| [mddi-dev] [Fwd: Re: [Fwd: Re: MDDi and GMT eclipse contrib]] |
Hi all,
How about Thursday morning / noon, after the ModelBus demonstration on the Tool track?
Please let us know.
Thanks,
Sébastien
-------- Message original -------- Sujet: Re: [Fwd: Re: MDDi and GMT eclipse contrib] Date: Wed, 28 Jun 2006 11:32:17 +0530 De: vinay.vkulkarni@xxxxxxx Pour: Sébastien Demathieu <sebastien.demathieu@xxxxxxxxxxxxxxx> Copie: sreedhar.reddy@xxxxxxx
Hi Sebastien,
We are happy to find that there is a general agreement between our ideas and approaches.
We would like to go through the detailed presentation on ModelBus that you mentioned. We can have a more detailed discussion during ECMDA.
Pl do go ahead and post our document on the mddi mail thread.
Vinay and Sreedhar
Vinay Kulkarni Tata Consultancy Services Limited Mailto: vinay.vkulkarni@xxxxxxx Website: http://www.tcs.com
*Sébastien Demathieu <sebastien.demathieu@xxxxxxxxxxxxxxx>*
06/26/2006 12:53 PM
To vinay.vkulkarni@xxxxxxx cc sreedhar.reddy@xxxxxxx Subject Re: [Fwd: Re: MDDi and GMT eclipse contrib]
Hi Vinay, Sreedhar,
Thank you for the reminder. The team did have a discussion on the document and I apologize for not replying sooner.
The document is in line, in general, with our idea of a model integration bus. Your contribution to project would be very welcome.
Among the most interesting points, we noticed that you consider a MDD bus as an extension of a generic service bus. This is exactly what we aim to do. The bus we are developing is currently tightly coupled to web services but, on the long term, we want to sit on top of various technologies (WS, OSGi, other middleware), providing a loose couple between platform-independent aspects and a particular technology.
You also provide interesting new ideas, such as "model deltas". Passing deltas and not complete models as arguments is very important if we want to avoid performance issues with XMI-based tool chains. That raises a series of questions: How delta models may be calculated (delta with respect to what baseline?) and how they can be transformed as fragments?
I am wondering about the view model integration. If we consider the context of QVT, a view can be calculated and maintained thanks to a source meta-model, a target meta-model and a transformation. What may be missing is to talk about the transformation definition, which permits to have a complete specification of a view. That may need further discussion, though.
We also defined tool categories. We have a bigger number of these but I think this is not an issue here.
I guess we need to figure out a way to collaborate now. How can we proceed to move on? Would you be interested in a detailed presentation on ModelBus? Do you want to show us a particular implementation?
I noticed that you guys are giving a talk at ECMDA. That can be an excellent opportunity to have a longer discussion. What do you think about it?
Also, are you okay with posting your document and following this mail thread on the mddi mailing list (https://dev.eclipse.org/mailman/listinfo/mddi-dev). External people have additional good insight.
Thanks,
Sébastien
vinay.vkulkarni@xxxxxxx a écrit : > > Hi Sebastien, > > Just a gentle reminder. Any update? > > Vinay > > Vinay Kulkarni > Tata Consultancy Services Limited > Mailto: vinay.vkulkarni@xxxxxxx > Website: http://www.tcs.com
Attachment:
TCS_MDDI Bus.pdf
Description: Adobe PDF document