Re: Connection with TSVN to nonApache SVN server

At 11:04 AM -0500 7/7/03, Ben Collins-Sussman wrote:
>Sure, everybody wants this. But it opens a whole new can of worms.
>Now every single ra_svn library needs to keep a list of server
>versions it can speak to? And vice versa?

I don't think so. If we're comfortable saying "one version back" in
response to the FAQ "what the heck does 'malformed HTML mean?????'",
then why would we not be comfortable doing version arithmetic in the
code?

If you're so optimistic as to hope that compatibility is sometimes
better than "one version," then make the message a warning on first
connection, so at least it's there to annotate the user-hostile
ultimate collapse.

>This is why most protocols
>exchange *features* rather than version numbers (cvs included).

I challenge your "most" statistics, but feature negotiation is
certainly a goo (post-1.0!) idea.

>In this particular case, I'm not even sure ghudson needs to implement
>either of these things; the easiest thing he could do is version the
>protocol itself, and exchange protocol versions.

That would work for me.

-- 
-==-
Jack Repenning
CollabNet, Inc.
8000 Marina Boulevard, Suite 600
Brisbane, California 94005
o: 650.228.2562
c: 408.835-8090
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org