mlhim-owners team mailing list archive

[Luciana Tricai Cavalini <lutricav@xxxxxxxxx>]

Hi mlhim-owners,

I am a strong supporter of having a single source for the MLHIM
Reference Model and the CCD Editor, because it is not sustainable in the
long run to implement the XSDs for the Reference Model and then make the
same changes on the XMind CDD template file.

While Tim is alive it will work, because he is so knowledgeable and
perfectionist, and let's pray for his health, but we know that MLHIM 2
is just like the heliocentric cosmological system: it will only be
understood and adopted in a century. We don't have The Inquisition
anymore, but we have something stronger: The Capital (in the form of
traditional software companies and other lobbies).

I was here thinking about my tiny experience with ecore models and how
to edit them after importing the XSDs (I made some exercises with my
colleagues from UFPR early this month about that). Then I had some
spontaneous thoughts that I would like to share with you, because
someone might extract something useful from them (I can't).

I would like to remember everybody that I am a physician with no
computational background. Everything I've been learning about computer
science I owe to Tim and Prof. Sergio. That's why my ideas sound just
like intuitions, because that's what they really are.

Software developers enjoy developing software based on ecore/acceleo
models. They were trained for that. It makes sense to them. So, having
the MLHIM Reference Model implemented as an ecore model is something
good to attract software developers.

But MLHIM-based software is not only code for the Reference Model. We
need to have the CCDs transformed in code as well. And now I think I
understand that, if your Reference Model is implemented in Python, your
CCDs need to be implemented in Python too. And so on: Java-Java,
Ruby-Ruby etc. And they need to be connected somehow into an
application. Not to mention the database underneath, but let's forget
about databases for a while.

OK. And the CCDs don't invent anything: they are combinations and
constraints of the classes of the Reference Model. So, I have this
intuition that every CCD will be another ecore/acceleo model, derived
from the Reference Model ecore/acceleo model.

My tiny experience told me that I can add packages, classes and other
things to the model. I can edit them too (which allows constraints). So,
probably I can copy and paste classes. The problems are: (1) can I do
that with the complicated hierarchy of the MLHIM (and openEHR too)
specs? and (2) can I copy and paste a certain class as many times as
required to express a given healthcare/socio-demographic/geographic concept?

The tool is powerful; so, I have this intuitive feeling that the answer
is yes for both questions. It is just a matter of our skills to work on
Eclipse and Acceleo. One can tell that it can't be done because Eclipse
is written in Java and what the MLHIM specs require to be implemented is
not possible to be done in Java. OK, if that's the case, we have the
most complicated computational problem ever and then we will need a new
"Eclipse", written in a language that allows us to do what we need. But
when I think about it, my intuition tells me that I am trying to cure
somebody's dandruff with decapitation.

So, in my Intuitive Paradise, we will have the MLHIM Reference Model
implemented in ecore/acceleo, and one model for each CCD, all of them
derived from the Reference Model ecore/acceleo model. We will upload
everything to HKCR, and software developers will pick up the models,
generate code on whatever OO language, and build MLHIM-based
applications that share information extracts from one side to another,
and everybody will be happy (some will be rich and famous too).

I hope reading that was at least entertaining.

Cheers,

Luciana.


Em 25/06/2011 22:35, Timothy Cook escreveu:
> Hi All,
> 
> Luciana and the team at UFPR found a possible solution to exporting
> ECore models to other languages such as Pythn, XSD, Ruby, PHP, etc.
> It is a project called Acceleo.   Reading the information about
> Acceleo is very promising.
> 
> Acceleo is a template based meta-model and can write (based on a
> template) any format desired.
> 
> However, attempts at installing Acceleo and it's dependencies into an
> existing Galileo (3.5) eclipse that is in the Ubuntu repositories lead
> myself and others into dependency hell.
> 
> The solution I have found is to manually install a new copy of the
> latest version of Eclipse Indigo (3.7).
> 
> I have reimported the MLHIM2 schemas and there are errors according to
> EMF capabilities so the model will need to change accordingly.
> After correcting the Ecore model I will begin to explore Acceleo.
> 
> I invite anyone to interested to join in a conversation here about
> moving (back) to using Eclipse EMF as a foundational modelling
> framework for MLHIM.
> 
> Here are the steps I have taken to install Eclipse Indigo.
> 
> Eclipse for MLHIM Install on Unbuntu 11.04
> ===========================================
> 
> The Ubuntu repositories have the 3.5 (Galileo) release.
> These steps are for a manual install of 3.7 (Indigo).
> http://www.eclipse.org/downloads/
> 
> Download (for your platform):
> Eclipse Modeling Tools
> 
> This package contains framework and tools to leverage models: an Ecore
> graphical modeler (class-like diagram), Java code generation utility
> for RCP applications and the EMF Framework, model comparison support,
> support for XSD schemas, OCL and graphical modeler runtimes. It
> includes a complete SDK, developer tools and source code.
> 
> Unpack the archive into your home directory and it will create a
> directory named 'eclipse'.
> There is an executable name 'eclipse' in that directory.
> 
> On first execution, click on Help->Check for updates
> 
> Install Acceleo according to instructions here:
> http://www.eclipse.org/acceleo/download/#installUpdate
> 
> Restart as suggested.
> 
> If you are going to be doing Python development then install PyDev as
> described here:
> http://pydev.org/manual_101_install.html#installing-with-the-update-site
> I did not install Mylyn Integration
> 
> There is an Acceleo Qucik Start and other tutorials here:
> http://www.eclipse.org/acceleo/documentation/
> Acceleo is the big game changer for using Eclipse or not.  It is worth
> exploring.
> ============================================================================
> 
> You can then get the mlhim-specs which includes an EMF project:
> bzr branch lp:mlhim-specs
> 
> Then in Eclipse go to
> File->Import->General->Existing Projects into Workspace
> 
> I think that should work.
> 
> 
> Cheers,
> Tim
> 
> 
> 
> 

-- 
Esta mensagem foi verificada pelo sistema de antivírus e
 acredita-se estar livre de perigo.