On Mon, Mar 17, 2014 at 6:33 PM, Christophe Bal <projetmbc@gmail.com> wrote:

I think that weak-left is a little strange, just think a little of the operators used by mathematicians that always follow a hierarchy.

A parser is mostly done using grammars : see http://docs.python.org/3.1/reference/grammar.html.

Defining *-product to have stronger priority than the @-product, and this last having stronger priority than +, will make the changes in the grammar easier.

I'm now convinced of the usefulness of @ and @@ too but I also think that you must think of other uses than only for numpy. In other words, numpy is a the good argument for this new operators, but this can also open new perspectives for other uses.

My main problem with weak-left (* higher) and tight-left (@ higher) compared to same-left is that I don't see any obvious choice between the weak and tight. I don't think I would have problems with readability. Wikipedia doesn't say anything about precedence of Hadamard versus matrix product. matlab, IDL and Gauss (I checked the manual) all use same-left, as Nathaniel pointed out. For scalar * together with dot product which is more common in formulas, we would just read it sequentially, i.e. same-left. I don't remember when I have seen dot-in-a-circle in a paper, but I don't think there was any precedence either. --- I guess the same applies for other (mis)uses of @ from math import sqrt class MyOp(object): def __init__(self, func): self.func = func def __at__(self, x): return [self.func(xi) for xi in x] myop = MyOp(lambda x: sqrt(x)) print myop.__at__(range(3)) # myop @ range(5) print myop.__at__(range(3) * 2) # myop @ (range(5) * 2) print myop.__at__(range(3)) * 3 # myop @ range(5) * 3 ''' [0.0, 1.0, 1.4142135623730951] [0.0, 1.0, 1.4142135623730951, 0.0, 1.0, 1.4142135623730951] [0.0, 1.0, 1.4142135623730951, 0.0, 1.0, 1.4142135623730951, 0.0, 1.0, 1.4142135623730951] ''' ------------- Josef

_______________________________________________ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion