Thanks, Tom. It is done by modifying coerce_type() and
can_coerce_type(). The reason I have to keep to verson 7.3 is I am
working on a research prototype that is built over pgsql 7.3. I need
the extra functions provided by that prototype.
On 1/25/06, Tom Lane <[EMAIL PROTECTED]> wrote:
> andrew <[EMAIL PROTECTED]> writes:
> > Sorry, I modified the parser code and forgot abt it. Now there is no
> > problem in creating the function. But there is another problem. I
> > create a function to accept record type parameter. But when I call it
> > on a specific composite type, error is reported. The followings are
> > what I have done:
>
> > backend> create function complete(record) returns int4 as
> > '$libdir/qualityudf' language C
> > QUERY: create function complete(record) returns int4 as
> > '$libdir/qualityudf' language C
>
> > backend> select *, complete(Person) from Person
> > QUERY: select *, complete(Person) from Person
>
> > ERROR: Function complete(person) does not exist
>
> Hmm. Looking at parse_coerce.c, 8.1 is the first release that thinks
> named composite types can be coerced to RECORD. I think you may be
> forced to upgrade if you want this to work. Changing 7.3's coerce_type()
> to allow this case would be simple enough, but I think you are still
> going to be minus a lot of infrastructure that's required to make it
> actually do anything useful :-(
>
> regards, tom lane
>
--
andrew
---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match
