Path: utzoo!attcan!uunet!husc6!bloom-beacon!EXPO.LCS.MIT.EDU!keith
From: keith@EXPO.LCS.MIT.EDU (Keith Packard)
Newsgroups: comp.windows.x
Subject: Re: X11 fenceposts
Message-ID: <8811162307.AA19309@EXPO.LCS.MIT.EDU>
Date: 16 Nov 88 23:07:43 GMT
References: <3595@pt.cs.cmu.edu>
Sender: daemon@bloom-beacon.MIT.EDU
Organization: The Internet
Lines: 59


>XDrawRectangle(display, drawable, gc, x, y, w, h) draws a rectangle defined
>by the coordinates (x,y)  (x+w,y)  (x+w,y+h)  (x,y+h)

>while

>XFillRectangle(display, drawable, gc, x, y, w, h) fills a rectangle defined
>by the coordinates (x,y)  (x+w-1,y)  (x+w-1,y+h-1)  (x,y+h-1)

>the manual (Xlib - C Library, pp 83 and 86) say that both functions
>should be behaving in the first manner.

The Protocol document is quite clear on this point, the two
requests will always draw different bits.  The R3 version of the
Xlib manual quotes from the protocol document, having no copy of
the R2 manual, I don't know what it says.

The trouble is that the PolyRectangle request draws a 5 point PolyLine.
A wide-line is defined by the rectangle centered along the specified
coordinates which is as wide as the line-width.  So, the verticle
portions of the PolyRectangle will extend 1/2 pixel to the left and
right of the given coordinates.  

Read the protocol document where it talks about filling polygons; as
wide-lines are defined in terms of polygons, the discussion applies
to them as well.

(for pixels on the boundary of the polygon) "the pixel is inside if and
only if the polygon interior is immediatly to its right (x increasing
direction)".  It goes on to clarify the case on the bottom of the polygon.

The method I used in R3 to compute bounds for span generation is:

	int_left_x = (int) ceil (double_left_x);
	int_right_x = (int) ceil (double_right_x - 1);

Applying these equations to the coordinates generated for
the filled rectangle makes the right edge move left one pixel:

ceil (x + width - 1) = x + width - 1

The outlined rectangle ends up at the same coordinates it started at:

ceil (x + width + 1/2 - 1) = x + width
                   ^
		   |
                   +-- line width / 2

This discussion also applies to the horizontal portions of the
rectangles, with similar results.

Wide lines in the R2 server used a different (and incorrect) algorithm
for calculating the span boundaries, however in this case, the
results would be the same.

						Keith Packard
						MIT X Consortium
						(617) 253-1428
						keith@EXPO.LCS.MIT.EDU