LC5 Discussion

Hi folks:

I took an action at the F2F to start some more email discussion about
issue LC5 [1], which if you'll recall is about the [source endpoint]
property and it's correlate the <wsa:From> header.  The problem is that
currently we don't have any testable assertions/rules regarding this
property.  At the F2F there were three (fairly evenly divided) camps, in
favor of the following choices:

1) Remove it (i.e. mark the feature "at risk")
2) Leave it as-is (it's good enough for people to do stuff with later)
3) Add some real semantics for it

Personally, I don't very much like option (2), so I'm going to avoid
that one in this particular discussion (although you should feel free of
course to champion it if you feel it's the right thing).  I believe that
if we don't want to define semantics for the property, we should just
remove it and let others define their own MAP if and when they need it.
So between choices (1) and (3) I'd like to offer the following
semi-proposal:

I think this issue hinges around use-cases.  At the F2F a few people
(Anish, others?) brought up some ideas for why the [source endpoint]
might be really useful (I recall NAT and auditing being discussed).  If
we could see these written up, I think it might be easier to decide
whether we'd like to keep the thing around or not.  If we *do* want to
keep it around, I would propose we work it into the defaulting semantics
for replies/faults - i.e. "for faults, use FaultTo.  If there's no
FaultTo, use ReplyTo.  If there's no ReplyTo, use From."

Personally, I'm fine with giving it some substance, and I'm fine with
removing it.  I just don't particularly like the unclear situation it's
in right now (then again, I suppose action is that way too....).  Can
those with use-cases for this property please post them?

Thoughts / comments appreciated.

Thanks,
--Glen

[1] http://www.w3.org/2002/ws/addr/lc-issues/#lc5

Received on Monday, 16 May 2005 20:15:17 UTC