I'm in two minds about the problem we have with Sinelabore.
I can't fault Michiels decision to apply a Well-formedness rule but I
do feel some sympathy for Peter at Sinalbore who has based his
application on the faulty XMI that we used to generate.
I wouldn't like anything to delay our 0.32.1 release which is due for
tomorrow. Please go ahead with that anyway Linus no matter what
direction this thread takes us.
Peter has supplied me some XMI from other tools to indicate that these
tools structure the XMI in the same way as ArgoUML used to. However
these look to me to be from UML2 tools rather than UML1.4. For UML1.4
I think Michiel is correct.
My suggestion would be that we add a runtime argoument to argo such as
the following
java -Dsinelabore=true -jar argouml.jar
Anyone planning on using the Sinelabore tool can do so with this
switch and all others will get the correctly formatted XMI.
I think this should just be a short term work around while Sinelabore
allow UML1.4 to use context to link the statemachine to its
class/operation instead. We can't be expected to support these feature
forever.
One thing to consider in Sinelabore would be to preprocess the XMI
with a stylesheet to move any statemachine as it wishes before it does
its current processing.
Regards
Bob
On 21 February 2011 08:09, Michiel van der Wulp
<[email protected]> wrote:
> On 02/20/2011 08:16 PM, Bob Tarling wrote:
>> Michiel - are you sure about the changes made? It would appear to be
>> correct to me but I'm no expert in this area.
>
> Yes, I am.
>
> It took me 4 years before I finally implemented the fix, because I
> feared consequences like the one we are having now - not because I was
> unsure about the validity of the defect.
>
> The 0.32 version of ArgoUML is setting the initial namespace of the
> statemachine according a WFR. As a workaround for the problem with the
> Sinelabore tool, you can move the statemachine and events in the
> namespace of the class (I did not test this).
>
> I located a reference to an old message from Eugenio A. about the
> subject, mentioning this:
>
> «So, a StateMachine is not owned by the class that is its context. The
> StateMachine is owned by a Package. It just has a relationship to the
> class.»
>
> See: http://tech.groups.yahoo.com/group/argo-dev/message/2028
>
> Kind regards,
> Michiel
>
> ------------------------------------------------------
> http://argouml.tigris.org/ds/viewMessage.do?dsForumId=450&dsMessageId=2706062
>
> To unsubscribe from this discussion, e-mail:
> [[email protected]].
> To be allowed to post to the list contact the mailing list moderator, email:
> [[email protected]]
>
------------------------------------------------------
http://argouml.tigris.org/ds/viewMessage.do?dsForumId=450&dsMessageId=2706639
To unsubscribe from this discussion, e-mail:
[[email protected]].
To be allowed to post to the list contact the mailing list moderator, email:
[[email protected]]
