Heikki Linnakangas <hlinnaka@iki.fi> writes:
> Index tuples have commit info hint bits in them, don't they?
Oooh, I forgot that one, but that's definitely strike 5.
And there's a strike 6: I'm pretty sure this idea breaks unique-index
checking. Somebody else trying to insert a tuple with a duplicate key
value might have just looked at your tuple and concluded it was dead,
hence they could go ahead and insert their tuple ... but at the time you
are looking, they haven't managed to insert their index entry quite yet.
Even if you can avoid that race condition, it will certainly take a
second index search to catch the problem.
regards, tom lane