Hi Dave,
Dave Pawson wrote:
> On 24/11/06, Michael Brauer - Sun Germany - ham02 - Hamburg
>
>> Dave (Pawson) noticed that the attribute is optional, and that ODF 1.2
>> document therefore may not identify themselves as such. I agree to him
>> that the version attribute should be mandatory for ODF 1.2 documents,
>> and therefore propose that we change the description and schema as
>> follows:
>>
>> > All root elements take an office:version attribute, which indicates
>> > which version of this specification it complies with.
>> >
>> > An application that stores a document conforming to this application
>> > *shall* use the attribute value "1.2".
>> >
>> op> <define name="office-document-common-attrs" combine="interleave">
>> > <optional>
>> > <attribute name="office:version" a:defaultValue="1.1">
>> > <choice>
>> > <value>1.0</value>
>> > <value>1.1</value>
>> > <value>1.2</value>
>> > <ref name="string">
>> > <param name="pattern">[1-9]+\.[0-9]+</param>
>> > </ref>
>> > </choice>
>> > </attribute>
>> > </optional>
>> > </define>
>>
>> Some notes:
>> - In order to achieve that ODF 1.0/1.2 remain valid ODF 1.2 instances, I
>> have kept the attribute optional in the schema. The description
>> however states that the attribute actually is mandatory for ODF 1.2
>> documents.
>
> I don't agree with keeping it optional.
> That would imply forward compatibility, which is clearly impossible to
> operate.
> If that is wanted, I'd suggest wording as used in XSLT from W3C.
> If a processor for version X meets a version >X then its response is
> undefined.
Isn't the situation I'm describing "If a processor for version X meets a
version <X"? What I want to achieve with keeping the version attribute
optional is that if a 1.2 processor validates a 1.0 or 1.1 document
against the 1.2 schema, that it then does not get validation errors.
Michael
>
>
>
>
>
>> - I have kept the possibility to use arbitrary version numbers in the
>> schema. This shall allow to validate future document against the
>> schema, if they don't use any new features. I'm not sure we need that.
>> - I have removed the sentences regarding validation, because it is
>> obvious and not specific to OpenDocument that documents could be
>> validated.
>
> But the 'optional' schema value makes it unverifiable?
>
> I would like to make it easy to verify compliance to a revision of the
> spec,
Maybe I have overseen something, but if the attribute is not existing,
then a processor would have to choose the 1.1 schema. Only if it is
present and has the value 1.2, it has to choose the 1.2 schema.
> not impossible?
Michael
>
> regards
>
>
