[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [Newsgroup Home]
[News.eclipse.technology.packaging] Re: Modeling package 
Eric,

EMF is still using its own local version of JET rather than reusing the JET project's version.

I personally find myself extremely frustrated when I debug and can't see code for everything on the stack. So I think that excluding the SDK likely makes sense for most packages, but probably doesn't make so much sense for tools used to generate code that extends a framework.


Eric Rizzo wrote:
Ed Merks wrote: 
Rich,

I certainly wouldn't be happy to exclude incubating components. I think it should be sufficient that any component that's conforming to the Ganymede rules be eligible for inclusion in the modeling package. There are an awful lot of valuable and cool EMFT components that should be included and it would be silly to exclude the UMLTools...

Doesn't EMF "core" use JET for its code generation already? In other words, if you're already including EMF doesn't that mean you are getting JET (even though it is <1.0)?


The SDK issue is a bit of a tricky one too. Source code is awfully useful when such a large percentage of the users will be generating code that extends the framework and will likely find themselves handicapped without the framework source code.

I'll second that - I would never want to work on an EMF-based project without having the EMF code handy. Debugging would be nigh impossible.

Eric




Richard Gronback wrote:
We’ve been discussing a Modeling package for a while, so as suggested, I’m posting here and to the Modeling newsgroup to move ahead with its proposal and definition.

In accordance with the proposed policies (http://dev.eclipse.org/newslists/news.eclipse.technology.packaging/msg00100.html), <http://dev.eclipse.org/newslists/news.eclipse.technology.packaging/msg00100.html%29,> it seems we should create a package that includes features shipping under the “Models and Model Development” category of Europa/Ganymede (and their dependencies). This would be:

    * EMF Runtime
    * EMF SDO
    * XSD
    * EMF Data Integrity Frameworks (Query/Transaction/Validation)
    * JET
    * GMF
    * EODM
    * OCL
    * UML2
    * UML2 Tools


To comply with policy 6a, which requires a 1.0 release or later, we’d need to exclude JET, EODM, and UML2 Tools. The most concerning aspect of this is the loss of UML2 Tools, which would mean the Modeling package would have no UML diagramming. How firm is this requirement? Note that we may also add components to Ganymede before the M4 cutoff on December 14th (e.g. M2M QVTO and M2T Xpand).

I suspect to keep with policy 3, which requires keeping the packages as small as possible, we should not include SDK features. Does this present a problem to anyone?

Another point to consider are those “modeling” features found outside of the Modeling project. For example, STP has a BPMN diagramming feature. Should we look to include all end user modeling/diagramming features in this package?

Thanks,
Rich