On Thu, Mar 06, 2003 at 04:43:12PM -0600, Karl Fogel wrote:
> Ben Collins-Sussman just pointed something out...
>
> Mark, your patch below makes it so that 0.19 servers won't break older
> clients. However, the other thing we wanted to do is make clients
> less strict about validating incoming XML in general, right? That's
> what caused this problem in the first place. Although we've now
> solved it for the UUID case specifically, we also want clients to
> relax a bit, so that this problem doesn't happen for other new
> properties.
Yes, that's right.
>
> But if that's the plan, it's important for us to get that relaxation
> into 0.19. Otherwise, we'll have to wait yet one more release cycle
> if we come up with some new property.
You're absolutely right; the sooner we fix that the better.
>
> In other words, the patch below is only one half of the change we
> really need. Does that seem right to you?
>
Actually, it's only a third, because we also want to start sending
and processing the special HTTP reponse headers. I just figured
breaking unused functionality was better than delaying the release.
> If so, then Michael should probably wait until you have a chance to
> commit that other step as well (and commit the patch below along with
> it).
>
> Or, I could be totally misunderstanding the thread between you and
> gstein about the solution. Just ~thwack~ me if so... :-)
>
No, you're dead on; I just broke off a tiny piece of the plan to
unblock 0.19.
I should be able to get the rest in tonight, barring unforseen OS/X trouble.
Even in that case, there is a burgeoning night-time #svn OS/X cabal who
I'm sure will support me. :-)
--ben
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Mar 7 00:39:03 2003