Philip, I've just committed the backend changes involved in setting up a
"default_tablespace" GUC variable for pg_dump to use, but I didn't
do anything to convert pg_dump to doing so instead of using explicit
TABLESPACE clauses. You had muttered something about wanting to add
a TOC entry field for this --- do you still want to do the work?
You can probably get it done faster than I could, but I dunno if you
have time at the moment. I'd like to get it in over the weekend so
that we can put out a new beta next week.
BTW, part of the backend changes was to stop emitting TABLESPACE
clauses in pg_get_indexdef() and pg_get_constraintdef() output,
so as of CVS tip pg_dump will in fact fail to restore index tablespaces
accurately. I assume this is the backend behavior you want, but
holler if not.
regards, tom lane