Re: wsa:FaultTo unusable for SOAP mustUnderstand faults

[ Switching to public-ws-addressing@w3.org
  The thread is at:
  http://lists.w3.org/Archives/Public/public-ws-addressing-comments/2005Jun/0003.html ]

* David Hull <dmh@tibco.com> [2005-06-08 20:26+0200]
> One of the outcomes of the Palo Alto meeting of the Async TF was Tony's
> Timeline.  On this timeline, there is a point at which the [fault
> endpoint] becomes "known", and faults can be sent there (actually, I
> think I'm oversimplifying somewhat here).  Before that, most bets are
> off.  For example, if I POST a request to an invalid URL, I'll get back
> a 4xx response before anything even has a chance to look at the [fault
> endpoint].  If I fire-and-forget a message to a bogus address, the
> transport layer may or may not be able to tell me I did this.
> 
> The upshot is that we only know for sure that the [fault endpoint] MUST
> be used in the case where a well-formed and properly addressed message
> arrives and the receiver produces a fault based on this.  This would
> certainly include, say, sending a negative number to a "real square
> root" operation.  It might or might not include a complaint brought on
> by an invalid refparam or some other bad header.
> 
> Your analysis shows that it /cannot /include errors with mustUnderstand
> headers, putting them in the same class as transport errors, malformed
> SOAP envelopes and such.  This may not be desirable, but it seems to be
> a consistent reading of what we have.

After several discussions over the past week, I'm OK with wsa:FaultTo
not applying to mustUnderstand faults, especially as I think that it
would be very hard to do so. However, we should make this clear in the
spec to avoid confusion and test it at CR.

Cheers,

Hugo

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