Tim Kientzle wrote:
> [...]
>
> Basically, the PostgreSQL developers have decided
> that any integrity violation is a serious error;
> therefore, PostgreSQL does not really permit
> tentative INSERTs within transactions. This violates
> SQL conventions that are pretty well-established
> in some circles, needlessly complicates
> applications that use complex transactions
> and introduces a fairly minor performance issue.
We haven't, it's just that our code inherited this feature by
the way, ROLLBACK is implemented - and if you'd know only a
little about the storage management in PostgreSQL, you'd
never tell it the way you did. With the current
implementation of tuple visibility there is only all-or-
nothing. Savepoints will get us out of there, but not before
7.2 or 7.3.
Jan
--
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#================================================== JanWieck@Yahoo.com #
