Actually, on "non broken" platforms, Perl's I/O is more efficient than stdio.h I/O because it knows how to peak inside stdio's internal structures to avoid extra copying of data.

Unfortunately, Linux is a "broken" platform in this regard so Perl's native readline (<>) is 4-times slower than it should be.

I suspect benchmarking this code will show that it is slower than plain <IN> when run on many old-style Unix platforms but faster than <IN> on Linux.

I'd like to encourage effort toward "fixing" Perl's I/O optimizations for Linux and other platforms (written in C).

Thanks tye.
I'd been wondering for a while why a read kludge like this would be faster than Perl for IO.
Saves me asking the question in another post. :)
I'll be keeping my eyes open for the fix if/when it appears, as a huge amount of my current work involves Perl and file IO.

Malk

You may want to use sysread() to spped up file I/O. Huameng