Ben Collins-Sussman <sussman@collab.net> writes:
> Ah, so then, if we went this route, one of two things would need to be
> true, because of the multiple pass issue:
>
> * the svn_client_status() would still need to construct a hash of
> status structures, but it would be totally private. As soon as a
> structure was deemed "fully mature", it could be sent back to the
> client via notification.
>
> -or-
>
> * svn_client_status() sends status information back to the client on
> the fly, with the understanding that it may send information more
> than once per path.
Right, I was assuming the latter.
> Are we still liking this? Keep in mind that this solution, along with
> my proposal above, still requires that we do the network pass first,
> which is a goodly bit of rewriting.
Understood.
> At the moment, I think I'm now shooting for the lowest-hanging fruit:
> I've got svn_client_status() sending a 'status' notification back to
> the client every time a statstruct is added to the hash. The client
> currently has no information, other than the fact that some path was
> statted. Still, this is enough information to show a procession of
> dots, a spinning cursor, or a throbbing icon. Maybe that's all we
> need for now -- just an indication that *something* is happening. We
> can "enrich" the status notification system later on.
Sounds good to me. Heck, as long as Subversion does what Subversion
documents, it can't be wrong, right? :-)
-K
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Nov 11 20:25:04 2002