I'm learning more about WPS with time, but I still think that is a valid approach for creating the resources that manage the OS/2-eCS desktop.
The Re-usability idea
When you create a new class to create a new object on WPS you don't start from scratch. You have several other classes that you can re-use to build your class (WPClass, WPObject...all)
Look for example the HWman class replacement. The goal of this was to improve some issues that the original "Hardware Manager" has. The author did not start creating a "Hardware Manager" by scratch. He made a "child" of the hardware manager class (HWManager) and replaced (overide) only the functionality that has the issues. So in fact he coded less.
You can also expand the functionality of objects on the WPS. By creating a "child" of the classes you can expand
Just look again the HWman class replacement. The author fixed the behavior of a close source program. I had never saw that, in general if you don't have the source code of the program, you can not expand it or improve it.
This is something that was never seen in the "close source software" world. If you want to add a feature to Windows ScreenSaver for example, you will have to write a complete replacement for it. While with OS/2 WPS you can copy the close sourced class and only code the feature missing.
But I also think there are some downsides:
- We depend on the old IBM somcompiler. (sc.exe)
- We need to learn how WPS works before improving it / replacing it with OSS components.
- Having this flexibility also grows the complexity. We need to have and understanding on how are the WPS Classes structured.
For the moment my actions are simple on this subject:
- Trying to include more documentation about WPS on the EDM/2 Wiki
- I'm trying to document the UML visual design of WPS Classes (attachment included).
If anybody wants to help me with this starting point I will appreciate it. It does not matter if only have some minutes at week.
My UML is very rusty.
- UML in PNG format.
- Argo UML file