Tom Lane wrote:
>
> inoue@postgresql.org writes:
> > Removed a check for REINDEX TABLE.
>
> As I commented before, what is the point of forbidding REINDEX INDEX
> if you are going to allow REINDEX TABLE? Or REINDEX DATABASE for that
> matter? You may as well strip out the corresponding checks in the
> other two cases too.
Simply because I've never tested REINDEX INDEX/DATABASE case
under postmaster and I'm afraid of the following.
1) I didn't make any device to not look up the reindexing
index itself in case of REINDEX INDEX.
2) REINDEX DATABASE generates multiple transactions internally.
There's no such command other than VACUUM.
regards,
Hiroshi Inoue