![]() |
Telelogic Rhapsody (steve huntington) | ![]() |
Topic Title: Reverse engineering interfaces Topic Summary: Created On: 10-Oct-2006 06:48 Status: Read Only |
Linear : Threading : Single : Branch |
![]() |
![]()
|
![]() |
|
I am designing a system that contains a mix of hand-written and automatically generated code. The system constitutes an extensible framework that can be customised by users. It defines a bunch of interfaces that users have to implement. As a fact of life, some of the users will work in the Rhapsody environment, while others will work directly in C++.
Consequently, I want to define interfaces in hand-written C++ to make them more human-readable, and use them from Rhapsody as external elements. This works ok except that reverse engineered interfaces appear as regular classes, rather than interfaces in Rhapsody, which means that I can't use them for example to describe port contracts. I can manually change the stereotype to <<interface>>, but then if I modify the interface in C++ and import it to the model again, it overrides my setting. So, my question is: Is there a way to tell Rhapsody to treat a class as an interface during reverse engineering? And if not, is there a way to use classes instead of interfaces in port contracts? |
|
![]() |
|
![]() |
|
Do you reverse engineer after the new changes, or roundtrip? I would think that once you've set up the classes/interfaces in Rhapsody, auto generate the code... that as long as you edit the classes between the //#[ and //#] markers, I would think when you roundtrip the changes in, it wouldn't change the stereotype. (I know this works when I choose a customized stereotype, but never actually tried it for interfaces, but it should do the same thing).
|
|
![]() |
Telelogic Rhapsody
» Rhapsody Category » Rhapsody
»
Reverse engineering interfaces
|
![]() |
FuseTalk Standard Edition v3.2 - © 1999-2009 FuseTalk Inc. All rights reserved.