As above (in merlyn's code) but with a swap

s/\G(.{1,80})/$1\n/gs;
print;
[download]

Maybe I should benchmark that. This doesn't have the advantage of not affecting long strings and it does put multiple lines into one variable, but that might be OK.

Couldn't help but benchmark this thing...

use Benchmark qw(cmpthese);
open(STDERR,">/dev/null");
cmpthese (10000,
            {
              match => sub { local $_ = "abcde    " x 100;
                             print STDERR "$1\n" while /\G(.{1,80})/gs
+;
                           },
              swap  => sub { local $_ = "abcde    " x 100;
                             s/\G(.{1,80})/$1\n/gs;
                             print STDERR $_;
                           },
            });
[download]

Produces

Benchmark: timing 10000 iterations of match, swap...
     match:  1 wallclock secs ( 1.26 usr +  0.01 sys =  1.27 CPU) @ 78
+74.02/s (n=10000)
      swap:  1 wallclock secs ( 1.01 usr +  0.01 sys =  1.02 CPU) @ 98
+03.92/s (n=10000)
        Rate match  swap
match 7874/s    --  -20%
swap  9804/s   25%    --
[download]

So the swap will save you time (if you are nitpicky about speed).