mrhhsg opened a new pull request, #64563:
URL: https://github.com/apache/doris/pull/64563
### What problem does this PR solve?
Issue Number: None
Problem Summary: Nested `array_map` expressions can capture variables from
an outer lambda while also defining their own lambda arguments. The BE
previously assigned lambda argument column offsets by recursively mutating
every `VColumnRef` under the outer lambda. This could make an inner lambda
argument reuse the outer lambda block offset and read the wrong column during
`INSERT ... SELECT`, causing a BE crash. This change serializes lambda argument
metadata to BE, preserves captured input columns in lambda blocks, and assigns
offsets only to the current lambda's formal arguments while respecting
nested-lambda shadowing.
### Release note
Fix a BE crash for nested `array_map` expressions that capture outer lambda
variables.
### Check List (For Author)
- Test:
- Manual test: `./build.sh --be --fe`
- Regression test: `./run-regression-test.sh --conf
output/local-regression/regression-conf-19700.groovy --run -d
query_p0/sql_functions/array_functions -s test_nested_array_map_insert
-forceGenOut`
- Regression test: `./run-regression-test.sh --conf
output/local-regression/regression-conf-19700.groovy --run -d
query_p0/sql_functions/array_functions -s test_nested_array_map_insert`
- Format check: `./build-support/clang-format.sh`;
`./build-support/check-format.sh`; `git diff --check`
- Static analysis: attempted `./build-support/run-clang-tidy.sh
--build-dir be/build_Release`, but local clang-tidy analysis was blocked by
pre-existing/toolchain diagnostics outside this change.
- Behavior changed: No
- 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]
