On Wed, 16 Mar 2005 05:17 am, Dave Cramer wrote:
> Shachar,
>
> I think with type oid 705 (unknown) it's safe to treat it as text.
> Certainly better than punting.
>
> On another note are you aware of any issues with transactions?
> Specifically with using the dated autocommit mode ?
>
> Dave
>
> Shachar Shemesh wrote:
>
> > Tom Lane wrote:
> >
> >> Dave Cramer <pg@fastcrypt.com> writes:
> >>
> >>
> >>> I just had a customer complain about this. My understanding is that
> >>> unkown is a constant which has not been typed yet. Is it possible
> >>> for it to be a binary type, if so how ?
> >>> I would think it should only ever be a string?
> >>>
> >>
> >>
> >> You can read "unknown" as "string literal for which the query does not
> >> provide sufficient context to assign a definite type". I dunno what the
> >> OLE driver really needs to do with the datatype info, but I suppose that
> >> treating this as equivalent to "text" is not unreasonable. Ask the
> >> complainant what *he* thinks it should do.
> >>
> >> regards, tom lane
> >>
> >>
[snip]
> >
> > On the good news front, Version 1.0.0.17 is about ready to be released
> > (initial schema support). I am resuming development after about half a
> > year of doing other stuff.
> >
> > Shachar
> >
>
I have complained about this in the past, and would also suggest that it be treated as a
string value.
CREATE table b AS SELECT 'unknown', col2 from a;
Will even create a table with a column type as unknown, which doesn't have any operators
to convert to anything, including text.
Regards
Russell Smith.
