voonhous opened a new pull request, #19985:
URL: https://github.com/apache/hudi/pull/19985
### Describe the issue this Pull Request addresses
`HudiCacheKeyProvider` on `release-1.2.1` implements the Trino 483
`CacheKeyProvider` contract (restored in 4983f9a9a8c5 so the branch compiles at
483). After 483, `getCacheKey` returns `Optional<CacheKey>` instead of
`Optional<String>`. The erased method signature is the same, so a 1.2.1
`hudi-trino` jar still loads in a plugin assembled against a newer Trino (for
example a Trino-side shim built at 484), then fails with `ClassCastException`
at the first cached read.
### Summary and Changelog
Release-only change; master is on the new contract and keeps the binding.
- `HudiModule`: stop binding `HudiCacheKeyProvider`, so Trino's default
provider is used. A comment at the old binding explains why and when to restore
it.
- `TestHudiAlluxioCacheFileOperations`: `@Disabled` on
`testSelectWithFilter` and `testJoin`, with the reason in the annotation. They
assert the provider's exclusions, which no longer apply.
`testReadsServedFromAlluxioCache` still checks that reads are served from the
cache.
### Impact
Caching stays on. Trino's default provider keys files by path, last-modified
time and length, so cached reads add a metadata lookup per file, and table
properties and index definitions are now cached (a new last-modified time or
length invalidates them).
### Risk Level
low
<details><summary>Verification</summary>
- Ghost RC2 E2E runs (connector built at 483, server and plugin at Trino
head, file cache on):
- control with the binding (expected to fail): TBD
- without the binding (expected to pass): TBD
- Connector CI on this PR runs the remaining cache tests.
</details>
### Documentation Update
none
### Contributor's checklist
- [x] Read through [contributor's
guide](https://hudi.apache.org/contribute/how-to-contribute)
- [x] Enough context is provided in the sections above
- [x] Adequate tests were added if applicable
--
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]
