On Tue, 22 Jan 2008, Graeme Geldenhuys wrote:
> On 22/01/2008, Mattias Gärtner <[EMAIL PROTECTED]> wrote: > > About: move from LCL to widgetset > > That was the goal of lazarus from the beginning. > > OK, I get that and respect the choice. I'm simply wondering (from a > personal point of view) if it's still the right way of doing things? > Considering you have years of experience with Lazarus development... > If you could do it (Lazarus LCL) over again, what would you change? > Hindsight is a awesome thing. :-) > > > Keep in mind that using the native widgets has several advantages: > > - native look. Even when user switches theme or OS. > > As I mentioned before. This is very easy to achieve - Trolltech has > proved this. Simply ask the native widget to draw itself on a memory > bitmap. All native toolkits send out a message when the theme > changes, so it's very easy to detect that as well. As far as I know, you cannot ask all widgets to do this ? > > - widgetset specific goodies: e.g. tab menu of gtk notebook, unicode input > > method, assistive technology, hardware acceleration, network support (X > > client/server modell). > > All this can be implemented in a custom drawn toolkit. At least that > way all platforms will have these features. Currently only GTK > Notebook has tab menu for example. LCL now needs to support a basic > set of features which are common to all widget sets - nothing more > otherwise it's not compatible across widget sets. Drawing bitmaps is very slow, so asking the widgetset to draw on a bitmap and then send this bitmap to the X server would be VERY slow indeed, bypassing the whole idea of a widget set in the first place. Try and work remotely with an application that does all custom drawing. Not easy. Everything has its ups and downs. But the nice thing about Lazarus is that it can use for instance FPGUI, which will work on all platforms, hence rendering the whole discussion moot. Michael.
