| Re: [mdt-ocl.dev] MDT/OCL Architecture wiki |
|
Hi Ed, I've finally read your architecture proposal for the next release. In general, I'm not sure which is the figure of the pivot model. It's an axuxiliary metamodel ? At the same level of the our - Generic - Abstract Syntax Metamodel ? It's a replacement of our - Generic - Abstract Syntax Metamodel ?. I guess that this doubt is also in your mind after reading the in-parenthesis sentence in the "Pivot AST Model" section. It also seems that there will be different kind of pivot models... Are they coming from the same Pivot -Meta-model ? I think that a general picture of the the different -meta- models (including the StdLib figure) in their different modeling layers could help. Apart from that, I some times got confused by the usually ambiguity of the modeling terminology or by usual terminology in the modeling context.... words such as "references", "defined", "based on", "declaration", "feature", "derivation", "realization" are not sufficiently auto-explanatory in the context they are used, or at least I'm not 100% sure about what they refer.... just guessing. Since this document should be crucial for the future of this component, I think that there should not be any doubt about it. I mean all of us (including clients) should have a clear idea of the architecture which is in your mind ;)....Anyway, I know that this is just a first draft, which should be evolved during the next release. Some other thoughts: - In the "Xtet integration" when you say "Xtext CST is considerably more accurate than the LPG CST" you are referring that XText CST is better aligned or resembles to the OMG's Concrete Syntax definition ? - I'm not sure about your plans concerning how the API would evolve having two parsers. Will the current API be deprecated, so that we create and evolve a new API using the ANTLr parsing infrastructure ? Would the current API "shift" in somehow to use the said Pivot models ? This could imply a hard adoption efforts by dependant projects such as QVTo..... Will the ANTLr parser only be used by the Editors, which would imply maintaining and evolving two different parsers ? P.S: After our contact with the Dresden-Team I've found in google a couple of interesting presentations which also mention the concept of the "Pivot Model". When you have time, you could have a look to the following links: http://st.inf.tu-dresden.de/Ocl4All2007/slides/braeuer_demuth.pdf http://dresden-ocl.sourceforge.net/gbbraeuer/presentations/final_presentation/final_presentation.pdf Cheers, Adolfo. El 21/05/2010 12:04, Ed Willink escribió: Hi --
|
