[GRASS5] g51 v.digit

[Glynn Clements]

Radim Blazek wrote:

> > It *is* just a hack. But you don't need to wait 5-10 years. Removing
> > the R_get_location_with* stuff altogether shouldn't take more than
> > 5-10 minutes.
> >
> > And removing it *is* the right solution for the monitor API. For
> > v.digit, the right solution is to use a real UI toolkit. Trying to
> > build something like v.digit on top of R_get_location_with* is a
> > mistake.
> 
> Which portable UI toolkit is or should be used for GRASS?

Will Tk not suffice?

I'm hesitant to suggest wxWindows due to its use of C++, although
having C++ calling C is usually less problematic than the reverse.

> > Of all of the possibilities which I've considered for a future display
> > architecture, none of them deal with input in any way; it's meant to
> > be a *display* architecture.
> 
> Do you mean different systems for display and v.digit?

No, I mean that the display system only handles display, not input. 
Much the same way that e.g. OpenGL and Display PostScript only deal
with display, not input.

Programs which need user input would use the display mechanism for
display, and handle the input themselves.

> So that if I have
> displayed map and discover that I need to edit it, I have to start 
> some other application with new display? That is not good from user point
> of view. It must be one system built on UI toolkit and that is what we have
> to wait for 5 years.

This problem already exists. The only real solution is an overall user
interface solution (tcltkgrass doesn't really cut it), which is
inherently a hard problem.

-- 
Glynn Clements <glynn.clements at virgin.net>