On Aug 31, 2007, at 2:08 PM, Tom Lane wrote:
> Alvaro Herrera <alvherre@commandprompt.com> writes:
>> Mikko Partio escribió:
>>> Off-topic question: the documentation says that XID numbers are
>>> 32 bit.
>>> Could the XID be 64 bit when running on a 64 bit platform? That
>>> would
>>> effectively prevent wrap-around issues.
>
>> No, because they would take too much space in tuple headers.
>
> It's worth noting that the patch Florian is working on, to suppress
> assignment of XIDs for transactions that never write anything, will
> make
> for a large reduction in the rate of XID consumption in many real-
> world
> applications. That will reduce the need for tuple freezing and
> probably
> lessen the attraction of wider XIDs even more.
>
> If he gets it done soon (before the HOT dust settles) I will be
> strongly
> tempted to try to sneak it into 8.3 ...
>
> regards, tom lane
Off topic and just out of curiousity, is this the work that will
allow standby servers to have selects run on them without stopping
WAL replay?
Erik Jones
Software Developer | Emma®
erik@myemma.com
800.595.4401 or 615.292.5888
615.292.0777 (fax)
Emma helps organizations everywhere communicate & market in style.
Visit us online at http://www.myemma.com
