- From: Noah Mendelsohn <noah_mendelsohn@us.ibm.com>
- Date: Thu, 3 Jan 2002 10:22:47 -0500
- To: "Jacek Kopecky <jacek" <jacek@systinet.com>
- Cc: "xml-dist-app" <xml-dist-app@w3.org>
Jacek Kopecky writes:
>> The lexical representation is not important
>> in our case since the value is not in an
>> XML file. What's important is the value
>> space - arrays of bytes.
The reason I'm not comfortable with this is that schema types are about
lexical and value spaces. You can't get one without the other. The schema
WG seriously considered a design that would have used some notion of
abstract types to capture commonality of value spaces, while allowing you
to then derive subtypes to control the lexical representations. We never
found a way to get it right, and the current version of schemas does not
support such a notion. There are two built in binary types, one for hex
and one for base 64, and each has a specific lexical representation. You
can restrict those representations in a subtype, but you cannot otherwise
change or replace them.
------------------------------------------------------------------------
Noah Mendelsohn Voice: 1-617-693-4036
Lotus Development Corp. Fax: 1-617-693-8676
One Rogers Street
Cambridge, MA 02142
------------------------------------------------------------------------
Jacek Kopecky
<jacek@systinet. To: Noah Mendelsohn/CAM/Lotus@Lotus
com> cc: xml-dist-app <xml-dist-app@w3.org>
Sent by: Subject: Re: issue 168 proposal: xsi:type of external references in Encodi ng
xml-dist-app-req (fwd)
uest@w3.org
01/03/02 05:21
AM
Noah,
my intention was indeed that we specify a new simple type or
that we reuse some XSD binary simple type. The lexical
representation is not important in our case since the value is
not in an XML file. What's important is the value space - arrays
of bytes.
And I also intended ASCII text files to be typed with this type
for I don't want to specify a gazillion of types.
Again, the type I talk about above would only be a default for
the case the actual type was not specified otherwise. In this way
we can keep the requirement that every simple value is typed by
an XML Schema simple (possibly sub-) type.
Regards,
Jacek Kopecky
Senior Architect, Systinet (formerly Idoox)
http://www.systinet.com/
On Fri, 21 Dec 2001, Noah Mendelsohn wrote:
>
> Jacek Kopecky writes (in relation to hrefs to .GIFs):
>
> >> Noah, from the point of view of
> >> the XML Schema type system, the
> >> type would be something binary
> >> - with the value space of any
> >> sequence of octets and the lexical
> >> space not important since it's
> >> not exactly in an XML message.
>
> Well, my point is that it certainly won't have any schema type unless we
> write something in our specification to make it so.
>
> Furthermore, I'm quite nervous about this direction. A .gif file is
surely
> not conformant with the lexical representation of a schema binary; even
if
> we wished to associate the two, we would have to write a spec saying
how.
> If I point to an ASCII text file, does that turn into binary? string?
> string with character repertoire restricted by regex to ascii only?
binary
> for the encoded characters?
>
> This all feels like a very slippery slope to me, hence my suggestion
that
> we allow such nodes to be untyped (they are obviously NOT of the schema
> anytype, since they need not conform to any of the limitations for XML
> character sets or other restrictions that characterize XML simple and
> complex types.)
>
> ------------------------------------------------------------------------
> Noah Mendelsohn Voice: 1-617-693-4036
> Lotus Development Corp. Fax: 1-617-693-8676
> One Rogers Street
> Cambridge, MA 02142
> ------------------------------------------------------------------------
>
>
>
>
Received on Thursday, 3 January 2002 10:54:06 UTC