Re: LC21 resolution issues

Hi Amy.

* Amelia A Lewis <alewis@tibco.com> [2004-10-27 15:11-0400]
> On Wed, 27 Oct 2004 15:03:59 -0400
> Hugo Haas <hugo@w3.org> wrote:
> 
> > We have decided that the HTTP binding would support all of our MEPs,
> > except for for robust-in-out and robust-out-in.
> 
> Never heard of 'em.
> 
> > The issue is that Part 2 does not define such MEPs AFAICT. We had
> > problem with those as they may involve 3 messages, but I don't think
> 
> Two of those listed below have this characteristic.
> 
> > that Part 2 MEPs have this characteristics, so I think that we want
> > all of them:
> > 
> >         2.2.1 In-Only
> >         2.2.2 Robust In-Only
> >         2.2.3 In-Out
> >         2.2.4 In-Optional-Out
> 
> Uses "message triggers fault", and as a result may have a fault in
> response to the optional out message.
> 
> >         2.2.5 Out-Only
> >         2.2.6 Robust Out-Only
> >         2.2.7 Out-In
> >         2.2.8 Out-Optional-In
> 
> Uses "message triggers fault," and as a result may have a fault in
> response to the optional in message.

Oh, so that must be the ones that we wanted to exclude from the HTTP
binding as the fault would not be transmittable in the context of an
HTTP interaction.

> Note: it is not acceptable to change the fault propagation ruleset for
> these MEPs.  It is, of course, perfectly feasible to define additional
> MEPs that use "fault replaces message" with otherwise similar semantics
> (a bit odd, in my opinion, but certainly doable).  "message triggers
> fault" is generally inappropriate for HTTP, in my opinion, except for
> the "robust" forms of single-message MEPs (also true of other strongly
> connection-oriented, client/server protocols).

Agreed.

Cheers,

Hugo

-- 
Hugo Haas - W3C
mailto:hugo@w3.org - http://www.w3.org/People/Hugo/