aboutsummaryrefslogtreecommitdiff
path: root/src/backend/tcop/postgres.c
diff options
context:
space:
mode:
authorTom Lane <tgl@sss.pgh.pa.us>2025-03-21 15:55:06 -0400
committerTom Lane <tgl@sss.pgh.pa.us>2025-03-21 15:55:06 -0400
commit16a3ae504e2030583851c8f45cbed14b0c546d6e (patch)
tree93dbd5fffe6134bb11286eb21aaad4f82143861b /src/backend/tcop/postgres.c
parent04ff636cbce4b91fba1f334e1bc0dc88686e7b2d (diff)
downloadpostgresql-16a3ae504e2030583851c8f45cbed14b0c546d6e.tar.gz
postgresql-16a3ae504e2030583851c8f45cbed14b0c546d6e.zip
Revert inappropriate weakening of an Assert in plpgsql.
Commit 682ce911f modified exec_save_simple_expr to accept a Param in the tlist of a Gather node, rather than the normal case of a Var referencing the Gather's input. It turns out that this was a kluge to work around the bug later fixed in 0f7ec8d9c, namely that setrefs.c was failing to replace Params in upper plan nodes with Var references to the same Params appearing in the child tlists. With that fixed, there seems no reason to continue to allow a Param here. (Moreover, even if we did expect a Param here, the semantically correct thing to do would be to take the Param as the expression being sought. Whatever it may represent, it is *not* a reference to the child.) Hence, revert that part of 682ce911f. That all happened a long time ago. However, since the net effect here is just to tighten an Assert condition, I'm content to change it only in master. Discussion: https://postgr.es/m/1565347.1742572349@sss.pgh.pa.us
Diffstat (limited to 'src/backend/tcop/postgres.c')
0 files changed, 0 insertions, 0 deletions