> > It seems like it would be so much more useful if the timestamp with > > time zone type actually stored the time zone in the record. > > This has been requested before, and it would be closer to the intention > of the SQL standard, but I guess it won't happen. > > For one, it would change on-disk storage, which would make it > impossible to use pg_upgrade. It also would require timestamps to > occupy more than 8 bytes. > > The best solution is probably to explicitly store the time zone as > an additional column. Would that not be a perfect candidate for a fully fleshed out, exemplary composite type ? (but, then, yes it would beg the question what the TZ field of the composite is to actually mean...) Karsten
