Re: Access to transaction status - Mailing list pgsql-hackers

From Tom Lane
Subject Re: Access to transaction status
Date
Msg-id 17571.1056116411@sss.pgh.pa.us
Whole thread Raw
In response to Re: Access to transaction status  ("Christian Plattner" <postgresql@sioux.ch>)
List pgsql-hackers
"Christian Plattner" <postgresql@sioux.ch> writes:
> From: "Tom Lane" <tgl@sss.pgh.pa.us>
>> How much later?  clog is not kept forever.

> ... Ofcourse one should not do a
> VACUUM FULL while not being sure about the status of a transaction in the
> past :)

As long as you haven't done a cluster-wide VACUUM, clog status will not
get recycled.  For the application you're describing I think this will
work fine.

You might want to set up the API of the inquiry function to include
specified return codes for UNKNOWN (older than beginning of clog) and
FUTURE (greater than NextXid) as well as COMMITTED, ABORTED, and
INPROGRESS.  The current implementation can't easily give you UNKNOWN
(it'll error out instead) but any general-usage function of this kind
would have to offer that.
        regards, tom lane


pgsql-hackers by date:

Previous
From: "J.R. Nield"
Date:
Subject: Subtraction carry bug in xlog.c in 7.3 and 7.4
Next
From: Tom Lane
Date:
Subject: Re: Ownership change doesn't change privileges