Re: [Chicken-users] Problems with srfi-4-comprehensions

[Alex Shinn]

On 3/16/07, Alex Shinn <address@hidden> wrote:
> On 3/16/07, Ivan Raikov <address@hidden> wrote:
> >
> > Could someone who actually understands Scheme macros and
> > comprehensions take a look at this? Thanks a bunch,
>
> I won't touch SRFI-42, but if you'd use it I'll provide a SRFI-4
> extension to loopy-loop :)

I almost forgot about this.  I just updated loopy-loop with support for
SRFI-4 by default (it's only one extra line of compile-time code per
vector type, not really bloat).

#;1> (use syntax-case loopy-loop srfi-4)
#;2> (loop ((x <- in-u8vector '#u8(1 2 3))) (print x))
1
2
3

I also added the in-cartesian-product iterator which takes a list of
lists and iterates over the full cartesian product (right joins, may add
left joins later) of those lists, with a fast algorithm using only O(n)
space where n is the number of elements of the initial lists.

#;3> (loop ((x <- in-cartesian-product '((a b) (c d e)))) (print x))
(a c)
(a d)
(a e)
(b c)
(b d)
(b e)
#;4>

For those of you currently using srfi-42 or CL's loop or otherwise
hesistant to try loopy-loop, let me point out some advantages of
loopy-loop:

1) It's simpler and easier to use.  It really is.  If the documentation
  doesn't reflect this well, tell me so I can improve it :)
2) It's more general.  It gracefully handles changes as your loop grows
  in complexity and accumulates ad-hoc special cases.
3) It's fast.  It produces very simple, tight code with no mutations.
  All the alternatives use mutation.
4) It's compile-time only.  There are no run-time dependencies, making
  it suitable even if you don't want to use any module system.
5) It's actively developed and tested to work with either syntax-case or
  syntactic-closures, whichever you prefer.  The latter is smaller and
  faster.
6) It preserves line number debugging information when used with
  syntactic-closures.
7) It allows implicit destructuring-bind on all loop variables.
8) It has lots of nifty iterators (e.g. hash-tables, permutations and
  combinations) not found by default in the alternatives.

--
Alex