RE: LC74c proposed resolution

FWIW: in UDDI Version 2 the xml:lang attribute (where it occurred) was required (one missing was allowed), and required to be unique. In Version 3 this restriction was lifted - it was legal to have multiple occurrences of the same xml:lang in the one list - people produced some reasonable use cases.
 
Let's skip the mandatory and unique phase :-)

	-----Original Message----- 
	From: www-ws-desc-request@w3.org on behalf of Jacek Kopecky 
	Sent: Mon 30-May-05 17:59 
	To: Amelia A Lewis 
	Cc: WS-Description WG 
	Subject: Re: LC74c proposed resolution
	
	


	Amy, I like the suggestion, I also think that making xml:lang mandatory
	and unique on repeated documentations would be overkill.
	
	I wonder though, for schema validation, do we actually need to allow
	extensibility attributes to allow xml:lang? I have a vague feeling that
	the xml: attributes are exempt from validation so they are allowed no
	matter what the schema says. I may be wrong, though, in which case
	adding the extensibility attributes on documentation would be good.
	
	Best regards,
	
	Jacek
	
	On Thu, 2005-05-19 at 13:40 -0400, Amelia A Lewis wrote:
	> Heylas,
	>
	> LC74c [1] raises issues related to internationalization of
	> documentation elements, and proposes a solution.  Excerpted:
	>
	> a) The <documentation> element require an xml:lang attribute.
	>    The attribute may be empty (xml:lang="")
	> b) The <documentation> element be allowed to be repeated,
	>    provided the xml:lang attributes in each of the elements be unique.
	>
	> I think that this is more than is necessary, on examination.
	>
	> I recommend that we do only the following:
	>
	> c) add maxOccurs="unbounded" to the reference to wsdl:documentation in
	> the definition of DocumentedType.
	>
	> We make no statements about how multiple documentations may relate to
	> one another, if present, although the use case presented above is
	> feasible and even likely.  We neither require xml:lang, nor require it
	> to be unique (multiple documentation elements could, in theory, share
	> the same xml:lang attribute value).
	>
	> *Optionally*, we could make xml:lang a required attribute.  However,
	> I'm not convinced that this is useful.  In my experience, much
	> documentation seems to be written in Klingon, or possibly in the
	> private languages of twins.  Permitting the recurrence of the
	> documentation element permits proper internationalization (and
	> potentially other use cases involving multiple documentation elements,
	> such as an ASCII presentation versus an algorithm in MathML, perhaps,
	> or different authorities for different documentation blocks); if it
	> allows xml:lang, we're done.
	>
	> Unfortunately, we do not allow attribute extension on a document
	> element.  *sigh*  So, we should *also* add anyAttribute
	> (namespace="##other") to the DocumentationType definition.
	>
	> In short: change the recurrence of wsdl:documented in DocumentedType to
	> *, add attribute extensibility DocumentationType, let usage of multiple
	> documentation elements be determined in practice.
	>
	> [1: http://www.w3.org/2002/ws/desc/4/lc-issues/#LC74c]
	>
	> Amy!