By the way, I had a teacher (J.P Reveilès) who had invented the theoretically fastest drawline algorithm ever: he simply figure out that, for a line, there is a pixel pattern which repeat itself (unless the variation is a transcendental number http://en.wikipedia.org/wiki/Transcendental_number) so you only have to compute that pattern once and copy/paste it. If I remember well, for example, a line which has a variation of 1/4, has a 4-pixel wide pattern only!
His discrete line equation is relatively beautiful (it is easily extractible from Bresenham's algorithm however): 0 ≤ ax − by < ω
Wouldn't the exception be for lines whose slope is irrational? (i.e. that can't be represented by a ratio of integers), so the repeating pattern is analogous to the repeating pattern in a decimal representation of a rational number.
His discrete line equation is relatively beautiful (it is easily extractible from Bresenham's algorithm however): 0 ≤ ax − by < ω