Since the message didn't come back to me via the list (I sent it last
Tuesday), I resend it, this time to pgsql-hackers, because I think the
discussion doesn't belong to PATCHES.
Bruce Momjian wrote:
> Applied. Thanks. You can report any ecpg problems to the bugs list.
Ok, I'll flood the list ;-)
WARNING: My patch breaks existing code!
If an ecpg program did not provide an indicator variable the library set
the variable to zero.
Now it will return an error.
I would also like to address a remaining problem:
should ecpglib touch the host variable if the result is NULL and an
indicator
variable is given?
Ecpglib so far zeroed the variable, my patch doesn't touch the bool
variable.
What does the standard say about this? (Adabas e.g. doesn't touch the
variable.)
Christof