Emmanuel Cecchet wrote:
> I just saw that this new patch was not considered because the previous
> version ended being rejected.
> Note that this version of the patch aims at supporting ONLY temp tables
> that are created AND dropped in the same transaction. We need to be able
> to use temp tables in transactions that are doing 2PC, but the temp
> table lifespan does not need to cross transaction boundaries.
> Please let me know if this patch could be integrated in 8.4.
IMHO, this is just getting too kludgey. We came up with pretty good
ideas on how to handle temp tables properly, by treating the same as
non-temp tables. That should eliminate all the problems the latest patch
did, and also the issues with sequences, and allow all access to temp
tables, not just a limited subset. I don't think it's worthwhile to
apply the kludge as a stopgap measure, let's do it properly in 8.5.
As a workaround, you can use a regular table instead of a temporary one.
If you create and drop the regular table in the same transaction (that's
the same limitation that latest patch has), you won't end up with a
bogus table in your database if the connection is dropped unexpectedly.
If your application uses multiple connections simultaenously, you'll
need a little bit of code in the application so that you don't try to
create a table with the same name in all backends. You could also create
a different schema for each connection, and do "set
search_path='semitempschemaX, public'", so that you can use the same
table name and still have separate tables for each connections.
(sorry for the late reply)
In response to
pgsql-hackers by date
|Next:||From: Simon Riggs||Date: 2008-12-23 16:48:42|
|Subject: Re: Synchronous replication, reading WAL for sending|
|Previous:||From: Kevin Grittner||Date: 2008-12-23 16:27:57|
|Subject: Re: incoherent view of serializable transactions|