"Jeroen T. Vermeulen" <[EMAIL PROTECTED]> writes:
> It does require that the application be meticulous in its checking though.
> Existing client programs, for instance, may ignore any errors coming back
> from PQexec() during the transaction and just see if the COMMIT succeeds.
> Such could would break in very nasty ways with this change.
I think it's a given that the *default* behavior will not change.
You'll have to do something --- at least set a SET variable --- to get
intratransaction error recovery to behave differently. The risk of
breaking existing clients in subtle ways is too great if we do
otherwise.
regards, tom lane
---------------------------(end of broadcast)---------------------------
TIP 4: Don't 'kill -9' the postmaster
