Albert Wagner wrote:
> The project is surely desirable and you've
> convinced me that it may be doable.  But it is simply to too large and
> complex.  Enthuisiasm can get you started, but only grit and persistence will
> finish it.  This is a project that can easily become a life's work.

I thought of it that way, and it's taken several years to grieve over its
commercial failure. But now I believe it can be done *much* more easily,
perhaps as little as two to three months hard work. My QTML prototype used
the expat XML parser, the Qt toolkit, and I was just about to start embedding
the Mozilla Javascript engine when I discovered Ruby. I'm still learning Ruby
and won't decide on the optimum model for a while. The final piece in the
jigsaw was to build local and remote (using SOAP encoding) asynchronous
message delivery services and the back-end API implementations.

Qt was a nuisance, since many of the layout attributes are stored on hidden
objects which attach a widget to a position on the geometry manager. The
attribute-based QTML language needs to see those as attributes, there's no
place for method calls in setting up this kind of stuff. Actually I thought
that the geometry attributes required by the children could be modelled as
namespace extensions, but the area of XML namespaces is a bit grubby so I
left that undecided for now. I haven't learnt enough about FOX and FXRuby to
know whether it's better, but the lack of Unicode support is a worry.

--
Clifford Heath