Re: problem with custom_variable_classes

hubert depesz lubaczewski <depesz(at)depesz(dot)com> writes:
> while manual in 8.4 still states:
> "When custom_variable_classes is set, the server will accept definitions of arbitrary variables within each specified class.".

It will accept *definitions*, yes. I can't imagine why you'd think
it's a good idea to not throw error when asked for an unknown variable's
value.

The whole custom-variable thing is being abused far beyond what the
facility was intended for, anyway. Rather than allowing variables to
spring into existence like magic, what we should have is some facility
for letting session-local variables be *declared*, complete with type
(int/real/string) and other info as needed. See the archives --- this
was discussed not too long ago.

regards, tom lane