1. damn, you start thinking 'wow a new kewl feature that does not break anything' and suddenly stuff is broken. Life just sux
     posted by verbat at 06:43:52 AM on September 04, 2004  
2. [not tested]
   
   def sieve2(stream):
   newstream = lambda p,stream: (n for n in stream if p(n))
   while 1:
   head = stream.next()
   yield head
   p = lambda a, b=head: not divisible(a, b)
   stream = newstream(p,stream)
   
   Genexp are meant to be passed to a function for
   immediate consumption, at least for Guido.
   So this is beyond that scope.
   
   That this problem would show up in this kind of usage was known. Lambda itself has a similar problem (workaround b=head).
   
   Other capture rules were also suggested.
   
   The PEP tracks some of this.
     posted by an anonymous coward at 10:31:00 AM on September 04, 2004  
3. """ damn, you start thinking 'wow a new kewl feature that does not break anything' and suddenly stuff is broken."""
   
   It doesn't break existing code, AFAIK...
   
   And yes, this was a known problem, but nevertheless it's a bit disappointing, that genexps cannot be used for this purpose. Not that it's really important, but this lazy name lookup rule makes it difficult to do interesting stuff with genexps.
     posted by Hans Nowak at 11:50:17 AM on September 04, 2004  
4. Just looking at the first example, in python2.3 you can already get essentially the same functionality:
   
   import itertools
   odd_numbers = itertools.ifilter(lambda x: x%2, numbers())
   
   Consequently, I don't really see the need for generator expressions...
     posted by Anonymous at 11:40:35 PM on September 04, 2004  
5. All in all the following version is not bad, I think:
   
   def sieve3(stream):
   def newstream( head, stream):
   return (n for n in stream if not divisible(n, head))
   
   while 1:
   head = stream.next()
   yield head
   stream = newstream(head, stream)
   
   Groetjes!
     posted by Wim Stolker at 08:25:17 AM on September 10, 2004  
6. Oops,
   
   sieve 2 ( and 3 ) non recursieve ???
   Try calculating the first 1000 primes!
   
   RuntimeError: maximum recursion depth exceeded
   
   Groetjes!
   
   
     posted by Wim Stolker at 08:46:34 AM on September 10, 2004  
7. Hm, yes, they are just as recursive... except in Scheme-land (where the original examples came from) this is considered "iterative". I guess it doesn't make much difference for Python.
     posted by Hans Nowak at 11:43:27 AM on September 10, 2004  
8. This blog entry might be useful: http://goldenspud.com/webrog/archives/2004/09/12/ah-so-thats-how-they-work/
   
   Maybe you could adapt it to use that (admittedly hackishly ugly) trick?
     posted by Moof at 05:42:19 AM on September 14, 2004  
9. i\'m liked :)
     posted by Joe at 11:06:00 PM on December 27, 2004  
10. hi friend, it seems that i am in different site,
    oh no its the same site where I wanna go but no comments
    are appearing here related to the topic. May be I am wrong
    but friend got your site while searching blog in search
    engine and it is nice to say that i visited ur site
    in my friend's house who is layout controller
    of website's. He told me that this particular
    site contain reasnable layouts and good contents
    which is relevant to this site. Rarely we get this
    type of site. Really appreciates your blog page. Thnks.
      posted by dafer at 04:14:22 AM on January 06, 2005