[GUI] Mission Statement + Widgets

[Simon Peyton-Jones]

|   I further believe that the item of primary
| importance is the user's haskell api.  I could easily
| write a backend using fltk, Qt, winapi, or whatever
| and we'd probably see that once the specification is
| finished.  The sample implimentation that is released
| with the gui-haskell standard need not work on three
| platforms, but only one.  Once it is released (if not
| before) the other backends will come quickly.  I
| suggest efforts are concentrated on the actual
| haskell-interface document as that will give the most
| bang for the buck.  Others can then impliment backends
| according to their preferences (ncurses, CORBA, COM,
| odbc?).

Yes, yes, yes. =20

I think (hope) the reason there's been so much discussion about
implementation routes (=3D back ends) is that there's no point in
specifying something that is unimplementable.  And actual prototype
implementations can feed back strongly into the specification.

But yes, the spec is the thing.

Simon