Peter Eisentraut <peter_e@gmx.net> writes:
> Tom Lane writes:
>> Add GUC parameter check_function_bodies to control whether validation
>> of function bodies is done at CREATE FUNCTION time.
> If it's only honored by SQL functions, then it should probably be called
> check_sql_function_bodies.
I thought about that while I was making the patch, but decided that it
would be a very un-forward-looking name. Someday we will probably have
syntax-checking validators for plpgsql, etc.
The original version of the patch actually suppressed calling the
validator altogether, but I soon realized that wouldn't do --- it
would allow you to create SQL functions with unsupported pseudotype
arguments or results, for example. Further thought led me to decide
explicitly not to suspend checking of internal and C function
references, on the grounds that if they are broken they'd still be
broken at the completion of the restore script, and so we'd only
be losing the ability to report the problem.
So the fact that it only affects SQL functions at the moment is IMHO
just happenstance; the scope of what it does will change as we add more
validation capability.
regards, tom lane