Re: Lock detection and resolution

From: Nuno Souto <nsouto_at_nsw.bigpond.net.au.nospam>
Date: Sat, 26 May 2001 02:56:02 GMT
Message-ID: <3b0f17ee.5168444@news-server>

On Fri, 25 May 2001 16:05:12 -0400, Thomas Kyte <tkyte_at_us.oracle.com> wrote:

>
>No, and even if there was -- it wouldn't tell you anything. It could tell you
>that as 12:01:01.000 the row was not locked -- but at 12:01:01.001 (a split
>instant later) the row could become locked. Asking "is row X locked" without
>locking it doesn't tell you anything. The split INSTANT after you ask the
>question -- the answer can change.
>

Exactly. That was the problem too with the Codasyl databases. There is no way of ensuring a lock will stay put after we ask if it's there.

The only safe way of handling this in any database architecture is to ask for a lock. It succeeds, all is well. It doesn't, we either queue (WAIT) or return a status (NOWAIT).

I suppose a case could be made for a "register intent of locking", but I can hardly see how that would improve things, let alone the overhead this would cause.

Cheers
Nuno Souto
nsouto_at_bigpond.net.au.nospam
http://www.users.bigpond.net.au/the_Den/index.html Received on Fri May 25 2001 - 21:56:02 CDT