Re: proposal: make binding/operation/(input|output)/@messageLabel optional

On Thu, 18 Mar 2004 15:16:29 +0000
paul.downey@bt.com wrote:
> i'm happy to +1 this so long as the default values for the 
> messageLabel attribute are 'obvious'. 
> 
> That is I don't want to have to interpret complex rules to infer the 
> default values for these attributes based upon the presence or order 
> of input and output elements within a binding/operation. 

The rules are not complex.  *If* the specified message exchange pattern
contains only one message for any given direction (this includes, as a
matter of fact, all the currently defined MEPs), then the label is
optional.  Order is not significant.  If the MEP is in-out, the output
element may appear before the input element in the WSDL, and there is no
ambiguity.  So long as direction unambiguously identifies the message,
the label is clearly not needed.  As soon as there are two messages in
any given direction, labels become required.

> I'm puzzled about the impact on each of the MEPs identified in part 2
> as well as any MEPs added in the future. What would tip the balance 
> for me is a simple cheat-sheet: a table of the possible default 
> messageLabel default values being considered. 

It would be better to think of these not as "defaults", but as
unambiguous identification via an alternative algorithm.

Amy!
-- 
Amelia A. Lewis
Architect/Principal Engineer
TIBCO/Extensibility, Inc.
alewis@tibco.com

Received on Thursday, 18 March 2004 10:37:28 UTC