At 14:36 +0300 on 8/7/98, Stuart Rison wrote:
> Yeah, I've rethough about it and it does seem a bit out-there ;) ... I'm
> sure there was one time where I encountered a situation where you would
> want to use the \d? commands in an SQL statement environment but I can't
> even begin to remember... so maybe it wasn't that important afterall and
> regex'ing will do just fine.
>
> I do feel though that the \df <regex> functionality would be useful in
> other catalog queries (thank james, I didn't know that what they were
> called) e.g. \do <regex> or \dt <regex> (which would get you all tables
> containing <regex> e.g. titles, tileauthors, titleditors etc, as opposed to
> \d titles which would get you just table titles).
In my (humble?) opinion, instead of having this as a specialized command in
psql, there should rather have been a read-only view in the system which
would make it possible to look at only part of the information.
This would make the information readily available from other interfaces
beside psql, and will not force the interface developers to rely on "inside
information" telling them which tables to query and what the internal oids
are to look for. Thus, if internal things like that are changed, the view
would stay alive and easily accessible.
So, instead of \d we would have something like SELECT * from SYSRELS_VIEW.
And instead of \dt, SELECT * from SYSRELS_VIEW WHERE relation_type =
'table' (which would imply that the view will have this in ascii, where the
internal table has it in some coding). Instead of \d my_table, we'll have
SELECT * from SYSCOLS_VIEW WHERE relation = 'my_table';
I think that's more or less how it's done in Oracle and the like, isn't it?
And it relies only on something we all know - SQL queries.
Herouth
--
Herouth Maoz, Internet developer.
Open University of Israel - Telem project
http://telem.openu.ac.il/~herutma
