Hi,
I want to add a floating point renderer for the memory views to CDT (via an established committer), and there are several options for how I might organize the new code in CDT. I’m interested in feedback on a couple of options I describe
below, as well as additional ideas folks might have.
Currently, we have the org.eclipse.cdt.debug.memory.traditional plugin, which contains the Traditional Renderer for the memory views. The Floating-point Renderer (which is already written) uses many of the same classes used by the Traditional
Renderer; much of the code is the same, and some is different. When adding the new renderer I can eliminate a lot of duplicate code by carefully re-organizing the existing renderer code. There are several options that occur to me:
-
Create a new “common” project that holds the common classes/elements that both Traditional and Floating-point Renderers depend upon. This keeps the renderer-specific files in their own projects, while files common to both are neatly
containing in another project. Of course, there are project dependencies that need to be maintained.
-
Create a new project that contains all common and renderer-specific files, i.e. all renderer files go into one project. This simplifies the dependencies. However, ALL renderer files are in one project. And, as other/future renderers
are added, the number of files grow, file naming rules (e.g. FPRendering.java vs. TraditionalRendering.java) need to be followed to avoid confusion, etc.
What do think of the options above? Are there are any other ideas that the community might have?
Thanks, Andre St. Laurent (Wind River Systems, Inc.)