with local NixOS rebuild; build succeeds with Bun 1.3.6.
Note: This PR is similar to #8865 but includes the nix/hashes.json update that #8865 lacks, ensuring it works in Nix builds. Alternative to PR #9597, which uses semver ranges instead of exact pinning.