increasing expressiveness in array elements test

Question

I like Python for it's expressiveness. I can't express everything as compact as I'd like to, though. For example this one I write quite often:

  def is_everything_okay(some_array):
    for s in some_array:
      if not is_okay(s):
        return False
    return True

But that's more Java then Python. How to improve the expressiveness (and probably execution speed) of that code snippet?

Is any better in the case of an early false result (which doesn't need the checks to continue? How to formulate it correctly? — erikbstack, Feb 07 '12 at 22:35
You have `timeit` to confirm the performance. If that's actually your question. — S.Lott, Feb 07 '12 at 22:39
My question is about if you would like to write up a comprehensive answer to get my upvote and if better then Sven Marnach's also get my `accept`. — erikbstack, Feb 07 '12 at 22:47
"My question is about "?? I still don't understand the question. I can't write up a comprehensive answer. Either it's simply about `any` and `all` or it's about the performance of `any` and `all`. Since `timeit` is available, you could easily post the timing so we can then answer a question. — S.Lott, Feb 07 '12 at 23:18

score 5 · Accepted Answer · answered Feb 07 '12 at 22:22

5

Use the built-in function all():

all(is_okay(s) for s in some_array)

answered Feb 07 '12 at 22:22

Sven Marnach

574,206
118
941
841

looks much more like what I want. is it also faster or does it do the same "under the hood"? – erikbstack Feb 07 '12 at 22:26
2

@erikb: Of course it does the same -- what else should it do? It's slightly faster since it is written in C, but for this example, it still needs to call the Python function `is_okay()` for each `s`. – Sven Marnach Feb 07 '12 at 22:28
Doing the same in well written C is of course something different in terms of speed! Also it can be parallelised for faster results with multicore processors and/or order/index the array before the start if it helps anything. I have no idea! – erikbstack Feb 07 '12 at 22:31
2

It can't re-order the tests: the defined semantics are that it operates on the s's in order. – Ned Batchelder Feb 07 '12 at 22:33
Also does it include the speedup from returning False right when the first False is found? – erikbstack Feb 07 '12 at 22:35
1

Yes, both all() and any() short-circuit. – DSM Feb 07 '12 at 22:35
1

@erikb: Don't expect too much speed-up because the function is written in C. The test the must be executed in every iteration is still Python code. – Sven Marnach Feb 07 '12 at 22:41

score 0 · Answer 2 · edited Feb 07 '12 at 22:43

Just to fill it in Sven's answer*... docs for 2.7 : http://docs.python.org/library/functions.html#all

all(iterable)
Return True if all elements of the iterable are true (or if the iterable is empty).

Equivalent to:
def all(iterable):
    for element in iterable:
        if not element:
            return False
return True

Almost an exact copy of the code you show...

So, using a comprehension of the form (is_okay(s) for s in some_array) creates the iterable which is parsed by all()

Without specifically testing, you won't know which is faster.

and because I'm trying to wrap up a Python class I'm taking and need to answer some questions!

increasing expressiveness in array elements test

2 Answers2