In fact by doing tests for this explanation I found a short way to use a growing glob that properly handles the first empty entry. Run as perl -ap kleene0.pl <<< "a b" (so add 2 bytes for -ap`)

In fact by doing tests for this explanation I found a short way to use a growing glob that properly handles the first empty entry. Run as perl -ap kleene0.pl <<< "a b" (so add 2 bytes for -ap)

All I still need to do is prime the loop with an empty string. That is done by effectively doing $#a = 0 (, in numeric context becomes 0) which causes the first and only element of @a to become undef which will behave like "" when I use it

All I still need to do is prime the loop with an empty string. That is done using$#a = 0 (, in numeric context becomes 0) which causes the first and only element of @a to become undef which will behave like "" when I use it

All these solutions will keep more and more of the output in memory and that will cause the program to fail after some time. You can also use perl globs for lazy generation by using them in scalar context, but that makes the programs longer....

Perl, 39 37 35 bytes

Explanation:

kleene.pl (this version is 33+2 bytes):

Perl, 39 37 35 bytes

Explanation:

Improvement

kleene0.pl (this version is 33+2 bytes):