RE: [uddi-spec] Basic Profile Version 1.0 - Working Group Draft

[Daniel Feygin <feygin@unitspace.com>]

Claus,

I support your argument that UDDI is a *particular* XML Web Service,
which may legally introduce limitations on what it considers its valid
input.  However, I would like the group to consider the one technical
argument for supporting UTF-16 by all XML Web Services, including UDDI,
which I perceive as compelling.

Character encoding support is something wired deep into the tools used
to produce both Web services and their clients.  While server and client
implementations have full control and responsibility over the semantics
of the service and of their behavior, they have no standard means to
dictate to their lower-level infrastructure that a certain *standard*
encoding is not acceptable for a given schema or service - unlike all
other technical details of message/interface definition, that is
impossible to express in XML schema and in WSDL.  Therein I see a
problem blocking UDDI registries from being *universally* accessible.

My argument is in favor of adopting UTF-16 support in principle.  I
believe that existing V2 and V3 specs should be adopted as OASIS
standards as they are first, otherwise we will end up with different V2
and V3 specifications released by different bodies (UDDI Consortium V2
vs. OASIS UDDI TC V2).

Daniel


> -----Original Message-----
> From: Von Riegen, Claus [mailto:claus.von.riegen@sap.com] 
> Sent: Tuesday, October 22, 2002 2:53 PM
> To: 'Daniel Feygin'; uddi-spec@lists.oasis-open.org
> Cc: 'Keisuke Kibakura'
> Subject: RE: [uddi-spec] Basic Profile Version 1.0 - Working 
> Group Draft
> 
> 
> By pointing to the WS-I Basic Profile, my intent was not to 
> restart discussions we already had but to give you an idea on 
> what the Basic Profile is all about. After sending the 
> results from our last conference call to the WS-I Basic 
> Profile WG (a copy is available at 
> http://lists.oasis-open.org/archives/uddi-spec/200210/msg00022.html
> ), which by the way summarize the reasons why we decided
> to not consider to adopt UTF-16 for UDDI Version 2, the Basic 
> Profile WG slightly changed their position.
> 
> The compelling reason why ARBITRARY Web services must support 
> both UTF-8 and UTF-16 (as specified in the Basic Profile) is 
> that a potential user cannot programmatically derive the 
> supported character encoding if only one of both is supported 
> except by trial and error. As far as UDDI (and other 
> WELL-KNOWN meta Web services) is concerned, potential clients 
> KNOW when they interact with a UDDI registry and since the 
> UDDI specification mandates UTF-8, they KNOW that they must 
> use UTF-8 (or use one of the tools available today that 
> implicitely use UTF-8). Thus, the WS-I Basic Profile WG is 
> currently in a better position to accept UDDI's exception of 
> the general rule.
> 
> Claus
> 
> -----Original Message-----
> From: Daniel Feygin [mailto:feygin@unitspace.com]
> Sent: Dienstag, 22. Oktober 2002 10:59
> To: uddi-spec@lists.oasis-open.org
> Cc: 'Keisuke Kibakura'
> Subject: RE: [uddi-spec] Basic Profile Version 1.0 - Working 
> Group Draft
> 
> 
> Keisuke,
> 
> Sorry, I did not mean to imply that UDDI registry is merely 
> an XML processor.  In fact, we are in complete agreement on 
> both of your statements.
> 
> However, we must decide whether UDDI client must be able to 
> use any valid XML conforming to UDDI schema or whether it is 
> a appropriate use of XML to artificially restrict usable 
> information set to XML instances encoded in UTF-8.  The 
> reason I now believe the restriction is artificial is because 
> I see no significant implementation issues (although I've 
> done no detailed analysis and there may indeed be some impact 
> with regard to things like storage, search and sorting), 
> since XML processors must already be capable of parsing 
> UTF-16 XML documents and this applies to both server and client.
> 
> It appears that there is no rationale behind UDDI 
> specifications' restriction on messages being expressed in 
> UTF-8, and therefore implementation impact on adopting UTF-16 
> support should be minimal, because XML processing is 
> delegated to lower level infrastructure. Another factor in 
> favor of resolving this issue by adopting UTF-16 support is 
> that UDDI clients may not be able to control precisely how 
> their requests are serialized into XML.
> 
> Daniel
> 
> 
> > -----Original Message-----
> > From: Keisuke Kibakura [mailto:kibakura@jp.fujitsu.com]
> > Sent: Tuesday, October 22, 2002 11:56 AM
> > To: uddi-spec@lists.oasis-open.org
> > Subject: RE: [uddi-spec] Basic Profile Version 1.0 - Working 
> > Group Draft
> > 
> > 
> > Dan,
> > 
> > UDDI implementation uses an XML processor, which must conform to
> > the XML specification, but UDDI itself is not an XML processor.
> > 
> > I don't think the current UDDI specification is illegal,
> > though it uses only UTF-8. 
> > 
> > --
> > Keisuke Kibakura
> > 
> > 
> > >>>>> On Tue, 22 Oct 2002 10:44:32 +0400,
> > >>>>> Daniel Feygin <feygin@unitspace.com> wrote :
> > 
> > > Working group draft states: 
> > >     All XML Processors must support UTF-8 and UTF-16, per
> > the XML 1.0
> > > specification.
> > > 
> > > According to this statement, UTF-16 support is mandated by
> > XML 1.0.
> > > If that is correct, then we cannot claim that UDDI Web 
> Service is a
> > > true XML Web Service - it is a service supporting only a 
> > subset of XML
> > > expressed in UTF-8.
> > > 
> > > This certainly implies that some errata is required to correct the
> > > problem - the question is whether we amend the specifications to 
> > > clearly indicate that UDDI is not in compliance with XML 1.0 (and 
> > > therefore with WS-I Basic Profile) or whether we become 
> > XML-compliant
> > > with respect to the UTF-16 issue.
> > > 
> > > Daniel Feygin
> > > UnitSpace
> > > 
> > > 
> > > > -----Original Message-----
> > > > From: Von Riegen, Claus [mailto:claus.von.riegen@sap.com]
> > > > Sent: Monday, October 21, 2002 7:11 PM
> > > > To: 'uddi-spec@lists.oasis-open.org'
> > > > Subject: [uddi-spec] Basic Profile Version 1.0 - Working
> > Group Draft
> > > > 
> > > > 
> > > > Here is the link to the WS-I Basic Profile that was 
> recently made 
> > > > public:
> > > > 
> > http://www.ws-i.org/Profiles/Basic/2002-10/BasicProfile-1.0-WG
> D.htm.
> > > As for the character encoding issue we already discusses, the
> > > requirement to support both UTF-8 and UTF-16 is listed as 
> > > "R1012" in section 4.1. Section 3.2 defines conformance 
> > > (against the profile) of Web services in general.
> > > 
> > > Claus