[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: Philip Martin <philip.martin_at_wandisco.com>
Date: Tue, 15 Mar 2011 14:18:51 +0000

Branko Čibej <brane_at_e-reka.si> writes:

> The temporary table is:
>
> * in a different database, so the read lock on it does not affect
> the main wc-db;
> * per-connection, so even the same process using a different
> database connection will not even see that temporary table.

OK, so the process is:

  - read lock on main database
  - write lock on temporary database
  - populate temporary table
  - release write lock
  - release read lock
  - read lock on temporary database
  - make callbacks
  - release read lock

without the temporary table it could be:

  - read lock on main database
  - make callbacks
  - release lock

What we gain is that the callback can modify the main database, because
there is no read lock. It still can't modify it using any functions
that require write access to the temporary database.

What we lose is that the callback cannot call any "read-only" functions
that use temporary tables because the step that requires a write lock
will fail.

-- 
Philip
Received on 2011-03-15 15:19:26 CET

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