yihua opened a new pull request, #759:
URL: https://github.com/apache/hudi-rs/pull/759

   ## Description
   
   closes #758
   
   `verify_src_release.sh` covers the source release and nothing covers the 
published binaries, so the only check on them was two manual lines in the VOTE 
email template. That is how `0.5.0-rc.1` reached a version that exists on 
crates.io and on PyPI with the macOS and Windows wheels, no Linux wheel, and no 
sdist, with nothing to surface it before the vote.
   
   `verify_published_artifacts.sh` checks all three crates on crates.io, the 
full wheel set plus the sdist on pypi.org, then installs the wheel in a stock 
`python:3.11-slim` and reads a table with it. The install deliberately happens 
in a different image from the one the wheel was built in, so it tests the 
platform tag rather than the build container, and the read exercises the merge 
path, since the rocksdb bindings come from the build container's libclang and 
bad bindings surface on import or first read rather than at compile time. The 
release guide now runs it between the tag push and the VOTE thread.
   
   ## How are the changes test-covered
   
   - [ ] N/A
   - [ ] Automated tests (unit and/or integration tests)
   - [x] Manual tests
     - [x] Details are described below
   
   Run against the two published versions that make the best fixtures, one 
complete and one known broken:
   
   | version | result |
   | --- | --- |
   | `0.4.0` (complete) | all 3 crates and all 6 files found, exit 0 |
   | `0.5.0-rc.1` (broken) | reports the two missing manylinux wheels and the 
missing sdist, exit 1 |
   
   On `0.5.0-rc.1` it prints `FAILED: 3 problem(s) found. Do not start the VOTE 
thread.`, which is the outcome that was missing when that RC went out.
   
   The functional half was exercised separately, since neither published 
version pairs with the fixtures on this branch: a wheel built from this tree in 
`quay.io/pypa/manylinux2014` installs in `python:3.11-slim` and reads a COW 
table and two MOR tables, 3 for 3. Running it against `0.4.0` correctly reaches 
the read and fails on `Only support table version 5 and 6`, which is the 
fixture being newer than that release rather than a problem with the check; the 
script notes that it should be run from the release branch for this reason.
   


-- 
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