• From: Olivier Rossel <olivier.rossel@g...>
• To: Michael Kay <mike@s...>
• Date: Mon, 11 Jan 2010 16:16:24 +0100

well, i must admit i am puzzled.
as far as i was understanding, a substitutionGroup was supposed to
create a OR between an existing ***group*** definition, and a new
***group*** definition.
(like the |= operator in RelaxNG).

but it seems that substitutionGroup only applies on an ***element***,
so its definition mimicks the definition of another element.

my conclusion at the moment is: substitutionGroup has nothing to do
with altering a definition in a newer iteration of a schema.

did i miss something?

On Mon, Jan 11, 2010 at 3:55 PM, Olivier Rossel
<olivier.rossel@g...> wrote:
> thank you so much, michael !
>
> does my idea of massive usage of substitutionGroup sound like a good
> idea, to modularize and ease the patching of my schemas in the future?
> (given the fact that i am stuck in a XSD 1.0 environment).
>
> On Mon, Jan 11, 2010 at 3:50 PM, Michael Kay <mike@s...> wrote:
>>>
>>> Unfortunately, this schema is invalid : It is illegal to
>>> create an empty "optionalAttribute_elements" group.
>>> What is the correct way to define a group that contains
>>> nothing? or just text?
>>>
>>
>> <xs:group name="optionalAttribute_elements">
>>  <xs:sequence/>
>> </xs:group>
>>
>> Regards,
>>
>> Michael Kay
>> http://www.saxonica.com/
>> http://twitter.com/michaelhkay
>>
>>
>

• References:
  • Modular schemas.
    • From: Olivier Rossel <olivier.rossel@g...>
  • RE: Modular schemas.
    • From: "Michael Kay" <mike@s...>
  • Re: Modular schemas.
    • From: Olivier Rossel <olivier.rossel@g...>