[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: wc_db performance

From: Stefan Sperling <stsp_at_elego.de>
Date: Tue, 15 Mar 2011 15:00:07 +0100

On Tue, Mar 15, 2011 at 12:52:42PM +0000, Philip Martin wrote:
> Yes, by the time we invoke the callbacks we have a read-lock. If we
> simply ran a single transaction and made callbacks within the
> transaction then it would have the same effect. I don't see what the
> temporary table achieves.

The temp tables live in a separate database.
Without them, lock contention is on a single database.
With them, the contention situation is different -- the callback cannot
read from or write to the temp table.

At least that is how I understand this text:
http://sqlite.org/lang_createtable.html: If the "TEMP" or "TEMPORARY" keyword
occurs between the "CREATE" and "TABLE" then the new table is created in the
temp database. It is an error to specify both a <database-name> and the TEMP or
TEMPORARY keyword, unless the <database-name> is "temp". If no database name is
specified and the TEMP keyword is not present then the table is created in the
main database.

So we use the temp table to create a separate database that contains
the data we need to present to the callback, and which is unaffected
by whatever else the callback might be doing.

Does that make sense?
Received on 2011-03-15 15:00:43 CET

This is an archived mail posted to the Subversion Dev mailing list.