Re: [xml-dev] Namespaces, Xml Schema Whitespace normalization, xs:anyURI,and URILiterals in XPath 2.0

[Jeff Rafter <lists@jeffrafter.com>]

> I agree that the uri of the example is a valid
> xs:anyURI following the specs. And also that
> this might be problematic.
> 
> About the namespace name being an URI or 
> an xs:anyURI or simply a string, I am not sure,
> parsers will be ok with any not empty 
> xml characters sequence as namespace "uri", 
> right ?

There are tests in the XML Test suite that check specifically for URI 
validation. I see that Mike Kay just responded saying the exact 
opposite... and when I am not sure about something I defer to people 
like Michael Kay. Checking the Namespaces in XML 1.1 rec confirms this:

"To conform to this specification, a processor MUST report violations of 
namespace well-formedness, with the exception that it is not REQUIRED to 
check that namespace names are legal IRIs."

*sigh*


> I was thinking more to the case of 
> xs:anyURI as base type for an URL type
> like the href attribute of html, to give
> the idea.

> Isn't the schema supposed to stay between
> the xml and my application ? I mean as user 
> someone that writes an xml document,
> for application I mean a program that uses the 
> xml of the user together with a 
> schema-enabled-processor that provides 
> the program a datatyped value,
> for example a number that can be summed or 
> a space collapsed text etc.

I see where you are coming from, but a schema can be used for a variety 
of different things in the XML landscape. I think we are agreeing here.

> For a form or an editor I agree that
> url-encoding of characters should be
> made by the (input) application.
> 
> I think a reasonable option
> could be that the schema does not touch
> the uri so that -is responsability  of the 
> application to build the uri, maximum
> that can be given is trimming-.

Many people would insert a comment about your use of "reasonable option" 
and "XML Schema" in the same sentence :). I agree it would be reasonable 
to expect that, but like others have pointed out... namespaces/uris and 
there use throughout XML has lots of warts.

>  
> Otherwise I don't know how can help 
> a validator that, if I have an url like this: 
> 
> <http://example.org/I%20have%20space>
> 
> tells me that these are all valid lexicals
> representations of that url:
> 
> "http://example.org/I%20have%20space";
> "http://example.org/I%20have space"
> "http://example.org/I%20have  space"
> "http://example.org/I%20have   space"
> 
> 
> I cannot distinguish between a valid uri
> like the first one and the others 
> (with a space) that are really wrong. 
> 
> If is data coming from an other app,
> I think there is something wrong with 
> the application that produces uris like 
> the last ones, but I cannot check it with
> the schema, I have to do it my self 
> when the schema tells me they are valid.
> 
> At the same time the only benefit that the
> user has is to have the space character escaped,
> all the other characters must anyway 
> be escaped by him or his input application,
> and also consecutive spaces.
> 
> May be there is some other use or apect
> I am not aware of. I noticed now that both IE 
> and Mozilla interperet this as valid url:
> 
> file:/D:/Music/bob/Catch%20A Fire

Well, those browsers accept a lot of things that aren't necessarily 
valid. I think that is the intent here as well... when you have 
something that can safely be converted to a URI the XML Schema rec 
allows for it. In terms of the multiple spaces being normalized, you 
could always create a derived type that constrained the whiteSpace facet 
to preserve. From the XML Schema view, if your application is consuming 
the PSVI it would only see the corrected URI anyway.


>  Generally, the whitespace normalized
>> value is not what you 
>> deal with in an editor... you deal with the lexical
>> value.
>>
>>> Just to be clear: this should be a valid xs:anyURI
>>> isn't it ? http://www.example.org/c a/c%20b
>>> and when I 'map' it to an URI becomes
>>>
>>> http://www.example.org/c%20a/c%20b
>>>
>>> Right ?
>> If by "map" you mean apply the algorithm from XLink
>> section 5.4, yes. 
>> Note that this kind of "mapping" is not done to
>> namespace URIs in 1.0 or 
>> 1.1, so in theory you would never have a (legal)
>> namespace that 
>> contained space characters.
> 
> Legal for what ?

Sorry, by legal I meant valid according to the URI rfc.

Cheers,
Jeff Rafter