Having thought this through I think that you are correct. If the content length can never be 0 then EVDP does not apply.  I think IBM DFDL is issuing an error when it should not be.

Regards
 
Steve Hanson
Architect, IBM DFDL
Co-Chair, OGF DFDL Working Group
IBM SWG, Hursley, UK
smh@uk.ibm.com
tel:+44-1962-815848



From:        Mike Beckerle <mbeckerle.dfdl@gmail.com>
To:        Steve Hanson/UK/IBM@IBMGB,
Cc:        "dfdl-wg@ogf.org" <dfdl-wg@ogf.org>
Date:        09/09/2014 15:37
Subject:        Re: [DFDL-WG] Catch 22.for EVDP & NVDP ?

---

Interesting.

To me if lengthKind is explicit, and non-variable (i.e., is fixed length), then EVDP doesn't apply, because by definition, such a representation is not ever empty.

I would assume that EVDP would be ignored in such a case, and an implementation might even generate warnings if dfdl:emptyValueDelimiterPolicy appeared directly on such an element or simpleType where a fixed length is also expressed.

This is one of the challenges that having lengthKind 'explicit' which handles both the fixed length, and variable-length using an expression, creates. Even though we don't have a lengthKind 'fixed', we still have fixed-length as a concept, and some other features interact with fixed length, like empty values.


Mike Beckerle | OGF DFDL Workgroup Co-Chair | Tresys Technology | www.tresys.com
Please note: Contributions to the DFDL Workgroup's email discussions are subject to the OGF Intellectual Property Policy


On Tue, Sep 2, 2014 at 10:31 AM, Steve Hanson <smh@uk.ibm.com> wrote:
        <xs:element name="choiceGroup" dfdl:lengthKind="implicit">
                <xs:complexType>
                        <xs:choice dfdl:choiceLengthKind="implicit" dfdl:initiatedContent="yes">
                                <xs:element name="myNum1" type="xs:int"  dfdl:lengthKind="explicit"
                                                                dfdl:length="2"
                                                                dfdl:textNumberPattern="#0"
                                                                dfdl:initiator="num:"
                                                                dfdl:terminator="~" />
                                                                 >
                                <xs:element name="myText1" type="xs:string" dfdl:lengthKind="explicit"
                                                                dfdl:length="2"
                                                                dfdl:initiator="text:"
                                                                dfdl:terminator="~" />
                        </xs:choice>
                </xs:complexType>
        </xs:element>

The above model is a choice of two elements, each of which is fixed length and has an initiator and terminator. Because the elements are fixed length, I must set dfdl:emptyValueDelimiterPolicy (EVDP) to be 'none' **.  I also want to discriminate using the initiator, so I set dfdl:initiatedContent 'yes' on the choice, so I must set EVDP to 'initiator' or 'both' ++.  (Same for NVDP if nillable). Catch 22.

** 'none' because the only valid empty representation is that there is nothing there at all. The element is fixed length X, so if the delimiters are present then the content must be length X and can't be zero length.

++ 'initiator' or 'both' because an initiator must always be present when dfdl:initiatedContent 'yes'.

The implication is that I can't use dfdl:initiatedContent for this choice. While I can switch off nillable to avoid the error with NVDP, there is no other way of avoiding the EVDP error.

Regards
 
Steve Hanson
Architect, IBM DFDL
Co-Chair, OGF DFDL Working Group
IBM SWG, Hursley, UK
smh@uk.ibm.com
tel:+44-1962-815848
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
  https://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