Field of view does not always include internal points near diagonal jumps of the circular boundary

[felix-clark]

The field-of-view algorithm traces out a circle and scans lines to the center point. The problem is that this can miss points adjacent to the boundary that should be within the field-of-view.

#55 provides an alternate version of the Bresenham circle that jumps either in x or y (not both) to include these interior points in the set of locations to scan from, and makes some corresponding changes to tests. I did not want to replace the classic Bresenham algorithm, but it doesn't appear to be used elsewhere in the library so it might make sense to remove it and only use the new version.

[Zireael07]

How about a picture/graph explaining what do you mean?

[felix-clark]

For instance, the upper left quadrant for the FOV with range 5 appears to be (where X are the visible squares):

   XXX
  X XX
 X XXX
X XXXX
XXXXXX
XXXXX@

The blank squares inside should be within the view radius, but happen to not quite be included in any of the lines scanned from the edge to the center.

The suggested solution I linked is to change the circle traced out from this (now the edge is marked with # for emphasis):

   ###
  # XX
 # XXX
# XXXX
#XXXXX
#XXXX@

to this:

   ###
  ##XX
 ##XXX
##XXXX
#XXXXX
#XXXX@

so now these points are included in the visible range.

[thebracket]

I just merged in the PR you sent: #55
It should be fixed now?

[felix-clark]

Indeed this appears to be fixed in 0.6.1. Thanks for the quick review!