Another option I found was to say that a GPL like EER can also be used as an extensible modeling language, to create domain specific versions of the GPL. 

Hi Sonny,

I would not completely agree with the statement "EER is a General Purpose Language (GPL)": EER is used for data modeling (and is an extensions to the original entity-relationship (ER) model, because it introduces subclasses und superclasses). It is more a DSL for (general) data modeling purposes. So yes, it is used "in general" for data modeling, but it is specific for data modeling. The UML (that consists of various diagram types) covers a lot of aspects for software specification and can be described as a GPL. UML class models and EER are quite similar. If you ask me, I would prefer UML class models because they the UML is MOF-based (s. also EMF als a meta-meta model, M2), so is easy to use it in an Eclipse-based Model Driven Development (MDD) approach (OVT and MOFM2T implementations are available).

I hope that helps you with the Sirius project that you plan. Since there are already tool implementations for UML (e.g. Papyrus), maybe you should use a language that is not completely implemented like IFML. That would be great.

Regards

Roland

I do agree with Roland and his suggestion.  Best of good luck with your research.

Hi Sonny,

I also agree with all the definitions given by Roland.

I think that implementing an EER modeling tool is a good pretext to play with DSM tools like Sirius. The EER metamodel is quite simple to implement (drastically much simpler than the UML metamodel). You will be able to explore more Sirius (and other EMF frameworks) functionalities with a simple metamodel like EER.

EMF is a wonderful framework to implement DSLs (Domain Specific Languages) like EER. There is a rich ecosystem of EMF-based eclipse projects to develop your Domain Specific components (diagram based or textual based modelers, file and document generators, validation rule engines, advanced persistence systems, etc., depending of your tool expected functionalities). You can mix great frameworks like Sirius, xtext, cdo and many others to implement all aspects of your DSM tool about EERs.

It is a good subject... Have fun !

Regards

Romain

Thank you very much for your answers, they help me a lot. 

Romain, would you have any idea if a theoretical metamodel exists for EER with the class diagram notation? I have found the EERMM from Robson Fidalgo, but it is specifically designed for the Chen notation. 

Thank you very much in advance.

Hi Sonny,

I don't know about a recognized standard for the EER metamodel... MOF and UML have emerged to kill the debate about a myriad of EER, OO, EER-OO modeling tools who claimed to be the right and ultimate methodology to develop softwares. EER models are suited to describe relational database, but it is a very limited aspect of a complete software.

There is no implementation in the zoo metamodels collection (http://web.emn.fr/x-info/atlanmod/index.php?title=Ecore).

You can find inspiration for your EER metamodel in the OMG's KDM metamodel (http://www.omg.org/spec/KDM/1.4/PDF/), chapter 18 'Data Package'...

You could also have a look on Modisco (https://www.eclipse.org/MoDisco/). It supports many metamodels to represent legacy software artifacts. Not sure, but I suppose they could have an implementation for relational databases.

Considering EER diagrams, I can definitely say that there is no standard available. You will have to implement your own diagram editor, and decide the way you want to represent your EER models on diagrams.

Regards,

Romain

Thank you very much for your answers! 

I am a little bit in a corner with the EER being an DSL. 

My professor (who will eventually evaluate me) insisted that EER is a GPL, while it does seem more logical that it's a DSL. So I tried saying that it's a DSL with the characteristics of a GPL for data modeling in my paper. I'm not sure if that is correct though. 

Should I just state that it's a DSL? If so, is there a resource that I could use to prove my case?

Thank you very much in advance.

EER is one ancestor of UML. Some versions of EER embark OO concepts, and are somehow equivalent to the "class diagram" part of the UML metamodel. I can understand that before UML, it was used, and could have been considered, as a GPL for describing software... You can generate most of a software code with the "class diagram" concepts of UML...

But now that UML exists, with its broader-scoped metamodel, EER seems to be at least a very skimpy GPL metamodel, that is mostly suited to data description (which is only one particular aspect of a software)... ;-)

"Would you have any idea if a theoretical metamodel exists for EER with the class diagram notation?"

Follow:

• Dimitrieski et al., "Concepts and Evaluation of the Extended Entity-Relationship Approach to Database Design in a Multi-Paradigm Information System Modeling Tool", 2015 - http://booksc.org/book/45731611/600c43
• Dimitrieski et al., "Extended Entity-Relationship Approach in a Multi-Paradigm Information System Modeling Tool", 2014 - https://annals-csis.org/Volume_2/pliks/239.pdf