W3C home > Mailing lists > Public > public-ws-addressing@w3.org > February 2006

Re: Why we shouldn't make [destination] optional

From: Anish Karmarkar <Anish.Karmarkar@oracle.com>
Date: Sun, 12 Feb 2006 23:50:04 -0800
Message-ID: <43F03A2C.40704@oracle.com>
To: David Hull <dmh@tibco.com>
CC: "public-ws-addressing@w3.org" <public-ws-addressing@w3.org>

David Hull wrote:
> As much as I would like to make as many properties optional as possible, 
> I don't think that this will work for [destination].
> 
> If [destination] is optional, then we have to define behavior for when 
> it is missing.

Why and whose behavior?
We do not specify any behavior wrt to the value of the [destination] 
property.
If the value is missing, it is just that and nothing more.

>  Unless I've missed something, that behavior will be 
> exactly what we're currently defining for anonymous [destination].  But 
> since anonymous is available anyway, there will be no real difference 
> between leaving out [destination] and using anonymous.  If leaving out 
> destination means the same as giving anonymous for [destination], then 
> anonymous is effectively the default value, and we might as well just 
> say that.
> 

Specifying a default  value that is undefined, as I said during the last 
concall, does not make much sense to me. I don't think we should define 
a default in the Core that says -- whatever the binding document(s) 
says. Keep in mind that the binding document is under no obligation to 
define what 'anon' means. If one really need a binding-defined default 
value, then one can define that in the binding document, there is no 
need for a default value in the Core spec. Additionally, not defaulting 
allows one to say -- this property has no value (as none is required).

Specifying a default in the Core means that for a binding that does not 
have use of the 'anon' URI has to either: (a) say that 'anon' means 'no 
value,' or (b) require that every serialized message that uses the 
binding specify the value for wsa:To (i.e., the defaulting case will 
never arise) -- defeats the purpose of defining a default.

> In other words yes, the two approaches (default to anonymous and make 
> optional) are equivalent, but given the structure already in place, 
> defaulting is clearer.
> 
> I'm pretty sure someone made essentially this argument on the telecon.  
> If so, +1.
> 
> If leaving out [destination] does /not/ have the same effect as making 
> [destination] anonymous, we need to be clear on the difference and the 
> reasons behind it.
> 
> If we want to consider making [destination] optional anyway, it would be 
> option 3 on the list of choices for CR 20 [1].
> 
> [1] 
> http://lists.w3.org/Archives/Public/public-ws-addressing/2006Feb/0058.html
> 
> 
Received on Monday, 13 February 2006 07:50:17 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:35:11 GMT