Re: Is NEW.ctid usable as table_tuple_satisfies_snapshot?

Kaiting Chen <ktchen14(at)gmail(dot)com> writes:
> On Fri, May 26, 2023 at 11:34 AM David G. Johnston <
> david(dot)g(dot)johnston(at)gmail(dot)com> wrote:
>> On Fri, May 26, 2023 at 8:04 AM Kaiting Chen <ktchen14(at)gmail(dot)com> wrote:
>>> 2. If I lookup the row by its ctid, will the visibility map be consulted.

>> No, but that doesn't seem to be material anyway. Your user-space pl/pgsql
>> function shouldn't care about such a purely performance optimization.

It'd be a waste of cycles to consult the map in this usage, since the
tuple of interest is surely not all-visible and thus the page couldn't
be either.

> Just to clarify, there's no way for SELECT FROM foo WHERE ctid = NEW.ctid
> to return a row that ordinary wouldn't be visible right? There's no magic
> going on with the qual on ctid that skips a visibility check right?

No, a ctid test isn't magic in that way; nodeTidscan.c applies the
same snapshot check as any other relation scan.

regards, tom lane