XMI Bug

[joshua]

According to UML 1.3, model elements don't have references to tagged values. But, the UML 1.3 DTD from the OMG allows ModelElement.taggedValue elements within ModelElement elements. So the DTD is incorrect.

Your XML 1.1 serialization is correct, according to the UML 1.3 DTD, but incorrect according to UML 1.3. Do you plan to address this, for example by making TaggedValue elements children of XMI.content with references to their ModelElement elements?

[angel-o-sphere]

I'm not sure if that is a bug (because I'm not sure if I understood you correct).

The dark side of the OMG is: MOF, UML and XMI nearly never have been standards which where in synch.

It regulary was the case that for the same version numbers of those standards, e.g UML 1.1 and XMI 1.1, no 1 : 1 mapping from UML to XMI  existet.

This is a big problem. Especially as tool vendors e.g. implement UML 1.4 and support XMI import for XMI 1.0 or XMI 1.1 or XMI 1.0 plus Unisys Extensions ... and so forth.

As far as I know UML 2.0 with MOF 2.0 (or eMOF ?) and XMI 2.0 will be the first standard where all aspects of modelling are unified. (That albeit where UML 2.0 is final now but MOF 2.0 still in work)

angel'o'sphere

[joshua]

Right. It's not a bug; It's a feature

But seriously, the existence of a ModelElement.taggedValue element implies that ModelElement instances have references to TaggedValue instances, which, according to UML 1.3 is not the case. Instead, TaggedValue elements should appear as children of XMI.content and they should have an attribute called 'modelElement' that has the value of the xmi.id attribute of the ModelElement element to which the the TaggedValue applies.

The whole reason I raise this is that I'm having trouble reading EA's XMI with the XMI from NetBeans MDR, which follows the UML spec quite closely.

[angel-o-sphere]

Oops, I was jsut about to forward some mails from the MDR mailing list and realized you are familiar with that tool as well. So I asume you are on the MDR list?

angel'o'sphere