[pivot] Provide EMFv Constraint providers for UML profiles and Complete OCL documents

[eclipse-ocl-bot]

| --- | --- | | Bugzilla Link | 397519 | | Status | NEW | | Importance | P3 enhancement | | Reported | Jan 06, 2013 07:19 EDT | | Modified | Mar 02, 2013 15:24 EDT | | Depends on | 397518 | | Reporter | Ed Willink |

Description

Papyrus use the EMFv framework for its validation.

In order for OCL constraints within UML profiles to be executable in UML models an EMFv ConstraintProvider is required that provides the profile constraints to the EMFv framework.

A similar provider can be used for Complete OCL documents.

[eclipse-ocl-bot]

By Ed Willink on Jan 06, 2013 11:16

Plausible code has been pushed and built on the bug/389120.

Rigorous testing is difficult until Papyrus validation has a coherent philosophy; see Bug 397518.

[eclipse-ocl-bot]

By Adolfo Sanchez-Barbudo Herrera on Jan 31, 2013 12:07

Hi Ed,

Some comments to this bugzilla which looks like you tend to integrate by M5.

Making OCL depend on EMFv ? In the case which it were completely necessary (which I don't know), shouldn't this contribution go in its own plugin rather than in the pivot one ? Going further ... in any other different project ?.... Seeing those extensions (plugin.xml) at the pivot plugin makes my eyes scratch. BTW where is the EMFv dependency ?

Before you elaborate any argument to reply, I recall your sensible comments to Bug 371024:

"b) Currently EMFv depends on OCL rather than OCL on EMFv, so any contribution using EMFv should be to the OCL extension of EMFv."

A similar provider can be used for Complete OCL documents.

AFAI understand it, is not that provider the contribution offered at Bug 371024 ?

Cheers,\ Adolfo.

[eclipse-ocl-bot]

By Ed Willink on Jan 31, 2013 12:57

Unfortunately MDT/UML2 introduced the EMFv optional dependency.

My contribution of an extensible Validations View to EMFv means that I'll probably be an EMFv committer. Maybe the EMFv OCL functionality should migrate to MDT/OCL as I offered a couple of years ago.

MDT/OCL -> MDT/UML2 -> EMFv -> EMF core is a sensible dependency order.

[eclipse-ocl-bot]

By Adolfo Sanchez-Barbudo Herrera on Jan 31, 2013 13:09

(In reply to comment #3)\

MDT/OCL -> MDT/UML2 -> EMFv -> EMF core is a sensible dependency order.

No. +1 projects depending on a +2 project is not sensible at all [1]...

We could escape right now, providing the pivot project is in examples (OCL Tools +3), but we both know that the pivot MM will have to be part of the OCL Core +1 stuff.... It could make more sense in its own plugin, similarly to what Franhofer guys did.

I'm not sure about MDT/UML2 but I'm not sure they hace a similar justification... Adding Kenn to the discussion

Cheers,\ Adolfo.

[1] http://www.eclipse.org/projects/releases/releases.php?release=kepler

[eclipse-ocl-bot]

By Kenn Hussey on Jan 31, 2013 13:23

(In reply to comment #3)

Unfortunately MDT/UML2 introduced the EMFv optional dependency.

I don't see why this is unfortunate. It's certainly not the only case of such a dependency; even EMF has optional dependencies on things like DTP, for example.

[eclipse-ocl-bot]

By Adolfo Sanchez-Barbudo Herrera on Jan 31, 2013 13:37

(In reply to comment #5)\

I don't see why this is unfortunate. It's certainly not the only case of such a dependency; even EMF has optional dependencies on things like DTP, for example.

To me also is. If EMFv changed any API you depend on between +1 and +2 during a milestone week, which is totally free to do it, I'm not sure if the aggregator feel good with it... Or at least, at runtime your would have some unusable functionality.... I think that the reasoning of the +1, +2, +3 time frames, is that.... establishing some reasonable dependencies among official projects, or ?

Cheers,\ Adolfo.

[eclipse-ocl-bot]

By Ed Willink on Mar 01, 2013 15:22

Need to verify the optional org.eclipse.emf.validation dependency.

[eclipse-ocl-bot]

By Adolfo Sanchez-Barbudo Herrera on Mar 02, 2013 15:24

For better or worse the dependency is already in M5. My project lead and one PMC member are aware of the issue so my game ends here.

Hopefully, what Ed claims at comment 3 will happen sooner rather than later. Currently we have cyclic dependency between OCL-UML-EMFv projects

Regards,\ Adolfo.