>>Yes Gregory:)
>>I generally revert to lowlevel whenever I need similar substr() thing (and to creating an FLL if that doesn't satisfy).
>>Cetin
>
>
>Cetin,
>
>Yes, (suggestion #3 I made this morning)
>
>I found that out when I was writing a function for base64 encoding I posted a while back
>
>suggestion #4 may be faster, but I have no time to try it. Late on my schedule already as it is ;-)

Yeah I'm sure there are faster ways. It sounded to me falling to 1.6 secs for 500K+ file is sufficient for now:) I've created a FLL doing much more complex things (namely creating a sorted dictionary of words in input stream) and its speed is 1Mb+ per second.
Optimization is a fun game:)
Cetin