MDA synchronisation skip attributes in target if they were deleted

[Gary Cleal]

What I would like to be able to do is delete attributes of a class, after that class has been created via a transform.
But then, if I run the same transform again from the source, then I do not want the attributes that were deleted in the target to re-appear.
On the other hand, attributes in the target that have been altered in the source with a new name or type for example, should propagate to the target.

I am trying to create a common logical model, which is transformed into a set of interfaces that use selective subsets of the common model, but where common attributes should remain synchronized and common across all the target interfaces.

Is there a way to manipulate the template language for transforms to be able to achieve this result?

Thanks

[Helmut Ortmann]

Hi,

have you looked in Tools, Options, Source Code Engineering, Attributes/Operations?

There you can control some behaviour without changing any transformation.

As far as I know transformations are a nice thing to forward generate. You can easily adapt them. The reverse engineering is as far as I am aware of hard coded. You can of course play with the settings in Options of the language.

Helmut

[qwerty]

Think about writing your own transformation via the API since the transformation script language is very limited. Pro: Much more flexible and you can use a language you understand. Con: You have to start at scratch.

q.

[Helmut Ortmann]

Hi,

there is also the possibility to define your own Grammer with EA. May be here is someone who has experiences with this way.

Helmut

[qwerty]

The grammars are used to reverse engineer a language, but they are not used during transformation. The good thing with those grammars is that there is a way to add other than just the "standard" languages. The drawback is simply that defining a grammar for common languages is extremely time consuming. Only if you define one for e.g. Brainfuck, you will be able to do that in an hour or so. But I guess the number of serious Brainfuck applications is just zero.

q.

[Glassboy]

Obviously Okk! would be a more useful grammar to implement