[Freeciv-Dev] Re: normalize_map_pos and invalid map positions

["Ross W. Wetmore" <rwetmore@xxxxxxxxxxxx>]

Wrong ... regular is *NOT* a proper subset of real.

Consider the iso-earth map with wrapping in only E_W direction.

The usual normal set is a pi/4 rotated rectangle. The regular 
set is the square that just contains all these points, i.e. 
in addition to the corners that touch the X and Y axis, the
remaining two corners just touch the upper bound X and Y
lines parallelling the rectangular axes.

This leaves 4 triangular sections inside the regular square
but outside the normal rotated rectangle.

Two of these correspond to unnormalized coordinates in the 
E-W direction, but the other two are unreal coordinates off
the N-S polar edges.

So assuming "regular == real" just sets your understanding level
back to that of the bad old void tile days.

Regular is a worse artificial designation than normal, as it 
corresponds to nothing more than a hack to try and establish
rectangular boundaries around a normal set with no regard to
any important properties or charateristics.

Cheers,
RossW
=====

At 07:00 PM 02/08/15 +0200, Raimar Falke wrote:
>On Thu, Aug 15, 2002 at 11:43:39AM -0500, Jason Short wrote:
>> Per I. Mathisen wrote:
>> > On Tue, 13 Aug 2002, Ross W. Wetmore wrote:
[...]
>> In theory the set of "normal" positions could include (-1,-1).  I think 
>> we all agree that it should not, though.
>
>Yes. Any normal position is a regular position (see
>common/map.h:regular_map_pos_is_normal). So we have
>
>real positions are a (proper) subset of all map positions
>  are equal if wrapping in both axises is enabled
>regualr positions are a (proper) subset of real positions
>  are equal if no wrapping is enabled
>normal positions are a (proper) subset of regualr positions
>  are equal in "classic" (non-iso) maps
[...]
>       Raimar
>--