Re: RIGHT/FULL OUTER hash joins (was Re: small table left outer join big table)

Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> On Thu, Dec 30, 2010 at 10:45 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> ... But we only need one bit, so what about commandeering
>> an infomask bit in the tuple itself? For the initial implementation
>> I'd be inclined to take one of the free bits in t_infomask2. We could
>> actually get away with overlaying the flag bit with one of the tuple
>> visibility bits, since it will only be used in tuples that are in the
>> in-memory hash table, which don't need visibility info anymore. But
>> that seems like a kluge that could wait until we really need the flag
>> space.

> I think that's a reasonable approach, although I might be inclined to
> do the overlay sooner rather than later if it doesn't add too much
> complexity.

Well, there's no "complexity" involved, it's just which bit do we define
the macro as. Any complexity is conceptual.

regards, tom lane