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

Re: svn commit: r25869 - in trunk/subversion: include/private libsvn_fs_fs tests/libsvn_fs

From: Blair Zajac <blair_at_orcaware.com>
Date: 2007-09-19 16:04:08 CEST

On Sep 19, 2007, at 5:31 AM, Malcolm Rowe wrote:

> Hi Blair,
>
> With respect to the FSFS transaction-current code you commited a while
> ago...
>
> On Fri, Jul 27, 2007 at 07:22:48PM -0700, blair@tigris.org wrote:
>> (get_and_increment_txn_key_body):
>> New function used as a callback for svn_fs_fs__with_write_lock().
>> This gets the current base 36 value in transaction-current and
>> increments it. It returns the original value by the baton.
>
> I'm curious at to whether there was any reason that you chose to
> use the
> fs-wide write-lock to update the file rather than locking the file
> directly? If not, is there any reason we shouldn't change it to do
> so?

No reason. Originally, I was going to have a transaction-
current.lock file and lock that, but after discussion on the list and
a test by Peter, IIRC, that showed you can get 70,000 locks and
unlocks on a single file, we decided to use the fs-wide lock.

Didn't think of locking transaction-current. So we could definitely
do that.

Blair

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Sep 19 16:04:44 2007

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

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.