Any of the -X operators involve a call to stat(). You can use the cached result by doing if -f $_ and ! -l _ ie the '_' char accesses the (last) cached stat result. As for readdir vs glob I would expect them to be similar for speed. glob *is* different from readdir in one important way that can bite you if you don't expect it. With glob you will have the full/partial path you fed it in every result element. Sometimes this is very convenient as it saves adding the path, sometimes of course it is not. If it is convenient I use glob, if I just want the file/dir names without any path I use readdir.

As for readdir vs glob I would expect them to be similar for speed.

I got caught by that assumption a while ago...but glob is still nicer.

P:\test>perl -MBenchmark=cmpthese -le" cmpthese( -1, {    
    glob=>q[
        @f=@d=();
        push @{ -d() ? \@d : \@f }, $_ for glob('*');
    ], 
    readdir=>q[ 
        @d=@f=();
        opendir D, '.'; 
        push @{ -d() ? \@d : \@f }, $_ while $_ = readdir D;
    ] 
});"

          Rate    glob readdir
glob    22.9/s      --    -48%
readdir 44.4/s     94%      --
[download]

---

Examine what is said, not who speaks.

"Efficiency is intelligent laziness." -David Dunham
"Think for yourself!" - Abigail
"Memory, processor, disk in that order on the hardware side. Algorithm, algoritm, algorithm on the code side." - tachyon