Re: Temporary tables versus wraparound... again

On Fri, Apr 14, 2023 at 10:47 AM Andres Freund <andres@anarazel.de> wrote:
> I don't think it's outright wrong, but it is very confusing what it relates
> to. For some reason I tried to "attach" the parenthetical to the "otherwise",
> which doesn't make a whole lot of sense. How about:

I suppose that it doesn't matter whether it's outright wrong, or just
unclear. Either way it should be improved.

>  * If rel is not NULL the horizon may be considerably more recent (i.e.
>  * allowing more tuples to be removed) than otherwise. In the NULL case a
>  * horizon that is correct (but not optimal) for all relations will be
>  * returned. Thus, if possible, a relation should be provided.

That seems much better to me. The most important part is the last sentence.

The key idea is that you as a caller should provide a rel if at all
possible (and if not you should feel a pang of guilt). That emphasis
makes the potential consequences much more obvious.

--
Peter Geoghegan