Paul Rudin wrote:
> Steven D'Aprano <[EMAIL PROTECTED]> writes:
>
>> On Sun, 16 Sep 2007 19:25:22 +0100, Paul Rudin wrote:
>>
>>>> The generator expression takes about twice as long to run, and in my
>>>> opinion it is no more readable. So what's the advantage?
>>> If you do it with a decent size list they take more or less the same
>>> time.
>> Did you try it,
>
> Yes.
>
>> or are you guessing?
>
> Well - I guessed first, otherwise it wouldn't have been worth trying :)
>
>> What do you call a decent size?
>
> I used 10,000.
>
>>
>>> You'd presumably expect the generator to use less memory; which
>>> might be an advantage if you have large lists.
>> Unfortunately, Python doesn't make it as easy to measure memory use as it
>> does to time snippets of code, so that's just a hypothetical.
>
> Well - as it turns out the list gets made anyway by the join method,
> so in this case there's probably little difference. However there
> (obviously) are siturations where a generator is going to save you a
> significant memory over the similar list comprehension.
>
>>
>>> Isn't it odd that the generator isn't faster, since the comprehension
>>> presumably builds a list first and then iterates over it, whereas the
>>> generator doesn't need to make a list?
>> Who says it doesn't need to make a list? string.join() needs a sequence.
>>
>
> The generator doesn't make a list. The implementation of str.join does
> apparently needs a sequence and so makes a list from the generator
> passed to it, which is presumably why you get essentially the same
> performance once you factor out setup noise for small lists.
>
> Although it's not clear to me why the join method needs a sequence
> rather than just an iterator.
I suspect it's to do with making memory allocation easier, though I
didn't read the whole gory detail in stringobject.c. Without a
pre-existing set of joinable items you have to build the string up using
some sort of allocate-and-reallocate strategy, whereas if you have the
joinables already in a sequence you can pre-compute how much memory the
result will need.
Ah, right, it's in the comment:
* Do a pre-pass to figure out the total amount of space we'll
* need (sz), see whether any argument is absurd, and defer to
* the Unicode join if appropriate.
regards
Steve
--
Steve Holden +1 571 484 6266 +1 800 494 3119
Holden Web LLC/Ltd http://www.holdenweb.com
Skype: holdenweb http://del.icio.us/steve.holden
Sorry, the dog ate my .sigline
--
http://mail.python.org/mailman/listinfo/python-list
