Re: Extensions

Jeffrey,

I think that most of the complexity is illusory and we can find satisfactory
ways to explain how extensions work. Most programming languages have
lots of complex rules, yet people use them routinely. It helps that, in the course
of their daily activities, programmers rely on books and examples more than
language specifications! With that, I don't mean that our spec will go unread,
just that it will be as readable as we can make it all while being rigorous.
I'm sure the primer will contain a good explanation of extensibility, possibly
along the lines suggested by Igor.

I'll be sending out an updated proposal covering both element and attribute
extensions in a few minutes.

Roberto

--
Roberto Chinnici
Java and XML Software
Sun Microsystems, Inc.


Jeffrey Schlimmer wrote:

> I too am concerned about the complexity of the proposal; the table just
> calls out all the possible situations, and if we're going to define an
> optional global EII, an optional AII within that EII, and an optional
> AII that may be attached to other EII's, then we're going to have to
> specify what happens in all the cases.
>
> Does the convenience of a global EII outweigh the complexity of having
> to specify how / when it overrides a local AII?
>
> --Jeff
>
> -----Original Message-----
> From: Sedukhin, Igor [mailto:Igor.Sedukhin@ca.com]
> Sent: Tuesday, May 28, 2002 8:14 AM
> To: Sanjiva Weerawarana; Jeffrey Schlimmer; WS-Desc WG (Public)
> Subject: RE: Extensions
>
> Sanjiva,
>
> >>If needed we can put an ed note in the working draft saying we're
> considering adding more flexibility and see whether we get feedback
> asking for it.
>
> By the time we get feedback, it may be too late to make it flexible. By
> then all WSDL processors will already be implementing logic that we have
> spec'ed out.
>
> I'm also concerned by the complexity of Jeffrey's wordings. May be, at
> the initial stages of defining extensions we can get a simple
> explanation of what it means before we get into formal specifications. I
> suggest adding this clause:
>
> "When declaring an extension (explicit), all its elements that appear in
> the WSDL document are by default required for understanding by WSDL
> processors (i.e. wsdl:required = true). Optional extensions do not have
> to be declared (implicit) or may be declared specifying wsdl:required =
> false.
> An extension element may override wsdl:required attribute, in which case
> the element is required or not regardless of the extension declaration.
> By default an extension element does not override the wsdl:required
> attrubute and rules apply according to the extension declaration
> (explicit or implicit)."
>
> It does not sound very complex (to me :) and WSDL processor
> implementations may easily take it into account.
>
> -- Igor Sedukhin .. (Igor.Sedukhin@ca.com)
> -- (631) 342-4325 .. 1 CA Plaza, Islandia, NY 11788
>
> -----Original Message-----
> From: Sanjiva Weerawarana [mailto:sanjiva@watson.ibm.com]
> Sent: Monday, May 27, 2002 11:42 PM
> To: Jeffrey Schlimmer; WS-Desc WG (Public)
> Subject: Re: Extensions
>
> Hi Jeffrey,
>
> Your table has too many cells for my liking! It looks like an interop
> nightmare is being created ..
>
> Why don't we go with the simpler model ala WSDL 1.1 for requiredness?
> You get per-element "gotta have it" capability and that's it. If needed
> we can put an ed note in the working draft saying we're considering
> adding more flexibility and see whether we get feedback asking for it.
>
> Sanjiva.
>
> ----- Original Message -----
> From: "Jeffrey Schlimmer" <jeffsch@windows.microsoft.com>
> To: "WS-Desc WG (Public)" <www-ws-desc@w3.org>
> Sent: Friday, May 24, 2002 4:57 AM
> Subject: Extensions
>
> Roberto, thank you for patiently explaining the current proposal during
> the teleconference this morning.
>
> Just to make sure I understand the proposal, can it be accurately
> restated as?
>
> -----
>
> A WSDL parser MUST recognize a foreign EII if and only if one of the
> following is true:
>
> (a) The foreign EII has a wsdl:required AII that is true, or
>
> (b) The foreign EII namespace is declared with a wsdl:extension EII,
> that EII has a wsdl:required AII that is true, and the foreign EII does
> not have a wsdl:required AII.
>
> -----
>
> (a) is what we have in WSDL 1.1 today. (b) adds the global declaration
> but allows a wsdl:required AII on the foreign element to override the
> global declaration.
>
> Attached is a table that I crunched down to the two rules above.
>
> For completeness, let's allow AII extensions via (something like)
> <xs:anyAttribute namespace="#other" processContents="#lax"/>.
>
> Of course, because the wsdl:required AII cannot be attached to a
> (foreign) AII, only the global setting in the wsdl:extension EII can be
> used to indicate whether a WSDL parser MUST or MAY recognize such an
> AII.
>
> --Jeff

Received on Tuesday, 28 May 2002 14:44:27 UTC