Re: [HACKERS] getting rid of SnapshotNow

Andres Freund <andres@2ndquadrant.com>

From: Andres Freund <andres@2ndquadrant.com>
To: "Inoue, Hiroshi" <inoue@tpf.co.jp>
Cc: Robert Haas <robertmhaas@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, pgsql-hackers@postgresql.org, pgsql-odbc@postgresql.org
Date: 2013-07-19T13:03:30Z
Lists: pgsql-hackers
On 2013-07-19 08:57:01 +0900, Inoue, Hiroshi wrote:
> >>I had the idea they were used for a client-side implementation of WHERE
> >>CURRENT OF.  Perhaps that's dead code and could be removed entirely?
> >
> >It's been reported that ODBC still uses them.
> 
> Though PostgreSQL's TID is similar to Orale's ROWID, it is transient
> and changed after update operations unfortunately. I implemented
> the currtid_xx functions to supplement the difference. For example
> 
> 	currtid(relname, original tid)
> 
> (hopefully) returns the current tid of the original row when it is
> updated.

That is only guaranteed to work though when you're in a transaction old
enough to prevent removal of the old or intermediate row versions. E.g.
BEGIN;
INSERT INTO foo...; -- last tid (0, 1)
COMMIT;
BEGIN;
SELECT currtid(foo, '(0, 1'));
COMMIT;

can basically return no or even an arbitrarily different row. Same with
an update...

Greetings,

Andres Freund

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