From: | "Michael Shulman" <shulman(at)mathcamp(dot)org> |
---|---|
To: | "Richard Broersma" <richard(dot)broersma(at)gmail(dot)com> |
Cc: | pgsql-general(at)postgresql(dot)org |
Subject: | Re: inserting to a multi-table view |
Date: | 2008-06-17 19:34:55 |
Message-ID: | c3f821000806171234w37963edaheff9d468cb2b2853@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
On Tue, Jun 17, 2008 at 11:50 AM, Richard Broersma
<richard(dot)broersma(at)gmail(dot)com> wrote:
> While your individual update rules are firing for each of your tables
> from T1 thru T[n] to change your OLD row to NEW. Another client could
> also at the same time be updating any of the other tables before and
> after your update Rules take affect. The net result is that, some of
> what you've changed could over write what the other client commited.
> And some of what you've commited could be over written by what the
> other client wrote. The end result is that the view's virtual "row"
> appears to be left in an inconsistant state.
Got it; thanks.
> Basically what you want to achieve is something like:
>
> begin:
> Select for update table T1 where id = old.id;
> Select for update table T2 where id = old.id;
> Select for update table (...) where id = old.id;
> Select for update table T[n-1] where id = old.id;
> Select for update table T[n] where id = old.id;
> if all the needed row lock are aquired, then
> begin the updates
> else rollback
> commit;
Would it be possible to actually do something like this in an update
rule? You couldn't write the "begin/commit", but it seems that you
wouldn't need to either, since the UPDATE command invoking the rule
will be wrapped in its own begin/commit (automatic or explicit).
Mike
From | Date | Subject | |
---|---|---|---|
Next Message | Richard Broersma | 2008-06-17 19:46:27 | Re: inserting to a multi-table view |
Previous Message | Richard Broersma | 2008-06-17 16:50:17 | Re: inserting to a multi-table view |