generator expressions: performance anomaly?
steve at holdenweb.com
Tue Jan 18 09:29:15 EST 2005
Antoon Pardon wrote:
> Op 2005-01-18, Diez B. Roggisch schreef <deetsNOSPAM at web.de>:
>>>Something else I was thinking about. I think it would be nice if the
>>>python compilor could figure out whether a genexp in a list or tuple
>>>expression always generates the same list or tuple and then instead
>>>of generating code would generate the list or tuple in place.
>>This won't ever happen in python - at least not in python otherwise similar
>>to the one we know...
>>The thing you're after is known as "common subexpression elemination" and
>>can only be done in purely functional languages. While that certainly is an
>>interesting property of a language, it e.g. forbids functions like
>>time.time() - a too huge paradigm shift for python.
> I don't see how generating byte code for a = 9; when seeing the
> expression a = 3 + 6, would be a problem for non-functional
> I agree that things like [time.time() for i in xrange(10)] shouldn't
> be pregenerated and that the problem is more complicated as I thought.
> But during compilation the compilor could do an anlysis of the code
> do determine whether there are side effects or not. If the compilor
> then would store a code in the byte code for functions that are
> guaranteed side-effect free and only pregenerated objects generated
> by expressions with no side-effect, some common subexpression
> elimination could be done even in a non-functional language.
Indeed, and it has been, which is a matter of well-recorded history.
Constant folding has been used since the early Fortran compilers.
But you've already stated that the problem is more complicated than you
thought. I presume you are already prepared to bail on any function that
calls into external modules or extensions, since clearly nothing can be
known about their behaviors vis a vis side effects.
Python is *designed* as a dynamic language. I wish you would embrace
this aspect rather than continually trying to shoehorn it into a static
straitjacket. Efficiency is good. Flexibility is better.
Steve Holden http://www.holdenweb.com/
Python Web Programming http://pydish.holdenweb.com/
Holden Web LLC +1 703 861 4237 +1 800 494 3119
More information about the Python-list