Tom Lane <tgl@sss.pgh.pa.us> writes:
> Greg Stark <gsstark@mit.edu> writes:
> > Here's a small patch to add the full suite of btree operators for tids and the
> > corresponding btree opclass.
>
> This has been proposed and rejected before, mainly on the basis that
> there's no conceivable application for an index on TID. What's your
> use case?
IIRC the use case that was previously suggested was the common Oracle idiom
for removing duplicates. The Postgrse equivalent would something like:
delete from tab where exists (
select 1 from tab as x where x.pk = tab.pk and x.ctid > tab.ctid
)
In my case I don't want to create an index on tid, I just have a list of tids
that I want to sort so I can process them in heap order.
I was going to just write a qsort comparator but this way I can use tuplesort
and not worry about what happens if it doesn't fit in memory.
--
greg