RE: Issue: ALLPROP_AND_COMPUTED

I think we might be able to close on this issue if we consider it 
pragmatically:

1. We are having a great deal of difficulty deciding on a single 
semantics for ALLPROP that fits well with the various related specs 
(e.g., delta-V).

2. We already seem to have general agreement on a change in the 
default semantics for ALLPROP (i.e., no depth infinity and the 
default if not specified changes to 0) that's incompatible with the 
current draft (and thus current versions of clients).

 From these two I conclude that there's no way that ALLPROP in the 
next draft is going to have the same semantics as ALLPROP in the last 
draft.  Thus existing clients who use ALLPROP are going to have to be 
rewritten to be compatible with servers going forward.

So we're faced with two alternatives---preserve ALLPROP and change 
clients, or give up on ALLPROP and change clients---which IMHO differ 
very little for client implementors but which are quite different for 
server implementors (the first being harder than the second).

As a server implementor, I vote for option 2: abandon allprop.  As a 
client implementor (which I also am), I can live equally with either 
(as long as it means that servers are CONSISTENT, which is easier 
with option 2).

Are there any client implementors out there who strongly believe that 
changing to a new semantics for ALLPROP is harder for them than just 
doing PROPNAME then PROPFIND?

     dan

P.S. Yes I realize that server implementors who implement 
cross-server COPY and MOVE are also client implementors, and should 
feel free to respond from that perspective! -d.
-- 
Daniel Brotsky, Adobe Systems
tel 408-536-4150, pager 877-704-4062
2-way pager email: <mailto:page-dbrotsky@adobe.com>

Received on Thursday, 3 May 2001 13:21:07 UTC