Martijn van Oosterhout <kleptog@svana.org> writes:
> OK, I've changed the vacuum code now so your index doesn't get
> departialised. The changes seem pretty obvious so they're probably right.
> They certainly didn't seem to break my simple tests. How does one test that
> VACUUM works properly?
Since I'm just about to start doing some rearrangements of the VACUUM
code, I'll first go ahead and apply the changes to make VACUUM use
ExecOpenIndices. I've wanted to make that change for quite awhile,
independently of partial-index considerations, but had not gotten around
to doing it.
The rest of this should probably be sent to pgsql-patches for more
review.
regards, tom lane