[GSoC] Container proposals by Ishan and Christian

Jonathan M Davis jmdavisProg at gmx.com
Sun Apr 3 15:36:26 PDT 2011


On 2011-04-03 15:11, Matthias Pleh wrote:
> On 2011-04-03 23:48, spir wrote:
> > (Meaning a standard GUI-programming interface for D.)
> 
> Yes, exactly! Just a really small Api to abstract the OS.
> - Application -> as a representation for the process
> - Window -> create, show, hide
> - Canvas/Image -> to draw on
> + No buttons/textbox or other widgets, this would be part of the higher
> level
>    (some GUI's on this basic API could use native controls / others may
> draw it itself / or draw it with GPU ...
> 
> I mean, there were several attempts to create a GUI, each with a
> different approach, and most of them are unfinished and abandoned.
> This could be a base for all GUI's
> 
>  > A fairly big project, I guess! (even for a toy...)
> 
> It haven't to be complete. Just a start, with the most important corner
> cases.

 I do _not_ think that generic GUI layer is a good idea. GUI toolkits all act 
very differently about a lot of things. Abstracting them is not easy and tends 
to result in you losing a lot of their power. It's better to just use the 
appropriate GUI toolkit and be done with it. Only a _very_ simple GUI 
application would ultimately be very well served by a generic GUI layer. Not 
to mention, that's just adding yet _another_ layer of abstraction on top of 
who knows how many layers of abstraction, and D is supposed to be a _systems_ 
programming language. Extra layers of abstraction generally harm efficiency, 
and in this case, they don't really help the programmer all that much in 
writing their application anyway. You just can't abstract diverse GUI toolkits 
in an efficient or particularly useful manner. Only the really simplistic 
stuff would work.

What we need is for projects such as qtd to get the proper support that they 
need to make progress and be completed.

- Jonathan M Davis


More information about the Digitalmars-d mailing list