Gregory Stark <stark@enterprisedb.com> writes:
> "Tom Lane" <tgl@sss.pgh.pa.us> writes:
>> Actually, rather than get into that sort of AI-complete project,
> Is it really AI-complete? ISTM the *only* place where a parameter is allowed
> is where the parser inserts a ColumnRef node?
The first problem is that we have to make the decision far upstream of that.
The second is that whether we *can* insert a parameter symbol is not the
same as whether we *should*. Even if we made that work, we'd only be
fixing the cases that currently result in trivial-for-users-to-recognize
syntactic errors. The deeper semantic errors arising from substituting
a parameter where the user meant to refer to a similarly-named column
can only be dealt with through user education.
regards, tom lane