But you can parse it.

If it's just the validation, you still have ambiguitities e.g., suppose I form a union of two types, each derived from xs:int, but with different DFDL representations. There's a conflict then, how is that resolved. Rules for "consistency" among two DFDL annotations will be complex. E.g., is an int with a 31 bit representation compatible wiht an int with 32 bit representation? - shouldn't be unless both are unsigned. Etc.

...mike

From: Steve Hanson [mailto:smh@uk.ibm.com]
Sent: Monday, March 09, 2009 4:46 AM
To: DFDL
Cc: dfdl-wg@ogf.org
Subject: Re: [DFDL-WG] Numbers with multiple ranges

It's a validation issue. I have a packed decimal number with ranges 0-100, 200-300, 400-500. I can't parse and validate this with DFDL. WTX can handle this today.

Regards

Steve Hanson
Programming Model Architect
WebSphere Message Brokers
Hursley, UK
Internet: smh@uk.ibm.com
Phone (+44)/(0) 1962-815848

DFDL <mbeckerle.dfdl@gmail.com>

07/03/2009 04:49

To	Steve Hanson/UK/IBM@IBMGB
cc	"dfdl-wg@ogf.org" <dfdl-wg@ogf.org>
Subject	Re: [DFDL-WG] Numbers with multiple ranges

What does this have to do with representation/ format ?

Seems totally out of scope to me.

...mikeb

On Mar 6, 2009, at 12:18 PM, Steve Hanson <smh@uk.ibm.com> wrote:

One of the existing non-XML parsers from IBM is capable of modelling an integer or decimal with non-contiguous ranges, eg, 0-100, 200-300, 400-500. It is also possible to model this using XML Schema, using a user-defined simple type which is a union of simple type restrictions.

However it is not possible to model this in DFDL because unions are not supported. A choice could be used, but that would give different named elements in the resultant infoset depending on the value, which is not ideal.

I believe this is use case that we should consider for DFDL 1.0. We could decide to allow unions, but with a constraint that the members of the union had to be restrictions of the same built-in type, and that clashing DFDL properties on the members was an error. I believe the existing scoping rules for visiting types would still apply ok. No annotation appears on the xs:union only on the xs:simpleType as currently. The net effect is that our parsing rules are the same as currently, but we gain validation capability.

I am not in favour of more widespread support for unions. I don't have a compelling use case for union of different logical types, eg, xs:dateTime and xs:string.

Your thoughts are welcome.

!-- integer range 0 - 100
<xsd:simpleType name="intRange1">
<xsd:restriction base="xsd:int">
<xsd:maxInclusive value="100"/>
<xsd:minInclusive value="0"/>
</xsd:restriction>
</xsd:simpleType>

!-- integer range 200 - 300
<xsd:simpleType name="intRange2">
<xsd:restriction base="xsd:int">
<xsd:maxInclusive value="300"/>
<xsd:minInclusive value="200"/>
</xsd:restriction>
</xsd:simpleType>

!-- integer range 400-500
<xsd:simpleType name="intRange3">
<xsd:restriction base="xsd:int">
<xsd:maxInclusive value="500"/>
<xsd:minInclusive value="400"/>
</xsd:restriction>
</xsd:simpleType>

!-- Union of above types
<xsd:simpleType name="intRange">
<xsd:union memberTypes="intRange1 intRange2 intRange3"/>
</xsd:simpleType>

!-- Union of anonymous types being restrictions of above types
<xsd:simpleType name="intRangeLocal">
<xsd:union>
<xsd:simpleType>
<xsd:restriction base="intRange1"/>
</xsd:simpleType>
<xsd:simpleType>
<xsd:restriction base="intRange2"/>
</xsd:simpleType>
<xsd:simpleType>
<xsd:restriction base="intRange3"/>
</xsd:simpleType>
</xsd:union>
</xsd:simpleType>

!-- Union of anonymous types being local restrictions
<xsd:simpleType name="intRangeLocalRestrictions">
<xsd:union>
<xsd:simpleType>
<xsd:restriction base="xsd:int">
<xsd:minInclusive value="0"/>
<xsd:maxInclusive value="100"/>
</xsd:restriction>
</xsd:simpleType>
<xsd:simpleType>
<xsd:restriction base="xsd:int">
<xsd:minInclusive value="200"/>
<xsd:maxInclusive value="300"/>
</xsd:restriction>
</xsd:simpleType>
<xsd:simpleType>
<xsd:restriction base="xsd:int">
<xsd:minInclusive value="400"/>
<xsd:maxInclusive value="500"/>
</xsd:restriction>
</xsd:simpleType>
</xsd:union>
</xsd:simpleType>

Regards

Steve Hanson
Programming Model Architect
WebSphere Message Brokers
Hursley, UK
Internet: smh@uk.ibm.com
Phone (+44)/(0) 1962-815848

Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU

--
dfdl-wg mailing list
dfdl-wg@ogf.org
http://www.ogf.org/mailman/listinfo/dfdl-wg

Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU