mrhhsg opened a new pull request, #63430:
URL: https://github.com/apache/doris/pull/63430
### What problem does this PR solve?
Issue Number: None
Related PR: None
Problem Summary: LINEAR_HISTOGRAM used raw floating-point arithmetic both
when assigning values to bucket indexes and when emitting bucket boundaries.
Decimal intervals such as `0.1` could therefore place a value exactly on a
decimal boundary into the previous bucket and expose artifacts such as
`0.30000000000000004` in the JSON result. This patch normalizes bucket indexes
that are within floating-point tolerance of an integer boundary and rounds
emitted boundaries only when the interval and offset have an inferable decimal
scale, preserving the previous behavior for intervals that cannot be
represented as a bounded decimal scale.
### Release note
Fix LINEAR_HISTOGRAM JSON output for decimal bucket boundaries.
### Check List (For Author)
- Test:
- Unit Test: `./run-be-ut.sh --run --filter=AggLinearHistogramTest.*`
- Regression test: `./run-regression-test.sh --conf
output/local-regression/regression-conf-46001.groovy --run -d
feature_review_repro/agg_function -s AGG-HIST-001`
- Format check: `build-support/check-format.sh`
- Static analysis: `build-support/run-clang-tidy.sh --build-dir
be/ut_build_ASAN` attempted, but clang-tidy could not analyze the changed files
in this local environment because the toolchain reported missing `stddef.h` and
a pre-existing unmatched `NOLINTEND` in `be/src/core/types.h`.
- Behavior changed: Yes. Decimal bucket boundaries no longer expose
floating-point artifacts and boundary values are assigned to the expected
bucket.
- Does this need documentation: No
--
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]
