On Tue, Dec 16, 2025 at 4:49 PM Tom Lane <[email protected]> wrote:
> Alternatively: I don't see any really good reason for a check_hook
> to be setting other GUCs, in fact it's probably a seriously bad idea.
> (It's a *check* hook, it's not supposed to be causing any
> side-effects.) Therefore, there can be at most one of these
> operations in flight at a time, so you don't need any dynamic data
> structure. A simple static variable remembering a not-yet-reparented
> context would do it.
Oh, yeah, I actually wondered if that would be an acceptable
restriction and had it in an earlier version of the email, but it got
lost in the final draft. Maybe with this design you just do something
like:
if (TempCheckHookConteck != NULL)
MemoryContextReset(TempCheckHookConteck);
else
TempCheckHookConteck = AllocSetContextCreate(...);
So then if the context survives, you just reset and reuse it, but if
it gets reparented, you set the variable to NULL and create a new
context the next time. Then you don't need any integration with
(sub)transaction abort at all, which seems nice.
--
Robert Haas
EDB: http://www.enterprisedb.com
