linliu-code opened a new pull request, #780:
URL: https://github.com/apache/hudi-rs/pull/780

   ## What this is
   
   Position **1 of 4** in a series porting work from Onehouse's `hudi-rs` fork 
onto upstream `main`. The series is linear and must land in order: **01 → 02 → 
03 → 04**.
   
   Because a PR to `apache/hudi-rs` must be based on a branch in this 
repository, the later positions cannot be stacked on GitHub. Each is therefore 
raised against `main` and its diff is cumulative. **This PR is the base of the 
series — its diff is its own work only** (10 commits, `3abc4d5..04b9235d`).
   
   ## Why
   
   Two independent problems, bundled because neither is useful alone on a fork:
   
   1. **The workflows do not run on a fork, or on a PR whose base is not 
`main`.** `pull_request.branches: [main]` means a PR onto a feature branch 
produces *zero* workflow runs — which renders as green rather than as "not 
run". That is the enabling change for everything downstream.
   2. **`demo/infra/mc` cannot build.** The `mc` binary is fetched from 
`dl.min.io`, which now returns **HTTP 410**, so every `integration-tests` job 
fails at image build. This affects upstream exactly as much as it affects a 
fork.
   
   ## What changed
   
   - Drop `pull_request.branches: [main]` from `ci.yml`, `code.yml`, `pr.yml`, 
`asf-allowlist-check.yml`.
   - Guard actions only `apache/hudi-rs` may take (coverage upload, 
PyPI/crates.io publish) behind `github.repository`, so they skip on a fork 
instead of failing.
   - `.github/scripts/workspace-version.sh` — print the single 
`[workspace.package] version` from one authority, replacing a `grep | head -n 
1`.
   - Disk-budget fixes for `rust-tests`, `python-tests` and the coverage job; 
`[profile.dev.package."*"] debug = false`.
   - A `cpp-ffi` job covering the `cpp` workspace member.
   - `demo/infra/mc/Dockerfile` — take `mc` from MinIO's published image 
instead of the dead download URL.
   
   ## Note for upstream reviewers
   
   Parts of this are fork-shaped and deserve a decision rather than a rubber 
stamp. The `github.repository` guards are no-ops here (the condition is always 
true upstream); dropping the base-branch filter widens when workflows run; the 
`cpp-ffi` job is new CI cost. **The unambiguously upstream-relevant piece is 
the `mc` Dockerfile fix** — that is broken on `main` today. I am happy to split 
this into the `mc` fix alone plus a separate CI discussion if the reviewers 
prefer.
   
   ## Testing
   
   `cargo check --workspace --all-targets` and `cargo fmt --all --check` clean 
at the source commit. In the fork this position is what first produces check 
runs at all: 18 checks, 16 passing, 2 skipped, 0 failing.
   
   🤖 Generated with [Claude Code](https://claude.com/claude-code)
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to