Re: How to check whether the row was modified by this transaction before?

From: Andres Freund <andres(at)2ndquadrant(dot)com>
To: Alvaro Herrera <alvherre(at)2ndquadrant(dot)com>
Cc: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Vlad Arkhipov <arhipov(at)dc(dot)baikal(dot)ru>, pgsql-hackers(at)postgresql(dot)org
Subject: Re: How to check whether the row was modified by this transaction before?
Date: 2012-12-06 18:39:37
Message-ID: 20121206183937.GB20926@awork2.anarazel.de
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On 2012-12-06 15:08:51 -0300, Alvaro Herrera wrote:
> Tom Lane wrote:
> > Vlad Arkhipov <arhipov(at)dc(dot)baikal(dot)ru> writes:
> > > In a BEFORE UPDATE trigger I need to know whether the row was previously
> > > modified by this transaction. Is it safe to use xmin and txid_current()
> > > for this purpose (xmin is 32-bit txid type but txid_current() returns
> > > 64-bit bigint).

I wonder if we shouldn't have a function txid_is_current(xid); I could
have used that previously to avoid queuing multiple external
cache-invalidations if something gets repeatedly updated in the same
transaction. And I seem to remember some people here asking this
question before on the lists.

> >
> > > IF OLD.xmin = txid_current() THEN
> >
> > Comparing to txid_current() mod 2^32 would probably work,
>
> I think we should be setting the initial epoch to something other than
> zero. That way, some quick testing would have revealed this problem
> immediately.

+1, currently the difference of xid vs bigint is hard to spot.

Greetings,

Andres Freund

--
Andres Freund http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Simon Riggs 2012-12-06 18:42:22 Re: ALTER TABLE ... NOREWRITE option
Previous Message Tom Lane 2012-12-06 18:36:47 Re: why can't plpgsql return a row-expression?