Re: locating schemas via public ids

[Paul Grosso <pgrosso@arbortext.com>]

At 22:09 2001 01 23 -0800, Lauren Wood wrote:
>On 23 Jan 2001, Paul Grosso wrote:
>> Below is what I'm suggesting we submit to the XML Schema WG.  We'd
>> need to send it soon, so please give comments asap.  
>> 
>> Also, to what extent do we need to check with the OASIS AC rep before
>> we send this to the XML Schema WG.  And, since OASIS is a W3C member,
>> but OASIS is an outside organization, should we be involving the XML
>> CG in this interchange or not?
>> 
>> My suggested answers:  Lauren should tell Laura what we're doing and
>> we should plan to cc her on the email we send.  We (Lauren or her
>> designate--I'm willing if designated) should inform the XML CG and cc
>> them on the email to XML Schema.
>
>I think this is a good plan. Normally I think I should be the person 
>to carry out all the parts of this; however since I will be out of email 
>contact from Thursday until Tuesday Feb 6th, I would prefer if 
>someone else can handle this, copying me on any 
>correspondence. Paul, I am happy to designate you for this 
>position. I will email Laura about our plans before I leave.

Okay, so you will email Laura and cc this group, correct?

Then I will inform the XML CG and send the email to them 
and the XML Schema.

I want to do this as soon as possible, but certainly no later
than the XML CG telcon next Tuesday the 30th, so get any
comments in asap.

>The suggested email seems fine to me.

Actually, upon reflection, I think I'll delete note [7] where
I suggest a possible solution.  While theoretically workable,
I realize it requires 3 levels of quotes, and it makes for
some pretty ugly attribute values.

>Lauren
>
>
>> ----------------------------------------------------------------
>> The OASIS Entity Resolution Technical Committee (OERTC) [1] is
>> chartered to developed an entity resolution catalog format in XML (XML
>> Catalog or xmlcat).  The purpose and functionality of this catalog
>> format is to cover that which the SGML Open/OASIS TR9401 [2] Entity
>> Management Catalog did, but using XML instance syntax and tailored for
>> use with XML.
>> 
>> Many implementors and users have found public ids and entity
>> management catalogs to be very usefulness in practical situations
>> ranging from individual use to major production environments, and
>> there is a desire to be able to use such techniques for accessing XML
>> resources, especially "public" resources such as published DTDs and
>> Schemas.  The OERTC has the support of several implementors, and its
>> work has received interest from the xml-dev community.
>> 
>> During our work, we realized that the current Schema Structures
>> draft appears to make it impossible to provide schema-locating
>> hints using anything other than URIs.  Specifically, public
>> identifiers [3] could not be used as the spec is currently written.
>> 
>> In Structures, 6.3.2 How schema definitions are located on the Web
>> [4], it says:
>>   [xsi:schemaLocation] records the author's warrant with pairs of URI
>>   references (one for the namespace URI, and one for a hint as to the
>>   location of a schema document defining names for that namespace
>>   URI). [xsi:noNamespaceSchemaLocation] similarly provides a URI
>>   reference as a hint as to the location of a schema document with no
>>   targetNamespace.
>> 
>> The problem is that each member of schemaLocation and the value of
>> noNamespaceSchemaLocation is required to be a URI.  Furthermore, the
>> members are undelimited and separated by spaces, and public
>> identifiers can contain spaces.
>> 
>> The XML Catalog (as did TR9401 before it) would allow a user
>> to locate a resource using all the information that might be 
>> known about it (name, system id, public id), and certainly schema
>> resources will be given public ids (several have already [5]).  But
>> this only works if there is some way to include that information in
>> schemaLocation.
>> 
>> Therefore, the OERTC asks [6] that the XML Schema WG make allowances
>> in schemaLocation for specifying both public and system ids [7].
>> 
>> paul
>> 
>> [1] http://www.oasis-open.org/committees/entity/
>> [2] http://www.oasis-open.org/committees/entity/9401.html
>> [3] http://www.w3.org/TR/REC-xml#NT-PubidLiteral
>> [4] http://www.w3.org/TR/2000/CR-xmlschema-1-20001024/#schema-loc [5]
>> http://www.oasis-open.org/committees/entity/ident.html#schema [6]
>> OASIS is a W3C member organization [7] Not to presume to constrain a
>> solution to the problem, but
>>     even something such as the following might be acceptable:
>>     Say that each member--that is currently a URI--is optionally
>>     delimited by quotes and consists of either a SystemLiteral
>>     (production 11 in XML 1.0) OR the second half of the
>>     disjunction in production 75, to wit:
>>       'PUBLIC' S PubidLiteral S SystemLiteral
>>     (which, since it contains spaces, would necessarily be quoted).
>
>
>