From: | Richard Huxton <dev(at)archonet(dot)com> |
---|---|
To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
Cc: | David Fetter <david(at)fetter(dot)org>, PG Hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Immodest Proposal: pg_catalog.pg_ddl |
Date: | 2005-12-14 15:55:34 |
Message-ID: | 43A04076.9070805@archonet.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Tom Lane wrote:
> David Fetter <david(at)fetter(dot)org> writes:
>> On Tue, Dec 13, 2005 at 11:33:20PM -0500, Tom Lane wrote:
>>> and I don't even see the argument for doing it via a table rather
>>> than via the postmaster log.
>
>> Simple. Postmaster logs can roll over or otherwise be lost without
>> damaging the DB. This would provide a non-volatile log of DDLs.
>
> In that case you have to provide a pretty strong argument why everyone
> should be forced to have a non-volatile log of DDLs. Or will there be
> a way to turn it off? What about applications that, say, create and
> delete tens of thousands of temp tables every day?
What about system-event-driven triggers as a mechanism for this? That
should make it simple for people to extend how they wish - e.g.
- setup default preferences when new users are added
- setup temp.y tables at session start
- monitor ddl (as David wanted)
Now that we have sub-transactions, we could wrap the call to the trigger
function so that errors didn't abort the user setup/login etc.
There's been demand for this sort of thing fairly regularly - I'd
probably use it myself.
--
Richard Huxton
Archonet Ltd
From | Date | Subject | |
---|---|---|---|
Next Message | Christopher Browne | 2005-12-14 15:59:03 | Re: Cost-based optimizers |
Previous Message | Tom Lane | 2005-12-14 15:48:05 | Re: Interesting speed anomaly |