Lazy, infinite recursive sequences in Bash (like in Haskell, if you squint). The result is ugly but interesting.

fibs.hs

fibs = 0 : 1 : zipWith (+) fibs (tail fibs)
-- see http://stackoverflow.com/questions/6273621/understanding-a-recursively-defined-list-fibs-in-terms-of-zipwith

fibs_lazy.sh

init() { mkfifo fib out; echo 0 1 1 >fib; }
gen_fibs () {
  while read i j k; do
    echo $j $k $(($j+$k))
    echo $i > out
  done <fib>fib
}
read_results () { 
  while read res; do 
    echo $res
    sleep .4 # otherwise it goes so fast the integers wrap before you can blink 
  done 
} 
init &
gen_fibs &
read_results < out

fibs_non_lazy.sh

mkfifo fib 
echo 0 1 1 >fib &
while read i j k; do
  echo $i >&2
  echo $j $k $(($j+$k))
  sleep .4
done <fib >fib

Lazy infinite streams in Bash

Lazy, infinite recursive sequences in Bash (like in Haskell, if you squint).

I was inspired by the beautiful Haskell zipWith implementation of the Fibonacci sequence `fibs = 0 : 1 : zipWith (+) fibs (tail fibs)`
to find an equivalent in bash using 'coroutines' and recursive pipes.

My original experiments were https://twitter.com/tavisrudd/status/367164339716751360
"fun w/ recursive pipes: e=echo;mkfifo fib;{ $e 0 1 1 >fib &};{ while read i j k; do $e $i >&2; $e $j $k $(($j+$k));sleep .4; done;}<fib>fib"

and https://twitter.com/tavisrudd/status/367142071489937408
"o=ouro;b=boros;mkfifo $o$b;e=echo; { $e $o > $o$b & }; { while read s;do $e $s>&2;case $s in $o)$e $b;;*)$e $o; esac; done; }<$o$b >$o$b"

:|