mediprtl commented on PR #4320:
URL: https://github.com/apache/arrow-adbc/pull/4320#issuecomment-4476889921

   Added a bind-path regression test as a follow-up commit (b5a8d88). Context 
(cross-posted from apache#4319): `PostgresCopyListFieldWriter` is reused by 
`BindStream` via the same `MakeCopyFieldWriter` factory 
(`c/driver/postgresql/bind_stream.h:166`), so the offset bug also drifts 
list-typed parameters on `INSERT` / `UPDATE` / `DELETE` / upsert when the bound 
Arrow array is sliced.
   
   The new `PostgresStatementTest.BindUpsertWithSlicedListParameter` exercises 
the bind path end-to-end: binds rows 3..5 of a 6-row Arrow batch (with 
`parent.offset=3` on the struct, id and tags children) as the `text[]` 
parameter of an `INSERT ... ON CONFLICT DO UPDATE` upsert, then asserts 
`array_length(tags, 1)` per id matches the slice (1/2/1), not the 
underlying-rows-0..2 lengths (2/1/2). Fails on the unpatched driver at the 
first `len` assertion (id=3 reports length 2 — drift to row 0); passes with the 
writer.h fix from this branch.
   


-- 
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