[GRASS5] XDRIVER pad list and d.rast

[Glynn Clements]

Andreas Lange wrote:

> > > after working in a GRASS session for some more time without
> > > d.erase I realized by resizing the monitor, that some
> > > 20 display commands had been accumulated (generally fine!).
> > >
> > > My question/suggestion is: If
> > > d.rast map
> > > is used, can't this be used to erase the command list in GRASS
> > > monitor? As d.rast will overwrite everything (am I wrong?) it's
> > > effectively the same as
> > > d.erase
> > > d.rast map
> > 
> > Have you never used d.rast to display more than one adjacent or
> > overlapping raster?  I think the answer to that question will
> > answer your question ;-)
> 
> The behaviour of d.rast was as Markus outlined above (calls
> D_clear_window if no -o option). 
> But i believe that this is not consistent, as no other display module
> wipes out the display. And i had problems with the window re-creation
> and d.save, because the pad structure was erased with this original
> setup. 
> And the manual page says:
> If the active graphics frame already contains text or graphics, and the
> user does not wish to use the -o option, it is wise to first invoke
> d.erase to clear the active graphics frame before running d.rast.
> 
> So i think the existing users of GRASS have to learn the new setup, and
> the new users will enjoy enhanced consistency. Just remember to run
> d.frame -e and/or d.erase to initialize the display.
> 
> But if you all disagree, i will revert my changes in
> src/display/d.rast/cmd/display.c .

Now that we're on the subject, I'm wondering why -o isn't the default
behaviour. It was understandable before the introduction of null
values, but now -o should probably be the default, possibly with a
separate option to fill null-valued cells.

Also, even without "-o", d.rast doesn't overwrite the entire display.

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