[XMI] How to use multiple references

[insane79]

I'm developing an XMI interface and i have
found some problems in the utilization
of the contex attribute.

For example if i have two classes (class_1,
class_2) and i want to associate the same
state machine to both class i will wrote
the following XMI declaration

....

<UML:StateMachine xmi.id="id_123" name="foo"
contex="id_class_1 id_class_2" >

....

I've tried to import the document in rose
and poseidon but only one class own the
state machine.

With poseidon i've used the element based
version of contex.

There is someone that can help me?

Thanks

[Gene Mutschler]

The problem is that, in UML 1.3, you cannot do what you are trying to
do. The "context" aggregation has a 0..1 multiplicity on the
"aggregating" end. That is, while a model element can aggregate any
number of state machines, a particular state machine can be in no more
than one context.

The Rose addin chooses the first context link in the XMI file as the
state machine owner. It does not look at any additional context links
for that state machine. (Why do I know this? I wrote it that way.)

My guess is that any other UML-conformant modeling tool will do the
same.

The "x"s in my return address are spam preventers.
On 11 May 2004 02:41:39 -0700, rr*****@libero.it (insane79) wrote:

I'm developing an XMI interface and i have
found some problems in the utilization
of the contex attribute.

For example if i have two classes (class_1,
class_2) and i want to associate the same
state machine to both class i will wrote
the following XMI declaration

...

<UML:StateMachine xmi.id="id_123" name="foo"
contex="id_class_1 id_class_2" >

...

I've tried to import the document in rose
and poseidon but only one class own the
state machine.

With poseidon i've used the element based
version of contex.

There is someone that can help me?

Thanks

[insane79]

Gene Mutschler <gm***********@sbcglobal.net> wrote in message news:<vi********************************@4ax.com>. ..

The problem is that, in UML 1.3, you cannot do what you are trying to
do. The "context" aggregation has a 0..1 multiplicity on the
"aggregating" end. That is, while a model element can aggregate any
number of state machines, a particular state machine can be in no more
than one context.
Many thank. I was i bit confused because into the dtd the contex tag
is defined as idrefs.
The Rose addin chooses the first context link in the XMI file as the
state machine owner. It does not look at any additional context links
for that state machine. (Why do I know this? I wrote it that way.)

I've found some problem in export sequence diagrams, for examples the
lifeline are broken and the delete command is not exported, are these
known issues?

[Gene Mutschler]

On 15 May 2004 14:22:35 -0700, rr*****@libero.it (insane79) wrote:

Gene Mutschler <gm***********@sbcglobal.net> wrote in message news:<vi********************************@4ax.com>. ..

The problem is that, in UML 1.3, you cannot do what you are trying to
do. The "context" aggregation has a 0..1 multiplicity on the
"aggregating" end. That is, while a model element can aggregate any
number of state machines, a particular state machine can be in no more
than one context.
Many thank. I was i bit confused because into the dtd the contex tag
is defined as idrefs.

As I recall, any aggregation is going to be specified as idrefs, even
one with a max multiplicity of 1. This is a characteristic of XMI 1.x
(XMI 2.x might be more precise about this sort of thing, but I don't
have my reference at hand to say for sure)

The Rose addin chooses the first context link in the XMI file as the
state machine owner. It does not look at any additional context links
for that state machine. (Why do I know this? I wrote it that way.)
I've found some problem in export sequence diagrams, for examples the
lifeline are broken

This is a known problem with Rose--the script interface has no means
to query or set lifelines..

and the delete command is not exported,

I can look into this.

are theseknown issues?

[Gene Mutschler]

On 15 May 2004 14:22:35 -0700, rr*****@libero.it (insane79) wrote:

Gene Mutschler <gm***********@sbcglobal.net> wrote in message news:<vi********************************@4ax.com>. ..

The problem is that, in UML 1.3, you cannot do what you are trying to
do. The "context" aggregation has a 0..1 multiplicity on the
"aggregating" end. That is, while a model element can aggregate any
number of state machines, a particular state machine can be in no more
than one context.

Many thank. I was i bit confused because into the dtd the contex tag
is defined as idrefs.

The Rose addin chooses the first context link in the XMI file as the
state machine owner. It does not look at any additional context links
for that state machine. (Why do I know this? I wrote it that way.)

I've found some problem in export sequence diagrams, for examples the
lifeline are broken and the delete command is not exported, are these
known issues?

I finally figured out that by "delete", you mean the symbol for ending
a time line. Unfortunately, like the other aspects of timelines, this
information is not available to the Rose XMI export.