Hi,

at present I'm writing my diploma thesis about MDA and model to code
transformation. The main tool I use for the practical part of my work is
AndroMDA. Today I want to start a little discussion about a "weakness" (in
my eyes) of AndroMDA.

I read that AndroMDA is meant to be a Tool, that transforms a PIM to
(Java-)Code. Apart from different definitions of a "PIM" in my opinion the
PIM AndroMDA uses could be more independent than it is. In AndroMDA I must
define the exact datatypes I want to use. For example an Integer must be
placed in a package structure (/java/math/Integer) equivalent to the Java
structure. If I (theoretically) want to switch to C++ and Corba, I would
have to change my whole model. I would prefer a type mapping according to
the mapping of jdbc and sql types. For now I solved the problem by defining
a dependency in my model from my abstract datatype "int" to a datatype
called java.math.Integer and so on. Not the best way I could imagine but
quite simple.

Secondly I want to mention the fixed way where my files are generated. The
file name can consist of pre-defined parts and exactly 4 variables. They
stand for (full) package name, class name, template name and stereotype
name. There is no way to select any different parts for the file name like
other attributes or dependencies. My suggestion would be to write a
vsl-script to generate the file name. Sounds a little bit like overkill, but
AndroMDA and its users are using these scripts anyway so it looks quite easy
to do.

Thanks for your annotations

Niko



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
Andromda-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/andromda-devel

Reply via email to