Leon <leon@udmnet.ru> writes:
> Tom! I tested your method of creating table with
> create table ww (aa int4, bb timestamp default text 'now'),
> and it didn't work either! (BTW, this is exactly the way docs suggest
> doing it.) See? I promised to deliver a real bug and I did it! :)))
By golly, you're right. It works as advertised for a DATETIME field,
which is the case I'd been testing. But for a TIMESTAMP field the
constant gets pre-coerced anyway :-(. Wonder why ... will look into
it, since I'm busy hacking on that part of the system now.
> Yes, docs mumble something about 'cacheable' and 'non-cacheable'
> functions, but it is not clear to me how Postgres discerns them.
The proiscachable field in table pg_proc is presumably supposed to
tell this. Doesn't look like it's set in an intelligent manner
for most of the built-in functions though. I don't think it has
anything to do with the bug for TIMESTAMP...
regards, tom lane