You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@subversion.apache.org by Stephen Peters <po...@portnoy.org> on 2004/03/10 17:24:20 UTC

Performance of `svn status '

I noticed that an 'svn status <filename>' command was taking a long
time to complete, which I found odd since it should only be checking
the status of one file and not using the network.  Performing an
'strace' on the command, however, indicates that every directory in
the subtree was having its 'format' and 'entries' files read.  Since
this was a significant number of directories, this was taking a long
time.  The same behavior occurs even when I do 'svn status -N
<filename>'.

Is there a reason why all the subdirectories have to be checked, or is
this a performance bug?

-- 
Stephen L. Peters                                  portnoy@portnoy.org
  GPG fingerprint: A1BF 5A81 03E7 47CE 71E0  3BD4 8DA6 9268 5BB6 4BBE
    "Celebrated heads of state or 'specially great communicators.
 Did they have brains or knowledge?  Don't make me laugh! They were...
	POPULAR!  It's all about pop-u-ooo-lar..." -- Wicked

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org

Re: Performance of `svn status '

Posted by "C. Michael Pilato" <cm...@collab.net>.
Stephen Peters <po...@portnoy.org> writes:

> I noticed that an 'svn status <filename>' command was taking a long
> time to complete, which I found odd since it should only be checking
> the status of one file and not using the network.  Performing an
> 'strace' on the command, however, indicates that every directory in
> the subtree was having its 'format' and 'entries' files read.  Since
> this was a significant number of directories, this was taking a long
> time.  The same behavior occurs even when I do 'svn status -N
> <filename>'.
> 
> Is there a reason why all the subdirectories have to be checked, or is
> this a performance bug?

It's a known perf bug, with (IIRC) a fix either in place, or existing
as a patch awaiting approval.

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org

Re: Performance of `svn status '

Posted by Ben Collins-Sussman <su...@collab.net>.
On Wed, 2004-03-10 at 11:24, Stephen Peters wrote:

> Is there a reason why all the subdirectories have to be checked, or is
> this a performance bug?

It's a known bug in 'svn status'.  A fix has been merged into the trunk,
so it will be out in svn 1.1.




---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org