seb wrote:
> On Aug 23, 6:18 pm, John Posner <jjpos...@optimum.net> wrote:
[ ... ]
>> How about using a generator expression instead of a list?
>>
>> for i in (x for x in range(10) if x > 5):
>> print i
>>
>> -John
>
> Indeed, but we could have the same syntax than for generators but
> directly in the for statement as in
> for variable in generator if condition:
> body
>
> Is there a special reason for not doing so ? A rejected PEP ?
Well, the Zen of Python does say
There should be one-- and preferably only one --obvious way to do it.
Beyond that, I refer you to Gerald M. Weinberg's _The Psychology of Computer
Programming_, specifically chapters 11 and 12, about Programming Languages,
and their design.
The proposal creates an case where one particular pair of syntactic
constructs can be mooshed together. OK for them, but everything else
becomes an exception; what about
while a==c if b != d:
why not
if b != d while a==c:
or
for a in range(7) if os.name == 'posix':
It winds up burdening the programmers with remembering which constructs are
and which are not mooshable. Weinberg gave an example: FORTRAN had some
stringent rules for what expressions were and were not allowed as array
subscripts. The result was that many programmers couldn't remember all the
rules, and often avoided using legal forms, having forgotten they were
legal.
Maybe the line was already crossed when list comprehensions came into being,
still, the damage is localized in a particular context: building a list. It
isn't out creating wild options in the program control flow at large.
Mel.
--
http://mail.python.org/mailman/listinfo/python-list
