Hi Henning, hi Antoine,
I think that having structureRef as EObject is ok and should be
changed.. I have even an example where the EObject is almost correctly resolved
from XSD schema.
It is only a matter of how good our URIHandler works: It is
responsible to use the QName, find the corresponding Import, load the file and
resolve the reference.
I don’t see why we need the “import” reference
for that resolution process – the Qname in the file is sufficient.
BTW: The current implementation only work if the target
namespaces in the Import are unique. I think this is a valid assumption, even
if the only example from the Spec (the Correlation Example Seller) violates it.
Else we would need to have a global QName index, which would be quite time and
memory intentensive.
Regards,
Reiner.
Von:
mdt-bpmn2.dev-bounces@xxxxxxxxxxx [mailto:mdt-bpmn2.dev-bounces@xxxxxxxxxxx] Im
Auftrag von Antoine Toulme
Gesendet: Mittwoch, 15. September 2010 18:19
An: Henning Heitkötter
Cc: BPMN2 Developers Mailing List
Betreff: Re: [mdt-bpmn2.dev] ItemDefinition's Import is transient ?
OK I'll change the model to use
a qname for now. Thanks for confirming.
On Wed, Sep 15, 2010 at 09:03, Henning Heitkötter <hheitkoetter@xxxxxxxxxxxxxx>
wrote:
Yes, judging from the "BPMN 2.0 by Example"
document, chapter 11, that's how it's supposed to work: structureRef as a QName
with a NS prefix pointing indirectly to the import element.
2010/9/15 Antoine Toulme <antoine@xxxxxxxxxxxxxxx>
The spec has a serious problem on this: the import attribute
is not in the schema, but it shows in the table of the attributes of Item
Definition.
Our code uses StructureRef on Item Definition as an object,
while the spec mentions it should be a qname, and casts it as an EObject.
I think we should leave import as it is, set structureRef as
a qname, and later on the getImport method should be rewired to get the right
Import object that matches the qname's prefix of the StructureRef attribute.
WDYT ?
On Wed, Sep 15, 2010 at 08:50, Henning Heitkötter <hheitkoetter@xxxxxxxxxxxxxx>
wrote:
Hi Antoine,
I guess it's transient because this attribute is not supported by the XSD.
Apparently the specification does not mention a replacement - at least I
couldn't find anything. Note that a normal reference by ID is not possible in
XSD, because Import elements do not have an ID.
Regards,
Henning