'for every' and 'for any'

[Andrew Dalke]

jepler at unpythonic.net:
>this is equivalent to
>    bool([None for x in args if not isinstance(x, str)])

Not quite the same as the original poster's request, which
raises a TypeError, and raises it at the first error location
without continuing to scan the rest of the args.  It's also
harder to understand this as a list comprehension.  Jeff's
style, of

for x in args:
    if not isinstance(x, str):
        raise TypeError, "arguments must be of type str"

is what I consider to tbe the prefered style, except I think the
exception should be raised

        raise TypeError("arguments must be of type str")

(That preferred change occured, I believe, in Python 1.4.)

>> valid = i>0 for every i in vector
>
>This is equivalent to
>    not [None for i in vector if not i > 0]

Jeff's equivalent also stopped at the firt match while yours scans
everything.  That may or may not be part of the consideration of
the original poster.

Oren Tirosh:
>> The words 'any' and 'every' can be non-reserved keywords (like the word
'as'
>> in "import foo as bar").  They are valid only after the keyword 'for'
when
>> it's used in a non-statement context.

Jython supports something like this, to allow "print" as a method name.
This is important because there are a lot of Java classes with that name,
as it isn't a keyword in Java.  As I recall, this came up recently in
python-dev and Guido was against implementing it for CPython, because
he didn't want a word which was a keyword in places and a variable in
others.  This change won't happen.

                    Andrew
                    dalke at dalkescientific.com