yihua opened a new pull request, #912:
URL: https://github.com/apache/hudi-rs/pull/912
## Description
closes #910
Stacked on #894, review that first; the diff against `main` shows both until
it merges.
`HFileWriter::write` now appends each structured key's parts (the `u16`
length, the content, the family and timestamp bytes, the type byte) straight
into the data block buffer instead of building a `Vec` per key and copying it
in. The bytes written are the same, and the 64 KiB key check still rejects the
same keys with the same error. On a 10M-key record index build this removes one
allocation and one copy per entry.
## How are the changes test-covered
- [ ] N/A
- [x] Automated tests (unit and/or integration tests)
- [ ] Manual tests
- [ ] Details are described below
The hfile writer and reader suites pass unchanged; a probe comparing the
output of this writer with the one on `main` for the same entries (empty,
single, 2,000 mixed entries with repeated and multi-byte keys, keys at the 64
KiB boundary, and the oversized error cases) found the files byte-identical and
the errors the same.
--
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]
