hcrosse opened a new pull request, #1561: URL: https://github.com/apache/arrow-go/pull/1561
### Rationale for this change When a bound field reference has the wrong type, `ExecuteScalarExpression` returns an error that names the wrong field or none. I bound `col` as `int64` over `int32` data at top level and `col.a` the same way inside a struct. Both schemas have a leading `int8` column `x`. On v18.8.0 and main, I got: ```text invalid: referenced field %!s(<nil>) was int32, but should have been int64 invalid: referenced field .field(0) was int32, but should have been int64 ``` The full references are `.field(1)` and `.field(1).field(0)`. Arrow C++ names the whole reference in the same error: https://github.com/apache/arrow/blob/976779b2aac587f51b87c3bd2e6ae4c36f75ab6c/cpp/src/arrow/compute/expression.cc#L763-L765 ### What changes are included in this PR? `execFieldRef` strips the first struct field segment to find the column, and the error formatted what was left. It now formats the original `e.Reference`. ### Are these changes tested? `TestFieldRefTypeMismatchReleases` gains a `wantMsg` column for its top level and nested rows. Both rows fail on main and pass with this change. With Go 1.26.10, vet and race tests for `arrow/compute/...`, `golangci-lint` 2.13.2 and the Release Audit Tool passed. ### Are there any user-facing changes? Yes. Type mismatch errors now name the full field reference. There is no API change. -- 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]
