Does Python need a '>>>' operator?

[Alex Martelli]

Martin v. Loewis wrote:

> Roy Smith <roy at panix.com> writes:
> 
>> Any operator can get replaced by a function call.  It's just a matter of
>> what's more convenient to write.
> 
> Indeed, and so it is desirable to have infix shorthand notations for
> the frequent operations. However, as the number of infix operators
> grows, the convenience loses: you now have to start remembering line
> noise.

Very well put.  Indeed, given the small amount of bit-twiddling in
typical Python usage, I think & ~ | ^ are already overboard -- they
make a lot of sense in C, a very low-level language that is used a
LOT for bit-twiddling, but, in Python, named built-in functions might
have been more appropriate in hindsight (fewer precedence levels to
remember) -- if not, even, functions in a separate module.

A simple histogram of non-alphameric tokens in Lib/*.py only partially
conforts this thesis:

     0  18495 )
     1  18495 (
     2  16820 .
     3  15467 :
     4  15327 ,
     5  12989 =
     7   4487 ]
     8   4487 [
     9   2426 +
    10   1434 ==
    11   1044 -
    12    699 %
    13    488 `
    14    481 *
    15    342 !=
    17    305 <
    18    280 }
    19    280 {
    20    256 >
    21    150 >=
    22    128 <=
    23    117 &
    24    100 /
    25     80 +=
    26     73 ;
    27     67 |
    28     34 >>
    29     29 **
    30     27 <<
    31     21 //
    35      9 ~
    [ cut here ]

I'll have to get a better / wider / more representative corpus of
Python sources before my dimly-remembered background in computational
linguistics can be used here, I guess -- these numbers are just too
small (and from a probably non-representative sample).  Hmmm, maybe
Parnassus...  It's all academic, of course, anyway.  But even though
we're looking at code so bit-oriented than it bitmasks (&) more than
it divides (or just a bit less, if you count both / and //), bitwise
xor is hardly popular in it -- _that_ much we can already state:-).


Alex