On Sat, Mar 15, 2014 at 11:44 AM, Robert Kern <robert.kern@gmail.com> wrote:
I tend to favor tight-right. The general scheme of precedence more or less puts "heavier" operations higher than "lighter" operations (+ < * < **) and @ is "heavier" than * in my mind. I think tight (either -right or -left) has a good correspondence with current dot() expressions, so it will make translation a bit more straightforward, IMO.
s * dot(A, b) == s * A @ b dot(s * A, b) == (s * A) @ b
I'm not sure if this is a convincing argument, but I'll throw it out there: in most of my programming fonts, @ is a bigger, more visually salient character than *, so in any expression that mixes the two, my visual system is going to end up grouping the @-connected terms anyways. -- Robert Kern