You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@subversion.apache.org by "Taylor, Richard" <rt...@t-tales.com> on 2007/04/10 10:36:48 UTC

files with large numbers of revisions

Hi,

Does svnserve slow down for files with large numbers of revisions ? 

In the documentation
(http://svn.collab.net/repos/svn/trunk/doc/user/svn-best-practices.html)
under the section "Be patient with large files" there is a description of
the svn client algorithm. This seems to imply that all the diff generation
is performed by the clients, leading to the conclusion that it is client
machine which would be the bottle neck in an svn system ?
Does the svn server perform any sigificant amount of work either generating
diffs or reconstructing files from deltas, either during an update or commit
?

Thanks
Rich
	 

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

Re: files with large numbers of revisions

Posted by Ryan Schmidt <ry...@macports.org>.
On Apr 10, 2007, at 05:36, Taylor, Richard wrote:

> Does svnserve slow down for files with large numbers of revisions ?
>
> In the documentation
> (http://svn.collab.net/repos/svn/trunk/doc/user/svn-best- 
> practices.html)
> under the section "Be patient with large files" there is a  
> description of
> the svn client algorithm. This seems to imply that all the diff  
> generation
> is performed by the clients, leading to the conclusion that it is  
> client
> machine which would be the bottle neck in an svn system ?
> Does the svn server perform any sigificant amount of work either  
> generating
> diffs or reconstructing files from deltas, either during an update  
> or commit ?

When a client commits, it does generate a diff between the working  
copy's pristine files and your modified files, and sends that to the  
server. Then the server goes and generates a different diff to store  
in the repository (based on the skip-deltas algorithm which is  
described elsewhere). When any client requests an update from the  
server, the server has to construct another different diff, between  
what the client already has in its working copy and what's now in the  
repository.

I don't have information about the performance of all this, or if  
it's affected by how many times a given file has been modified.

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