Peter N. Lundblad wrote:
> On Thu, 31 Mar 2005, Max Bowsher wrote:
>
>> Barry Scott wrote:
>> Patches require log messages, but more importantly, in what situation
>> would
>> this new function be worth it compared with calling
>> svn_ra_open/svn_ra_get_repos_root directly? (I'm not saying it's not
>> worth
>> it, just wanting to be sure it is before promising to create it.)
>>
> I'm not Berry, but I can understand that someone just wanting
> an RA session would be glad to avoid having to mess with RA callbacks and
> authentication stuff, when that's already handled by the client APIs. But
> instead of new special-case functions, maybe we should export some general
> purpose client function to open an RA session, such as
> svn_client__ra_session_from_path?
This seems like a good thing to do - we even have a private function to
model it on: svn_client__open_ra_session.
It occurs to me that if in the future we plan to cache the repos root in the
entries file, there would be justification for a repos_root analogue of
svn_client_uuid_from_path, which would take advantage of the cache, and fall
back to a RA access only if it was not present.
Max.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Mar 31 14:31:56 2005