On Tue, Feb 13, 2007 at 05:55:11PM -0500, Bruce Momjian wrote:
>
> Is there a TODO here?
>
> ---------------------------------------------------------------------------
>
> Jim Nasby wrote:
> > On Oct 2, 2006, at 6:28 PM, Tom Lane wrote:
> > > "Jim C. Nasby" <jim@nasby.net> writes:
> > >> However, the test right above that means that we'll fail if the user
> > >> tries something like "row_variable := NULL;":
> > >
> > > The patch you seem to have in mind would allow
> > > row_variable := int_variable;
> > > to succeed if the int_variable chanced to contain NULL, which is
> > > surely
> > > not very desirable.

Well, that's Tom's objection, though I'm not sure if by 'int_variable'
he means 'internal' or 'integer'.

Personally, I think it would be useful to just allow setting a row or
record variable to NULL as I showed it above; ie: no variables involved.
This is something you might want to do to invalidate a row/record
variable after taking some action (perhaps deleting a row).

You'd also think that you should be able to detect if a record variable
is null, as you can with row.

So, I suggest:

* Allow row and record variables in plpgsql to be set to NULL

It's not clear if it's a wise idea to allow this assignment from a
variable. It may be better to only allow explicitly setting them,
ie:

row_variable := NULL;

* Allow testing a record variable to see if it's NULL

Currently works for row variables, but not record variables
--
Jim Nasby jim@nasby.net
EnterpriseDB http://enterprisedb.com 512.569.9461 (cell)

---------------------------(end of broadcast)---------------------------
TIP 2: Don't 'kill -9' the postmaster