Re: Fixing a few minor misusages of bms_union()

Greg Burd <greg(at)burd(dot)me> writes:
> On Oct 3 2025, at 10:04 am, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> This change in substitute_phv_relids_walker is *not* safe according
>> to the routine's head comment:

> I'll have to remember to scroll up a bit more when reviewing and always
> read the header comments. I missed that one entirely, apologies. When I
> read the bitmapset_del() below the bitmapset_union() I incorrectly
> assumed that it was okay to modify it in-place. Maybe a short comment
> above that line would be useful?

After we've done the union(), we know we have a singly-referenced
bitmapset, so it's safe for the second change to be in-place.

regards, tom lane