On Fri, Mar 15, 2024 at 10:15 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:
Daniel Gustafsson <daniel@yesql.se> writes: > I can't see how refusing to free memory owned and controlled by someone else, > and throwing an error if attempted, wouldn't be a sound defensive programming > measure.
I think the argument is about what "refusal" looks like. An Assert seems OK to me, but anything based on elog is likely to be problematic, because it'll involve exit() somewhere.
Yeah, I agree an Assert seems safest here.
I'd like to get this done ASAP so I can rebase my incremental parse patchset. Daniel, do you want to commit it? If not, I can.