Hi all,

I tested the v2 patch for the issue related to incorrect varnullingrels
handling when executing MERGE on a trigger-updatable view.

I was able to reproduce the reported error before applying the patch. After
applying the patch and rebuilding PostgreSQL, I ran the same test again,
and the MERGE statement completed successfully without the previous wrong
varnullingrels error.

The output was:

MERGE 0

I also reviewed the code changes in prepjointree.c. The patch appears to
address the issue by ensuring that the nulling-relid information is updated
consistently across the relevant MERGE expressions.

I ran the MERGE regression tests as well, and all tests passed.

Based on the reproduction, code review, and regression test results, the
patch looks good to me.

Regards,

solai

Reply via email to