On 3/27/07, David Glasser <glasser@mit.edu> wrote:
> On 3/26/07, Daniel Rall <dlr@collab.net> wrote:
> > While I haven't reviewed this patch thoroughly, from the description
> > I'm not in favor of it. Rather than having the global configuration
> > file override any values specified for per-repository configuration,
> > shouldn't this work the other way around, as it does for our client
> > library configuration files? Per-repository configuration should be
> > overlayed on top of a global configuration file.
>
> My thoughts were similar. From Blair's comments in STATUS, it seems
> like a major incentive here is to not require parsing config on every
> svnserve access. I'm not sure if this optimization saves much, but
> perhaps the best way to handle this is to make a global config file
> which is overridden by per-repo files, and a new global-only
> configuration option "ignore-repository-configuration"?
>
[sorry, probably I'm not understand something]
As I understand Blair's arguement to implement global repository
configuration file is performance. I think we could implement it
better using some caching of configuration files in svnserve. And
check only time stamp of configuration and authz files.
--
Ivan Zhakov
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Mar 27 13:00:57 2007