Skip to main content


Eclipse Community Forums
Forum Search:

Search      Help    Register    Login    Home
Home » Eclipse Projects » GEF » GEF, EMF and semantic vs. graphical model
GEF, EMF and semantic vs. graphical model [message #159319] Wed, 24 November 2004 12:46 Go to next message
Eclipse UserFriend
Originally posted by: usenet.jevopi.de

Hi,

I've followed some of the threads concerning GEF and EMF integration.
I'm glad that other people have the same problems with two
implementations of the Command pattern and other things (e.g. the
naming problem with ConnectionEditPart.getTarget() and
Adapter.getTarget()).

Are there any concrete approches yet integrating these two frameworks?
Frank Budinsky wrotes about a prototyping work with "support for
generating GEF views in a generated EMF editor". I think this would be
very interessting.

IMHO this idea is related to a more general problem: adding graphical
information to domain (or semantic or business) models. The Red Book
mentions three approches solving that problem:

1) Create a modified version of the (domain) model with graphical information
2) Use two separate models
3) "Use modelling techniques" to make a link between the two models

I'm not sure if I understand point 3) since it appears to me as if 1)
and 2) are modelling techniques as well -- and the example given for 3)
seems to me pretty much the same as the frist approach.

I'm wondering if there exists any work about this issue. For example a
generator for GEF views requires at least two information:
a) A mapping from model elements to nodes or connections
b) Information about how to get and set layout constraints

The "UML Diagram Interchage" (UMLDI) is an approach for answering the
question for UM2 models (and maybe others as well). I'm currently
developing a small class diagram using (my own) EMF based
implementation of UMLDI in combination with UML2. I'm running into a
lot of problems with this solution, and I'm wondering if there exist
any work about how to combine domain model and graphical information.
Besides that, I'm not too happy with UMLDI, since the specification
isn't finished yet and too many questions are left unanswered.

Are there any (other) works on that issure? If there are any works on
integrating GEF and EMF, is this a topic there? Do you know samoe
articles about this subject?

Jens

PS: Is it called "semantical model" or "semantic model".. (or graphical
vs. graphic model)? Too many problems everywhere %-)
PPS: Maybe I should have posted this article in the EMF newsgroup, but
I thought it's more a graphical related problem, isn't it?
Re: GEF, EMF and semantic vs. graphical model [message #159469 is a reply to message #159319] Sat, 27 November 2004 21:29 Go to previous messageGo to next message
Pratik Shah is currently offline Pratik ShahFriend
Messages: 1077
Registered: July 2009
Senior Member
https://bugs.eclipse.org/bugs/show_bug.cgi?id=37716

"Jens v. P." <usenet@jevopi.de> wrote in message
news:co1vtb$kr8$1@www.eclipse.org...
> Hi,
>
> I've followed some of the threads concerning GEF and EMF integration.
> I'm glad that other people have the same problems with two
> implementations of the Command pattern and other things (e.g. the
> naming problem with ConnectionEditPart.getTarget() and
> Adapter.getTarget()).
>
> Are there any concrete approches yet integrating these two frameworks?
> Frank Budinsky wrotes about a prototyping work with "support for
> generating GEF views in a generated EMF editor". I think this would be
> very interessting.
>
> IMHO this idea is related to a more general problem: adding graphical
> information to domain (or semantic or business) models. The Red Book
> mentions three approches solving that problem:
>
> 1) Create a modified version of the (domain) model with graphical
information
> 2) Use two separate models
> 3) "Use modelling techniques" to make a link between the two models
>
> I'm not sure if I understand point 3) since it appears to me as if 1)
> and 2) are modelling techniques as well -- and the example given for 3)
> seems to me pretty much the same as the frist approach.
>
> I'm wondering if there exists any work about this issue. For example a
> generator for GEF views requires at least two information:
> a) A mapping from model elements to nodes or connections
> b) Information about how to get and set layout constraints
>
> The "UML Diagram Interchage" (UMLDI) is an approach for answering the
> question for UM2 models (and maybe others as well). I'm currently
> developing a small class diagram using (my own) EMF based
> implementation of UMLDI in combination with UML2. I'm running into a
> lot of problems with this solution, and I'm wondering if there exist
> any work about how to combine domain model and graphical information.
> Besides that, I'm not too happy with UMLDI, since the specification
> isn't finished yet and too many questions are left unanswered.
>
> Are there any (other) works on that issure? If there are any works on
> integrating GEF and EMF, is this a topic there? Do you know samoe
> articles about this subject?
>
> Jens
>
> PS: Is it called "semantical model" or "semantic model".. (or graphical
> vs. graphic model)? Too many problems everywhere %-)
> PPS: Maybe I should have posted this article in the EMF newsgroup, but
> I thought it's more a graphical related problem, isn't it?
>
Re: GEF, EMF and semantic vs. graphical model [message #159806 is a reply to message #159319] Tue, 30 November 2004 16:53 Go to previous messageGo to next message
Eclipse UserFriend
Originally posted by: none.us.ibm.com

> I've followed some of the threads concerning GEF and EMF integration.
> I'm glad that other people have the same problems with two
> implementations of the Command pattern and other things (e.g. the
> naming problem with ConnectionEditPart.getTarget() and
> Adapter.getTarget()).

You should ask why getTarget is on Adapter. Adapter is the interface used
*by* the TARGET. Surely the target already knows about itself. If a
specific adapter needs to allow clients to reach in to the EMF object, that
adapter should provide the unwrap() method on the interface which the
adapter is exposing to clients (the actual interface being adapted).

> Are there any concrete approches yet integrating these two frameworks?
> Frank Budinsky wrotes about a prototyping work with "support for
> generating GEF views in a generated EMF editor". I think this would be
> very interessting.

Both teams are very busy. This is not likely to happen anytime soon.
Personally I question whether this is really more productive, or if it just
gets people interested in the technologies, then they have to learn the
generated code as well.

> PS: Is it called "semantical model" or "semantic model".. (or graphical
> vs. graphic model)? Too many problems everywhere %-)

I use the terms "business model" and "view model" . "semantic model" is the
same as business, but it take me longer to process that term in my head :-P
Re: GEF, EMF and semantic vs. graphical model [message #159970 is a reply to message #159806] Wed, 01 December 2004 11:31 Go to previous messageGo to next message
Eclipse UserFriend
Originally posted by: usenet.jevopi.de

On 2004-11-30 17:53:59 +0100, "Randy Hudson" <none@us.ibm.com> said:

>> I've followed some of the threads concerning GEF and EMF integration.
>> I'm glad that other people have the same problems with two
>> implementations of the Command pattern and other things (e.g. the
>> naming problem with ConnectionEditPart.getTarget() and
>> Adapter.getTarget()).
>
> You should ask why getTarget is on Adapter. Adapter is the interface used
> *by* the TARGET. Surely the target already knows about itself. If a
> specific adapter needs to allow clients to reach in to the EMF object, that
> adapter should provide the unwrap() method on the interface which the
> adapter is exposing to clients (the actual interface being adapted).

I didn't want to start a discussion about the adapter pattern here :)
I've already posted a message in the EMF newsgroup
(Pattern trouble: Adapter, Observer, Extension Object, Date: Mon, 18
Oct 2004) where I tried to understand the EMF implementation of the
adapter pattern. IMHO Adapter.getTarget() should be called
Adapter.getSubject() (using the observer pattern role name) or
Adapter.getAdaptee() (using the adapter pattern role name). I also
think that the Adapter interface should be splitted into two
interfaces, a NotifierListener (or something) and the Adapter
interface. But I don't think that this interface is going to be
refactored, is it?
>
>> PS: Is it called "semantical model" or "semantic model".. (or graphical
>> vs. graphic model)? Too many problems everywhere %-)
>
> I use the terms "business model" and "view model" . "semantic model" is the
> same as business, but it take me longer to process that term in my head :-P

OK. Since I'm not a native speaker, it makes not difference to me using
"semantic", "domain" or "business" :)
Do you know any works (atricles, patterns) on combining business and
view model?

Maybe it would be a nice idea of creating a, maybe EMF/UMLDI based,
layer (or framework) handling the view model. Using some kind of
mapping mechanism this layer could be connected to the business model
and the EditParts. I'm wondering if it's possible to move all view
model related stuff into this framework and provide some super classes,
e.g. ConstraintChange commands for moving the GraphNodes around.

Jens
Re: GEF, EMF and semantic vs. graphical model [message #160026 is a reply to message #159970] Wed, 01 December 2004 17:02 Go to previous message
Eclipse UserFriend
Originally posted by: none.us.ibm.com

I have seen a generic EMF-based diagram model implemented in the past and it
was not successful. I don't know of any ongoing efforts for this type of
work. It doesn't fall under the scope of the GEF project charter. The VE
project might include a base view model like what you are suggesting.

> OK. Since I'm not a native speaker, it makes not difference to me using
> "semantic", "domain" or "business" :)
> Do you know any works (atricles, patterns) on combining business and
> view model?
>
> Maybe it would be a nice idea of creating a, maybe EMF/UMLDI based,
> layer (or framework) handling the view model. Using some kind of
> mapping mechanism this layer could be connected to the business model
> and the EditParts. I'm wondering if it's possible to move all view
> model related stuff into this framework and provide some super classes,
> e.g. ConstraintChange commands for moving the GraphNodes around.
>
> Jens
>
Previous Topic:GEF and RCP - Where do they meet?
Next Topic:Weird problem with ICellValidator
Goto Forum:
  


Current Time: Fri Apr 26 12:42:31 GMT 2024

Powered by FUDForum. Page generated in 0.02894 seconds
.:: Contact :: Home ::.

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

Back to the top