Re: [emergency] Schema Test Results

[R. Allen Wyke <emergency-tc@earthlink.net>]

I have spent some time looking into this as well, and I admit that I am 
having trouble finding any kind of definitely answer from the XML 
Schema spec. Maybe I am over looking it - not sure. Anyway, I have 
included what I have found as well as the remaining questions that, if 
we can answer, will resolve these....

On Jul 16, 2004, at 5:55 PM, Kon Wilms wrote:

> Thanks Art,
>
> Using the XSD file from the document, Here's the run-through:
>
> 1. Microsoft .NET DOM and Borland DOM fail on all <simpletype name=nnn>
> fields. Since simpletype is nested within <element name=nnn> it 
> inherits
> the name and thus <simpletype name=nnn> should simply read <simpletype>
> -- the name is already declared. So elements must be named, but nested
> element simpletypes not (where the name is identical).

Kon: you say "but nested element simpletypes not". Where in the XML 
Schema spec does it say this? Not talking about what parsers do - but 
what the spec says. If we can find where the spec definitely says they 
MUST NOT, then this issue is resolved. Our schema is wrong (wrong = 
does not conform to the XML Schema spec) and we need to publish in 
errata.

I have looked all over the spec and it appears to be an area of 
ambiguity, although I am not confident my research 100% covered what it 
needed. I found a lot of places where it looked to be "allowed", but 
then others that said names must be unique. Problem is that "unique" 
did not specify if that was needed in parent-child situations whereby 
the child is actually only a child by nature of nesting, as it serves 
the purpose to further define the parent.  Here, let me show you what I 
mean:

"Except for anonymous simple types (those with no {name}), since type 
definitions (i.e. both simple and complex type definitions taken 
together) must be uniquely identified within an ·XML Schema·, no simple 
type definition can have the same name as another simple or complex 
type definition."

Ok, that make sense, but it does not necessarily address our question. 
Not too mention the fact that it is in a non-normative section of Part 
1 of XML Schema 
(http://www.w3.org/TR/xmlschema-1/#Simple_Type_Definition_details - 
scroll down to first written paragraph), which was included as a cross 
reference to Part 2 of XML Schema 
(http://www.w3.org/TR/xmlschema-1/#Simple_Type_Definitions - see NOTE 
under list of links for Section 3.14). Well, if you jump over the Part 
2 (http://www.w3.org/TR/xmlschema-2/#dc-defn - again, scroll down to 
first written paragraph) the language is slightly different, references 
anonymous types, but neither clarifies or reinforces our situation.

So, where are we? Well, I think we are dancing all around the right 
spots in the XML Schema spec, but I, for one, am just not knowledge 
enough at this level of detail on XML Schema to definitely offer a 
recommendation on or even a solid pointer to the solution. Probably 
time, if no one else knows, for someone to ask the W3C XML Schema 
Working Group (http://www.w3.org/XML/Schema) for some guidance.

Rex - as our historical liaison person, can you facilitate this?

> The .NET / Microsoft DOM parsers will validate the spec with the above
> modification.
>
> 2. I have had some older DOM parsers still cough up blood (Borland XML
> Mapper for C++Builder and friends is one). They can be repaired by
> changing:
>
> <simpleType>
> <list itemType = "string"/>
> </simpleType>
>
> to:
>
> <simpleType>
> <restriction base = "string"/>
> </simpleType>
>
> One reason could be that these incorrectly parse the XPath absolutely
> instead of relatively. Since there are simpletypes nested within
> complextypes and complextype by w3c definition cannot contain a list
> item, these fail.

Correct me if I am wrong, if I understand correctly what you are 
saying, but just because <complexType> does not directly allow <list> 
does not mean that a valid child or grandchild of <complexType>, such 
as our <sequence> -> <element> - > <simpleType> situation, cannot have 
<list>. In either case, this may also be a good question to ask the W3C 
XML Schema WG.

Additional thoughts from anyone?

Allen
-------------------
R. Allen Wyke
Chair, OASIS Emergency Management TC
emergency-tc@earthlink.net