Peter Eisentraut <peter_e(at)gmx(dot)net> writes:
> I was wondering whether, as a future project, we could make this more
> convenient by parsing the body of the function with the binding of the
> function already in effect.
Seems like a simple rearrangement of the code. First insert the pg_proc
entry, then CommandCounterIncrement, then do the parsing/checking of the
function body. Given the CCI, the new entry will be visible for the
checking --- and if we error out, it rolls back just fine anyway.
regards, tom lane