From: Chris C. <ca...@al...> - 2003-01-18 17:35:08
|
On Saturday 18 January 2003 16:49, Guillaume Laurent wrote: > On Saturday 18 January 2003 17:29, Richard Bown wrote: > > The point is - do we trust Qt? > > Yes. For two reasons : [...] Oh, I have no problem "trusting" Qt. I just don't want to use Qt in=20 our core library. We use the Standard Library for everything else in=20 there, rightly I think, so why not just stick with standard strings? Actually (I've mentioned this before) I'd really like to refactor=20 somewhat so as to introduce at least one more library layer between=20 base and gui, to contain things like notation and layout-related=20 classes that are not actually specific to the Qt GUI. There are lots=20 of classes in gui that don't necessarily need to be there. I would=20 also probably avoid using Qt in this library. It's got nothing to do with the availability of the code, etc., it's=20 just a matter of design principles. I do not believe that a core=20 library that's fundamentally non-GUI-related should depend on a large=20 GUI library just because one of the developers thinks it'd be nice=20 not to use the standard (standard!) C++ string class. I don't care=20 whether the only application for that core library happens to be in a=20 GUI environment, the point still stands. Frankly, I'm also motivated partly by irritation at the fact that I=20 can't just use my perfectly good std::strings directly with Qt=20 classes. I know exactly why that is, but it's still exasperating,=20 and I don't like feeling coerced into changing good design into bad=20 to accommodate a defect of a GUI library. Chris |