extended list comprehensions

[Oliver Steele]

fxj at hotmail.com (F. Jamitzky) wrote in message news:<f32195af.0205261316.2482d4ac at posting.google.com>...
> It would be great to have something like a list comprehension for the
> reduce function. It would work in the same way as the comprehension
> for the map function, which is:
> 
> [foo(x) for x in xs]      <-is the same as->       map(foo,xs)
> 
> and maybe it could be written as:
> 
> {y=y+x for x in xs}      <-would be->         reduce(operator.add,xs)
> 
> what do the experts think about that ? 
> 
> ferdinand

What about using ellipsis?  These could do foldl (reduce):
  ...+x for x in xs
  operator.add(..., x) for x in xs

This leaves an obvious extension for foldr (reduce from the right):
swap the identifier and the ellipsis:
  x+... for x in xs
  operator.add(x, ...) for x in xs

The bug in this (putting aside the question of whether it's worth
adding anything to the language) is that it suggests that ... is a way
to write curry (similar to (x+) and (+x) in Haskell), and then:
  [...+x for x in xs]
becomes ambiguous between a singleton list containing a reduction, and
a list of lambdas.  Any way to save this?