Re: Feature Request - Table Definition query - Mailing list pgsql-hackers

From Svenne Krap
Subject Re: Feature Request - Table Definition query
Date
Msg-id 49081CC8.4010008@krap.dk
Whole thread Raw
In response to Feature Request - Table Definition query  (Wilfried Schobeiri <schobeiriw@videofurnace.com>)
List pgsql-hackers
> It could be useful to have a command that returns the table definition 
> (like pg_dump -st) from within the query interface. This could be 
> particularly useful if one doesn't have access to or prefers not to 
> manipulate results coming from a shell. If I have an API from which to 
> query the database, it seems silly to have to spawn a shell and run a 
> command to query the very same database.
Yes. That functionality would be really appreciated by myself and a few 
pgsql-users I know.
> (I'm not exactly familiar with how pg_dump manipulates or extracts 
> that data, so this could be a very, very stupid question. If so, 
> please forgive me.)
Me neither, but I think one of the problems is, that during upgrade you 
are supposed to use pg_dump from the new version on the existing 
database instance.
The server could most probably only generate its own format (not the 
new) which might be a problem.

Perhaps it could be implemented by stored function (autoinstalled and 
available in contrib or perhaps installed by a psql on first run) in a 
namespace like pg_dump_84 (i.e. tagged with server version). If you want 
to upgrade you could then install the new set of functions..
Psql could then be a thin wrapper (which calls the needed functions).

/Svenne



pgsql-hackers by date:

Previous
From: "Pavel Stehule"
Date:
Subject: some problem with casting unknown to smallint
Next
From: Simon Riggs
Date:
Subject: Re: Proposal of PITR performance improvement for 8.4.