|
From: M N. <m.n...@wa...> - 2009-07-26 17:31:40
|
On Sun, Jul 26, 2009 at 7:15 PM, Esteban Aguilar B. <nab...@ya...>wrote: > > > I have always said that we need to keep the > > "LOOK AND FEEL" of wxDevC++. How we implement it > > (e.g. Tstringlist/wxArrayString) is, for ME, unimportant. As > > long as it looks like wxDevC++ and feels like wxDevC++ > > (especially the designer) then I'm a happy > > chappie. > > I agree with Mal. In fact, I think the point of wxDevIDE is the chance of > getting rid of tons of things wxDevCpp does wrong. Sure, the current code > may be helpful in understanding several things, like how the compiler, the > debugger or the designer stuff works, etc. but in my opinion it would be a > big mistake just taking the code and try to translate it to C++. Designing the IDE should be reasonably simple, since we are just replicating a lot of dialogs in the first place. And I do agree that attempting to merely translate the Delphi to C++ is fraught with problems. > > > You don't want to try to deal with the huge ass TMainForm mess again. In > fact, a good desing separates the GUI stuff from the business domain > objects; in that regard, the TMainForm implementation is a fundamentally > flawed desing. > > I can give several other examples of things wxDevCpp does wrong: for one, > event-driven sections of code are very difficult to understand. I've had > lots of pain looking for the right place in code that gets executed after > some obvious action in the GUI; there are several places that use things > like the repaint event of some component to check on the state of some other > component to trigger some action. Again, a good design requires the control > of the flow of actions to be condenced in some well defined objects in which > these tasks are clearly identified. I too found it difficult to fathom when some of the code was actually being run. > > > > > I've written a proposal on how the plugins support could be modeled. > > http://wxdevide.wiki.sourceforge.net/wxDevIDE+API?f=print > > I know it is a little confusing, but my idea could be summarized in > observing which parts of the IDE are good candidates for plugin replacement, > and implement all the comunication between them and the cotroller objects > throwgh interfaces (c++ abstract classes). > > At this point I think I would like to contribute in this area of the > project, if you guys agree. > > I say go ahead. Any hard and fast documentation is good when we are all so spread out as we are. Best regards Mal |