From octave-sources-request at bevo dot che dot wisc dot edu  Sat Mar 27 02:19:20 1999
Subject: Re: inpoly.m clean up
From: Ariel Tankus <arielt at math dot tau dot ac dot il>
To: doolin at cs dot utk dot edu (David Doolin)
Cc: octave-sources at bevo dot che dot wisc dot edu
Date: Sat, 27 Mar 1999 10:19:27 +0200 (GMT+0200)

> 
> 
> Code clean, documentation clean, minor 
> corrections.
> 
> 
> ----------------snip snap------------
> 
> % [ inside ] = inpoly(polygon, point)
> %
> % Determines whether a given point is inside a polygon.
> % The polygon is determined by point pairs on its 
> % boundary.  There is no error checking, and to be safe
> % only use convex polygons.
> %
> % input:
> %       polygon: column vector of (x,y) pairs representing the vertex
> %                points of the polygon.
> %       point: (x,y) pair representing point that is either
> %              in or out of the polygon
> %
> % output:
> %       inside: boolean true for inside, false otherwise.
> %
> % todo:  
> %       Implement the ability to handle an array of points
> %       and return an array of boolean values.
> %
> % bugs and limitations:  
> %        
> %         Need to check whether convexity is a necessary 
> %         condition for this algorithm (don't think it is).
> %
> %         Behavior with multiple loops, i.e., if sides cross
> %         over, is unknown.  
> %          
> %         Algorithm implemented in integer, first quadrant
> %         numbers.  Need to check generality of algorithm.
> %
> %         Points on the boundaries of polygons may not be handled
> %         correctly or consistently.
> 
> %
> %         This code was derived from a public domain code
> %         copyright 1995-1996 Galacticomm, Inc., modifications
> %         allowed for any purpose provided redistribution is 
> %         not restricted
> %
> % This script properly belongs in octave/compgeom/
> % (computational geometry).
> %
> % This modified code is copyright David M. Doolin and placed in
> % the public domain, with the exception that redistribution 
> % is not restricted in accordance with the copyright of unmodified
> % (original) C code.
> %
> % $Author: doolin $  doolin at ce dot berkeley dot edu
> % $Date: 1999/03/26 18:42:00 $
> % $Source: /shag/homes/doolin/cvsroot/octave/compgeom/inpoly.m,v $
> % $Revision: 1.2 $
> 

   I suggest some minor but very important changes before the code is
included in the distribution.  The important change is the name
of the function.  It is called: inpolygon in MATLAB, and in my
opinion, we should follow the naming convention.  In addition,
the parameters received by the function do not follow MATLAB's.
MATLAB uses: in = inpolygon(x,y,xv,yv) where x and y are matrices
of coordinates (x(p,q), y(p,q)) and xv and yv are the coordinates
of the polygon.  I understand that (x,y) being matrices is still
not supported, and this indeed can be a future task.
   Though I find *your* input format more convenient, I still
think the input format should match MATLAB's, so the users do not
get used to a different interface, which is difficult to change in the
future.  I do think, however, that leaving your interface as an additional
option is a good idea (switching the position of the polygon and point:
point should be the first argument).


	-- Ariel Tankus
	Tel-Aviv University.