XiaoHongbo-Hope opened a new pull request, #1393:
URL: https://github.com/apache/paimon-rust/pull/1393

   ## What
   
   Decode a validated Variant object offset table in bulk, dispatching once on 
its 1–4 byte width instead of calling the checked, byte-at-a-time decoder for 
every offset.
   
   The existing root-size, offset-order, field-layout and selected-value 
validation remains unchanged. No format, API, configuration or dependency 
changes.
   
   ## Validation
   
   - Tests cover all four widths, empty objects, numeric and cast extraction, 
repeated projection, non-monotonic offsets, truncation, duplicate offsets, 
oversized offsets and trailing payload bytes.
   - `cargo fmt --all` and `git diff --check`.
   - `cargo test --release -p paimon --lib variant::` is running; results will 
be updated before marking ready.
   
   ## Performance
   
   An earlier-base microbenchmark (`fb5e8f7`, Arrow 58) exercised the 
production `variant_get_numeric_fields` entry point on resident Arrow input: 
146 rows, 200 projected fields, 5,569 fields per Variant. Three independent 
processes with alternating baseline/candidate rounds measured median process 
CPU of 12.21–12.40 ms versus 7.99–8.06 ms (34–35% less CPU). Complete Arrow 
outputs matched.
   
   This measures numeric extraction only, excluding Parquet decoding, I/O and 
planning. It is not an end-to-end throughput claim or a performance measurement 
of the latest main dependency set. No customer fixture is included.
   


-- 
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