Eclipse Community Forums: EMF » [Eclipse 2019-03 EMF.Editor] : regression ?

Search

Help

Login

Home » Modeling » EMF » [Eclipse 2019-03 EMF.Editor] : regression ?(EMF Generation process doesn't create all source code it used to. Some EMF editors features are absent.)

Show: Today's Messages :: Show Polls :: Message Navigator

Switch to threaded view of this topic

Create a new topic

Submit Reply

[Eclipse 2019-03 EMF.Editor] : regression ? [message #1806394]

Wed, 08 May 2019 12:03

Didier Garcin is currently offline

Friend

Messages: 68
Registered: April 2013

Member

I don't know why but while my EMF.edit and EMF.editor are installed as plugin, my example wizard item doesn't appear.

I note that some java file in EMF.editor are not generated (nevertheless, all compile well) as in the past. Extensions are missing too.
For example,
presentation/*EditorAdvisor.java (new)
presentation/*EditorPlugin.java

but not :

presentation/*ActionBarContributor.java
presentation/*Editor.java
presentation/*ModelWizard.java

Absence of ModelWizard could explain but why is it absent ?

What's wrong, please ?

[Updated on: Wed, 08 May 2019 14:30]

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806408 is a reply to message #1806394]

Wed, 08 May 2019 14:51

Ed Merks is currently offline

Friend

Messages: 33137
Registered: July 2009

Senior Member

It sounds as if you've accidentally you changed your GenModel's Runtime Platform property from IDE (the default) to RCP (or RAP).

Ed Merks
Professional Support: https://www.macromodeling.com/

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806415 is a reply to message #1806408]

Wed, 08 May 2019 16:43

Didier Garcin is currently offline

Friend

Messages: 68
Registered: April 2013

Member

Yes, I've corrected and now, presentation/*EditorAdvisor.java has deseappered.
But,
presentation/*ActionBarContributor.java
presentation/*Editor.java
presentation/*ModelWizard.java
are still missing.

I've tried to redo from scratch. And same result.

FYI, the .ecore and .genmodel files, here, attached.

I can't understand what happens.

Attachment: types.ecore
(Size: 2.11KB, Downloaded 131 times)
Attachment: digital.ecore
(Size: 1.90KB, Downloaded 139 times)
Attachment: digital.genmodel
(Size: 4.48KB, Downloaded 161 times)

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806416 is a reply to message #1806415]

Wed, 08 May 2019 17:20

Ed Merks is currently offline

Friend

Messages: 33137
Registered: July 2009

Senior Member

You seem to have a UML2-based GenModel. It's best to ask about that on the UML2 forum:

https://www.eclipse.org/forums/eclipse.modeling.mdt.uml2

Ed Merks
Professional Support: https://www.macromodeling.com/

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806418 is a reply to message #1806416]

Wed, 08 May 2019 17:59

Didier Garcin is currently offline

Friend

Messages: 68
Registered: April 2013

Member

Humm, strange ... sounds like being a defeact from EMF generator as it doesn't perform generation as usual and without error.

[Updated on: Wed, 08 May 2019 18:00]

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806421 is a reply to message #1806418]

Wed, 08 May 2019 18:38

Ed Willink is currently offline

Friend

Messages: 7655
Registered: July 2009

Senior Member

Hi

There is something wrong with digital.uml or your conversion thereof. Without the relevant file, it is hard to comment usefully.

When you write that it used to work was that before UML 2.5, because the problem seems to be that you have created a clone of the standard (new in UML 2.5) Primitive Types package, which might account for the subsequent trouble.

Regards

Ed Willink

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806425 is a reply to message #1806421]

Wed, 08 May 2019 19:42

Ed Merks is currently offline

Friend

Messages: 33137
Registered: July 2009

Senior Member

If there is a problem with the EMF generator itself I'll need to see such a problem occur (a test case) without having to install UML2's generator extensions to reproduce it. And of course I use EMF's generator in many other places, e.g, to generator Oomph's setup editor so a problem as glaringly obvious as you describe would not be easily overlooked.

Ed Merks
Professional Support: https://www.macromodeling.com/

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806493 is a reply to message #1806421]

Thu, 09 May 2019 20:16

Didier Garcin is currently offline

Friend

Messages: 68
Registered: April 2013

Member

Ed Willink wrote on Wed, 08 May 2019 18:38

Hi

There is something wrong with digital.uml or your conversion thereof.

Conversion is performed by 'New EMF generation model wizard'.

Ed Willink wrote on Wed, 08 May 2019 18:38

When you write that it used to work was that before UML 2.5, because the problem seems to be that you have created a clone of the standard (new in UML 2.5) Primitive Types package, which might account for the subsequent trouble.

types imported (through 'New EMF generation model wizard') are all of type EDataType.

And here is the original uml file.

Attachment: digital.uml
(Size: 10.51KB, Downloaded 157 times)

[Updated on: Thu, 09 May 2019 20:17]

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806495 is a reply to message #1806425]

Thu, 09 May 2019 20:21

Didier Garcin is currently offline

Friend

Messages: 68
Registered: April 2013

Member

[Updated on: Thu, 09 May 2019 21:58]

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806502 is a reply to message #1806495]

Fri, 10 May 2019 03:33

Ed Merks is currently offline

Friend

Messages: 33137
Registered: July 2009

Senior Member

If I create a new GenModel direct from your *.ecore files, indeed no editor is generated. But that's because there are no insubstantial ECasses in either model. All the EClasses you do have are marked as Interface = true so they don't even generate implementation classes. There's nothing to edit if there's nothing that can be created to edit, so no editors. Perhaps there is a regression in UML2 to Ecore conversion, though from the names (even in the digital package), this just looks like a bunch of "data type" definitions to me, so there's nothing you could possibly edit, and even programmatically, there are no EObjects you can create.

Ed Merks
Professional Support: https://www.macromodeling.com/

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806515 is a reply to message #1806502]

Fri, 10 May 2019 06:34

Ed Willink is currently offline

Friend

Messages: 7655
Registered: July 2009

Senior Member

Hi

I think you've confused us by titling the message "regression".

There seems to be absolutely nothing wrong; everything is working as intended.

However you are being caught by the EMF's surprisingly high bar between interesting/boring. It has caught me out before too, and it takes some time to debug why generation works for one model and not another.

The solution is to add a dummy 'interesting' Class so that your 'boring' Interface / Enumeration content is not ignored.

Regards

Ed Willink

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806524 is a reply to message #1806515]

Fri, 10 May 2019 07:16

Ed Merks is currently offline

Friend

Messages: 33137
Registered: July 2009

Senior Member

It's not an issue of interesting versus boring it's an issue of there being any object that can be created and hence edited. After all, the wizard to create a resource for the model needs to put some root object in that resource, but there aren't any. Of course a dummy class could be edited, but then the editor only edits a dummy thing and nothing else from these models are used by that dummy thing, so it's hard to imagine how that would be useful. This just all looks bogus to me.

Ed Merks
Professional Support: https://www.macromodeling.com/

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806536 is a reply to message #1806524]

Fri, 10 May 2019 09:05

Ed Willink is currently offline

Friend

Messages: 7655
Registered: July 2009

Senior Member

Hi

IIRC I hit the problem when generating a model that only had enumerations. Nothing was generated so another model that referenced them hit missing classes. A dummy class fixed the limitation.

In this case, I suspect that another model whose editor tries to use this model may be disappointed that some of the edit support is missing.

Regards

Ed Willink

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806597 is a reply to message #1806502]

Sat, 11 May 2019 09:51

Didier Garcin is currently offline

Friend

Messages: 68
Registered: April 2013

Member

Ed Merks wrote on Fri, 10 May 2019 03:33

If I create a new GenModel direct from your *.ecore files, indeed no editor is generated. But that's because there are no insubstantial ECasses in either model.

To say, no concrete class. OK. I totally agree with you. I've missed that.

Because no editor source code is generated, it means that somewhere in the EMF generator , a control-flow instruction controls creation of such java files depending on the presence of concrete class, I imagine something like :

if (!concreteClasses.isEmpty())
// Create *Editor.java, ...

It would be nice, as usual in software, at the contrary, to notify user by a warning, there is no concrete class in the model :

if (!concreteClasses.isEmpty())
// Create *Editor.java, ...
else
// Emit a warning : "no concrete class => no editor"

A good idea ?

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806598 is a reply to message #1806394]

Sat, 11 May 2019 10:45

Didier Garcin is currently offline

Friend

Messages: 68
Registered: April 2013

Member

I've corrected my UML model by replacing uml:Interface by uml:Class.

I must confess I'm surprised by the resulting Ecore model.
For instance, I can't understand why UML composite relationships are not renewed in the Ecore model (they are totally ometted).

Suggestions ?

Attachment: digital.uml
(Size: 7.21KB, Downloaded 70 times)
Attachment: digital.ecore
(Size: 2.68KB, Downloaded 107 times)

[Updated on: Sat, 11 May 2019 10:46]

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806602 is a reply to message #1806598]

Sat, 11 May 2019 12:02

Ed Merks is currently offline

Friend

Messages: 33137
Registered: July 2009

Senior Member

The generator just generates. It doesn't report to the user about all the optional /conditional things that aren't created nor do I plan to rework the generator framework to produce a report of what it did.

And when it comes to UML2, Ed W might help, but this in not the UML2 forum an I'm far from an expert in UML2 to comment on what it should or shouldn't do.

Ed Merks
Professional Support: https://www.macromodeling.com/

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806610 is a reply to message #1806602]

Sat, 11 May 2019 18:38

Ed Willink is currently offline

Friend

Messages: 7655
Registered: July 2009

Senior Member

Hi

If you want to continue this correspondence, it is really time that you moved to the UM2 newsgroup as Ed M directed, in his second response.

Regards

Ed Willink

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806632 is a reply to message #1806610]

Sun, 12 May 2019 21:28

Didier Garcin is currently offline

Friend

Messages: 68
Registered: April 2013

Member

Doesn't seem to be a UML2 defeact but EMF generator one.
I've understood...
My understanding is that UML -> Ecore translation is experimental and at the contrary, Ecore -> UML seems working better (I can make a mistake too).

As a result, I give up by this way.

Thank you.

Have a nice week.
Regards

[Updated on: Sun, 12 May 2019 21:58]

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806638 is a reply to message #1806632]

Mon, 13 May 2019 04:59

Ed Merks is currently offline

Friend

Messages: 33137
Registered: July 2009

Senior Member

No, UML -> Ecore is not experimental, but you're welcome to believe that without bothering to ask in the appropriate forum about it. Although why you take that approach is beyond my understanding.

There are a great many technologies that extend the basic ones provided by EMF, with code generation starting from a UML2 model being just one such example. I can't know about them all nor investigate all issues relating to all downstream technologies, but there is a forum where you can ask, but you just don't seem to want to go there for no apparent good reason.

Ed Merks
Professional Support: https://www.macromodeling.com/

Report message to a moderator

Re: [Eclipse 2019-03 EMF.Editor] : regression ? [message #1806639 is a reply to message #1806638]

Mon, 13 May 2019 05:19

Ed Willink is currently offline

Friend

Messages: 7655
Registered: July 2009

Senior Member

Hi

UML to Ecore is far from experimental.
UML is a standard model specification language and so is more powerful paradigm than EMF. UML offers more opportunities for misunderstandings. UML has no direct implementation. UML to Ecore has many conversion options.
Ecore is a very practical model implementation language that can also be used as a specification.
If you just want an implementation then Ecore may well be a better choice.

Regards

Ed Willink

Report message to a moderator

Switch to threaded view of this topic

Create a new topic

Submit Reply

Previous Topic:	Cross Reference in Ecore Editor
Next Topic:	Serializing attribute defined ad EDataType

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

]

Current Time: Fri Apr 19 23:34:40 GMT 2024

Powered by FUDForum. Page generated in 0.04309 seconds

.:: Contact :: Home ::.

Powered by: FUDforum 3.0.2.
Copyright ©2001-2010 FUDforum Bulletin Board Software

Back to the top