Re: BETWEEN Node & DROP COLUMN

Tom Lane wrote:
> Rod Taylor <rbt@zort.ca> writes:
> >> We could change pg_attribute to another name, and create a view called
> >> pg_attribute that never returned isdropped columns to the client.  That
> >> would allow clients to work cleanly, and the server to work cleanly.
> 
> > Another case where having an informational schema would eliminate the
> > whole argument -- as the clients wouldn't need to touch the system
> > tables.
> 
> This is a long-term solution, not a near-term one.  I suspect it's
> really unlikely that pg_dump, pgAdmin, etc will ever want to switch
> over to the SQL-standard informational schema, because they will want
> to be able to look at Postgres-specific features that are not reflected
> in the standardized schema.  Certainly there will be no movement in
> that direction until the informational schema is complete; a first-cut
> implementation won't attract any interest at all :-(
> 
> I thought about the idea of a backward-compatible pg_attribute view,
> but I don't see any efficient way to generate the consecutively-numbered
> attnum column in a view; anyone?

No, we can't, and because our client coders want consecutive, it is a
dead idea.  Even if we could do it, we would be feeding clients attno
values that are inaccurate, causing problems when attno is joined to
other tables.

--  Bruce Momjian                        |  http://candle.pha.pa.us pgman@candle.pha.pa.us               |  (610)
853-3000+  If your life is a hard drive,     |  830 Blythe Avenue +  Christ can be your backup.        |  Drexel Hill,
Pennsylvania19026