On Mon, Oct 18, 2010 at 3:26 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Andres Freund <andres@anarazel.de> writes:
>> On Monday 18 October 2010 20:06:01 Tom Lane wrote:
>>> Not unless you have some credible concept for how it might ever be
>>> implemented. You can't create temp tables because you can't modify
>>> system catalogs, and if you did somehow create them you couldn't put
>>> anything in them because you can't generate XIDs on a slave ... much
>>> less commit them. We have talked about ways that temp tables might be
>>> created without touching the "real" system catalogs, but the XID issue
>>> seems a complete showstopper.
>
>> Hm. Wouldnt it be possible to use virtual xids for that purpose? They are
>> never seen outside of that session anyway...
>
> Well, maybe, but then you need infrastructure to track whether VXIDs
> committed or aborted.
Seems like this would wreak havoc with the HeapTupleSatisfies* functions.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company