|
Re: [Metamodel] XPDL [message #24950 is a reply to message #24809] |
Wed, 14 March 2007 16:37   |
Wojciech Zurek Messages: 8 Registered: July 2009 |
Junior Member |
|
|
Hi Florian,
The only problem I can see here, is that there is several way of doing the
same thing in XPDL2.
For example you can invoke a web service directly from the activity (using
TaskService etc.) or you can use an application that define the web
service - and invoke an application from the task. And I believe most models
will have a few such cases (like what is the difference between BPMN's
Recieve Task and intermediate message event).
The question here is not how to translate XPDL into AgiPro, or other way
around, that is easy, both models seams to be capable to define and hold
process definition. But if we want to support roud-trip between those
models - so if we import and then export model it will look the same. This
include vendor specyfic extension that should be persisted.
Wojtek
"Florian Lautenbacher" <lautenbacher@informatik.uni-augsburg.de> wrote in
message news:4d936fd2a4b7c947f18ba78aa74f0ae9$1@www.eclipse.org...
> Hi all,
>
> any points that we need to consider when adapting the meta-model regarding
> XPDL? As we pointed out in former discussions we try to have a "Pivot
> meta-model" which can be adapted to different representations and hence,
> we don't simply take the XPDL metamodel. But it is necessary for our
> further work to know the requirements that come within XPDL and that are
> currently missing in the AgilPro (discussion) meta-model which can be seen
> on our Wiki.
>
> So, please can anybody summarize the requirements that XPDL has?
>
> Thanks in advance,
>
> Florian
>
>
|
|
|
Re: [Metamodel] XPDL [message #25215 is a reply to message #24950] |
Mon, 19 March 2007 08:22  |
Florian Lautenbacher Messages: 61 Registered: July 2009 |
Member |
|
|
Hi Wojtek,
I totally agree with you, that vendor specific extensions need to be
captured in the current model (and thats not an easy task!).
That one thing can be achieved using different ways is something that is
not a preferable solution in my eyes. I know that many standards include
different ways to do the same thing, but I can't see an advantage by
that. We should try to keep our meta-model simple and if we can generate
one of the ways using code generation it is enough - or what do you think?
I don't think that after importing and exporting a model it needs to
look the same as before, but the execution semantics must stay the same.
Best regards,
Florian
Wojciech Zurek schrieb:
> Hi Florian,
>
> The only problem I can see here, is that there is several way of doing the
> same thing in XPDL2.
> For example you can invoke a web service directly from the activity (using
> TaskService etc.) or you can use an application that define the web
> service - and invoke an application from the task. And I believe most models
> will have a few such cases (like what is the difference between BPMN's
> Recieve Task and intermediate message event).
>
> The question here is not how to translate XPDL into AgiPro, or other way
> around, that is easy, both models seams to be capable to define and hold
> process definition. But if we want to support roud-trip between those
> models - so if we import and then export model it will look the same. This
> include vendor specyfic extension that should be persisted.
>
> Wojtek
>
>
> "Florian Lautenbacher" <lautenbacher@informatik.uni-augsburg.de> wrote in
> message news:4d936fd2a4b7c947f18ba78aa74f0ae9$1@www.eclipse.org...
>> Hi all,
>>
>> any points that we need to consider when adapting the meta-model regarding
>> XPDL? As we pointed out in former discussions we try to have a "Pivot
>> meta-model" which can be adapted to different representations and hence,
>> we don't simply take the XPDL metamodel. But it is necessary for our
>> further work to know the requirements that come within XPDL and that are
>> currently missing in the AgilPro (discussion) meta-model which can be seen
>> on our Wiki.
>>
>> So, please can anybody summarize the requirements that XPDL has?
>>
>> Thanks in advance,
>>
>> Florian
>>
>>
>
>
|
|
|
Powered by
FUDForum. Page generated in 0.01943 seconds