Sorry for all the misspellings... 2014-03-17 22:32 GMT+01:00 Christophe Bal <projetmbc@gmail.com>:

Hello, and what about something like that ?

a @ b @ c -> (a @ b) @ c a * b @ c -> (a * b) @ c a @ b * c -> a @ (b * c)

Easy to remember. The *-product has priority to @-product, and then we just to @-product from left to right.

An advantage of this is that parsers do job from left to right so I realy think that is a better choice than the weak-right.

Christophe BAL

2014-03-17 21:37 GMT+01:00 Russell E. Owen <rowen@uw.edu>:

In article

<CAPJVwBkLww7-ysZB76LMRZ+mmbyN_5T=ym_VU1pJGakRLBqOkw@mail.gmail.com>, Nathaniel Smith <njs@pobox.com> wrote:

OPTION 1 FOR @: Precedence: same as * Associativity: left My shorthand name for it: "same-left" (yes, very creative)

This means that if you don't use parentheses, you get: a @ b @ c -> (a @ b) @ c a * b @ c -> (a * b) @ c a @ b * c -> (a @ b) * c

OPTION 2 FOR @: Precedence: more-weakly-binding than * Associativity: right My shorthand name for it: "weak-right"

This means that if you don't use parentheses, you get: a @ b @ c -> a @ (b @ c) a * b @ c -> (a * b) @ c a @ b * c -> a @ (b * c)

OPTION 3 FOR @: Precedence: more-tightly-binding than * Associativity: right My shorthand name for it: "tight-right"

This means that if you don't use parentheses, you get: a @ b @ c -> a @ (b @ c) a * b @ c -> a * (b @ c) a @ b * c -> (a @ b) * c

We need to pick which of which options we think is best, based on whatever reasons we can think of, ideally more than "hmm, weak-right gives me warm fuzzy feelings" ;-). (In principle the other 2 possible options are tight-left and weak-left, but there doesn't seem to be any argument in favor of either, so we'll leave them out of the discussion.)

After seeing all the traffic on this thread, I am in favor of "same-left" because it is easiest to remember: - It introduces no new rules. - It is unambiguous. If we pick option 2 or 3 we have no strong reason to favor one over the other, leaving users to guess.

To my mind, being able to easily reason about code you are reading is more important that hoping to increase efficiency for one common case when not using parenthesis.

It also has the advantage that it needs the least justification.

-- Russell

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