[Freeciv-Dev] Re: Corecleanup patch updates

[Jason Dorje Short <jshort@xxxxxxxxxxxxx>]

Mike Kaufman wrote:
> On Sun, Aug 12, 2001 at 11:52:24PM -0400, Jason Dorje Short wrote:

> > To repeat: not even all rectangles allow for independent normalization
> > of X and Y coordinates.  The orthogonal system proposed a few weeks ago
> > was a simple rectangle tilted at a 45 degree angle so that it would
> > appear flat with an orthogonal tileset.
> >
> >   X X
> > X X X X
> >   X X X X
> >     X X X X
> >       X X X X
> >         X X X X
> >           X X
> >
> 
> If it were just rectangular maps like the one above, I would say that 
> it wouldn't be worth the trouble, but I foresee slightly more complex 
> topologies where wrapping of this type would be needed.

What's wrong with rectangular maps like the one above?  Although less
interesting, it has a lot more practical use than the more complex
topologies you describe.

> If your claim is that more complex topologies will require different 
> guts to normalize_map_pos(), I'm certainly not going to disagree with 
> you (the guts of is_real_tile() would have to change too).

No, my claim is that even a simple topology like this one cannot use
map_adjust_x/map_adjust_y.  Obviously any different topology will
require a change to normalize_map_pos and friends - most likely a switch
statement like you describe.  While some topologies will have very
simple cases, others may be quite complex.

> But under the present---and slightly future---system of flat, single 
> and double wrap rectangular systems, we can normalize x and y independently. 
> Until the underlying topology changes, lets not make things more difficult.

I don't understand the confusion.  The orthogonal rectangle I describe
cannot have X and Y normalized independently.  You agree with me on
this, right?

Are you saying that this rectangle is not a worthwhile topology?  If so,
I disagree with you - it should be more graphically pleasing under the
orthogonal view than the current map setup, and is a good intermediate
step to more complex topologies.

> > Incidentally, I would call this a 12x2 rectangle.
> 
> hmm, what would you call this:
> 
>      X
>    X X X
>  X X X X X
>    X X X X X
>      X X X X X
>        X X X X X
>          X X X
>            X

A good question.  Under the same counting scheme, I guess I would call
it an 11x2.5 rectangle.  I see now, though, that the first rectangle
should be called 12x4 and this one 11x5.  The units are definitely not
the standard one-tile-width ones, though - each unit is 1/sqrt(2) the
length of a standard tile-width (thus the 12x4 rectangle has 24 tiles;
the 11x5 one has 27 or 28 tiles).  Note that this rectangle cannot wrap
in either direction since the grids won't match up - you need an even
value for the dimension for it to be wrappable.

How to represent all of this information as choices to the player
controlling the server I don't know.

jason