On Wed, 2003-03-12 at 10:02, Dave Page wrote:
> > -----Original Message-----
> > From: Dave Cramer [mailto:Dave@micro-automation.net]
> > Sent: 12 March 2003 14:55
> > To: Dave Page
> > Cc: Zeugswetter Andreas SB SD; Hiroshi Inoue; Tom Lane; Bruce
> > Momjian; pgsql-hackers@postgresql.org; pgsql-interfaces@postgresql.org
> > Subject: Re: [HACKERS] Roadmap for FE/BE protocol redesign
> >
> >
> > This may be problematic in the future if we change attrelid, attnum.
>
> I suggested using names to Tom for this reason, but he preferred to use
> attrelid/attnum.
Ok, but I am willing to bet that this will be a problem later. However
we can deal with it later, as well.
>
> > My preference would be to be able to query the backend for
> > the information if I need it. It is rarely required.
>
> Rarely required by most applications, but most interfaces do need this
> ability, and so do apps like pgAdmin and OpenOffice that allow the
> execution of arbitrary queries and subsequent updating of the results.
Yes, the interface requires it but how often? If I just do a select *
from foo, I don't need all the metadata, it's only if I do getColumnName
or something that I need metadata.
However qualified column names would be nice, as in the instance when
two column names are the same from different tables.
>
> Regards, Dave.
--
Dave Cramer <dave@fastcrypt.com>
Cramer Consulting
