Python's "only one way to do it" philosophy isn't good?

[Steve Howell]

--- Steven D'Aprano
<steve at REMOVE.THIS.cybersource.com.au> wrote:

> On Tue, 12 Jun 2007 15:51:07 -0700, Steve Howell
> wrote:
> 
> > 
> > --- "Anders J. Munch" <2007 at jmunch.dk> wrote:
> >> 
> >> Converting tail-recursion to iteration is
> trivial,
> >> and perfectly reasonable for 
> >> a human to do by hand.  You add an outer "while
> >> True"-loop, the recursive call 
> >> becomes a tuple assignment, and other code paths
> end
> >> with a break out of the 
> >> loop.  Completely mechanical and the resulting
> code
> >> doesn't even look that bad.
> >> 
> > 
> > I have to ask the stupid question.  If a human can
> do
> > this completely mechanically, why can't a machine?
> 
> 
> They can and do -- some compilers optimize
> tail-recursion into iteration.
> 
> Python doesn't, as a deliberate design decision,
> because to do so would
> lose traceback information.
> 

Ok, that didn't occur to me.  What does occur to me,
though, is that tracebacks for recursive algorithms
can get kind of, well, recursive, so I wonder if
there's a tradeoff somewhere.




       
____________________________________________________________________________________
Looking for a deal? Find great prices on flights and hotels with Yahoo! FareChase.
http://farechase.yahoo.com/