Jon Stevens <jon@latchkey.com> writes:
> on 6/19/2000 10:36 PM, "Ben Collins-Sussman" <sussman@newton.collab.net>
> wrote:
>
> > Writing a native java client library is a HUGE task. I think writing
> > a java class that accesses the C client library thru JNI is relatively
> > straightforward. :)
>
> something that speaks the client protocol is a huge task?? why? i'm not
> talking about writting a gui client or anything. just the communication
> protocol. there is one for CVS
> <http://www.gjt.org/javadoc/com/ice/cvsc/package-summary.html>, why
> shouldn't/couldn't there be one for subversion? i think that would be a
> major benefit to subversion to have that.
>
Sorry, let me clarify.
The Subversion network protocol is flexible; the initial
implementation will probably be a modified form of WebDAV, in fact.
However, there's a lot more to writing a Java Subversion client than
speaking the protocol! The Subversion C client is split into three
libraries:
* routines to manage a working copy & administrative files
* routines to make repository filesystem calls over the
network
* routines for commands that use both of the other two libraries
(such as add, commit, diff, checkout)
In addition, you'll need code that can both produce and interpret the
delta formats being transmitted back and forth. (!)
I suppose it all depends on what you want to do. A *true* native Java client
would have to implement all these things. But if you just want to do
simple network filesystem reads from the repository, then the protocol
is just about enough.
Received on Sat Oct 21 14:36:05 2006