kamcheungting-db opened a new pull request, #976:
URL: https://github.com/apache/iceberg-cpp/pull/976
## Summary
This is PR 1 of the split, grouped, and parallel table-scan feature. A
FileScanTask can now represent a validated half-open byte range. Whole-file
tasks keep their existing constructor and behavior.
For a file of 10 bytes, two tasks can cover it without overlap:
```json
{"start": 0, "length": 4}
{"start": 4, "length": 6}
```
Core and REST JSON preserve both endpoints. REST defaults to a whole-file
task only when both fields are absent; a lone endpoint or invalid range is a
parse error. Split tasks share their data-file, delete-file, residual, and
lineage context. Range sizes and estimated row counts no longer count the full
file for every split.
Partial reads are **not enabled in this PR**. FileScanTaskReader rejects a
partial task explicitly until PR 2 passes ranges to format readers; it cannot
silently reread the full file.
## Stack
This PR depends on #940, which is the last PR in the existing serde stack:
#974 → #975 → #939 → #940 → **this PR**. It is based on the #940 head. Like the
existing stack, this PR targets `main` and therefore includes preceding stack
commits; review the new commit `2014413f` for this PR. The planned follow-ups
add range-aware reading, lazy split/group planning, the bounded parallel
reader, and end-to-end performance validation.
## Testing
- `json_serde_test`, `scan_test`, `rest_catalog_test`, and `data_test`: all
passed.
- `clang-format --dry-run --Werror` on all changed C++ files and `git diff
--check`: passed.
- The repository `pre-commit` wrapper could not be installed locally because
PyPI DNS resolution failed; the available direct formatting and whitespace
checks passed.
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
