Re: Proposed change, fundamental invariant

[Stan Shebs <shebs at shebs dot cnchost dot com>]

Hans Ronne wrote:
> 
> I ran into the same problem a long time ago with the Mac code. I put a hack
> in xform_cell_flat that bypasses this annoying error for "wide viewports"
> (meaning Macs). What this hack does is to provide a safety margin of 8
> cells around the area edge where the error is ignored. Maybe something
> similar would be a good general solution? That way you would still catch
> most garbage xforms since a random int is unlikely to fall within such a
> margin.

I think we'd ultimately have to rely on callers to do the right thing,
because a large map with large scaling will have values that are way
outside the area, possibly by hundreds of cells.

> The mac drawing code works fine with cylindrical areas now, though it took
> some blod, sweat and tears to get there. Maybe some of that stuff, such as
> the UNWRAP macros, could be generalized? Or at least reused in the tcltk
> code.

I should take a look at those, I've forgotten the theory involved...

> P.S. The isometric stuff looks nice, but could we have the solid color
> option back for those terrains that lack isometric images? Right now,
> nothing is drawn in that case.

Just fixed in the fix 5 minutes ago, along with hash tables for unit
views, so your speed problems should be solved too.

Stan