Could Tk be D's ideal widget set?

[Josh Stern]

On Mon, 09 Oct 2006 14:43:46 -0700, Walter Bright wrote:

>
> D is a multiparadigm language, and that's important for 
> library development because appealing designs shouldn't be shut out 
> because they are impractical with D. For example, C++ can't do GUI 
> toolkits that are based on gc. C++ can't do toolkits based on signals 
> and slots (Trolltech had to write a C++ preprocessor to make that work). 
> Java can't do template metaprogramming based designs. C can't do OOP 
> based designs.

Just wanted to point out in case it wasn't clear that QtJava (and I think
PyQt) are GC'd in the normal way on the application side and allow
signals/slot use via different mechanisms which are then bridged to the
C++ library side. As written, QtJava uses weak references in a hash map to
auto-delete C++ roots when their corresponding Java objects get finalized,
but there are other ways that this could be done even without weak
reference support.  If D ultimately developed a native mechanism for
runtime signals/slots then it could be used instead and bridged.