Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [] Notes of the preccomitting meeting

Thibault LANDRE wrote:
After reading again my previous answer, I saw that I didn't fully answer to Cédric : I don't mind renaming those plugins ;) There are here for development purpose to help developers by generating code for Papyrus. So they need to be as understandable as possible for the developpers. I think oep.codegen is ok. Codegen refers to "Code Generation" and it is commonly used.
It is not so bad, we it still missing codegen "for what ?"
For the oep.def, you're right, it is not clear enough. Why not oep.TemplateGen or oep.GenTemplate instead (or others) ?
ok, but without Capital ==> templategen.
I still maintain that we need a global namespace for all gen ...

Etienne Juliot a écrit :
My opinion:
- yes, we want to have a clean architecture, without too much element forced by framework constraints - but if there are some name conventions or plugin organisations we don't know how to do with GMF and generators constraints, lets be realist and use them, but with keeping in mind to improve them by submit feature request to these frameworks to allow to use several Template directories.
In my opinion, it is not necessary to use several template directories.
I think it is clearer to have all the templates in the same place. But you're right, we can raise a feature for this against GMF and discuss it there.
Each part or generator should be reusable individually in other projects, whithout embedding all the stuff.

But I currently lack arguments to open this feature and I don't see well what will be the benefits of it. Do you have arguments that I can push forward to initiate the discussion ?
For the moment, I see :
- possibility to add different templates coming from different project (ex: Templates from Papyrus and Templates from Moskitt Project). But this could be dangerous, I think, if the two projects work on the same template.
- a clearer organisation.
Any other ideas ?

- for IP submission, I think we can to NOT include these generatos, if there are several problems. It will not be a problem for end users.
I agree. These generators are not necessary for the submission.
- for Properties generators, as Acceleo 4 will be based on MTL (OMG's standard, like UML), I think the easiest way to commit them will be to use this new engine already included in Eclipse Foundation. But it's too early to speak about that, during this rush for IP submission.



Thibault LANDRE a écrit :
I know that Papyrus is different from GMF and from UML2Tools.

The GMF Papyrus generator are an extension of the GMF tool to simplify the work for Papyrus developers. Those two plugins will be used only by developers.

OAW is used because some templates from GMF have to be overriden and those template are done with OAW.
More, it is necessary to keep the same folder hierarchy than the bundle
"org.eclipse.gmf.codegen" to override existing templates.
Besides, to take into account the new templates defined, it is necessary to defined the property "Template Directory" with the parent folder containing all specific templates.
I don't know how to indicate more than one "Template Directory"
Additionnally, I don't think that it will change anything for the developpers to have an unique plugin well structured (templates are organized with folders int this plugin) instead of several plugins with few templates in each one...

That's some of the reasons why I think it will be difficult to divide into several plugins the generators...

For Properties, I know that generators exist. But as far as I know (because I have never seen them, they are not present on the CEA svn), they are realized with Acceleo and not OAW.



GERARD Sebastien 166342 a écrit :
I second the comment of Cedric. It is clear that Papyrus is different from the UML2 MDT Tool, is clearly that Papyrus do not force the usage of the GMF generator.

-----Message d'origine-----
De : [] De la part de Cedric Dumoulin
Envoyé : mercredi 12 novembre 2008 16:04
À : thibault.landre@xxxxxxxxxxxxxx; Papyrus Project list
Objet : Re: [] Notes of the preccomitting meeting

Thibault LANDRE wrote:

I have named the plugin like they were denominated in the GMF and UMLTools project (for the same functionnality). I have kept this naming to avoid confusion for developers familiar with GMF and UMLTools.
The idea was good, but we are not a GMF demonstrator :-). They have named their plugins like that because they are gmf centric. This is not our case. GMF is just a tool that we use to produce diagrams. The artefact used for the generation should not appear abruptly in the plugins. And for me the names of the plugins should gives clear indication of their purpose/contents. 'def' and 'codegen' are not meaningful for me. This is why they should be renamed. Also, we may have such def/codegen for properties, diagrams and others. How we will do ? Mix all the stuff in the same 'def' plugin ? I think it is a bad idea.


_______________________________________________ mailing list

_______________________________________________ mailing list
_______________________________________________ mailing list
_______________________________________________ mailing list

Back to the top