I processing a very transient directory in an infinite loop. Basically, I read the first 64K of a file to look for specific information. If the information is present, the file gets moved out of the directory. If not, it is eventually moved by another process that I am in a race condition with. In between each cycle, I go to sleep for a period of time as to not chew up too much CPU. About once a minute, I go update my list of criteria for processing as it changes over time. What I would like to do is cache the file names I have already processed, so that I do not process the same file twice. The file names eventually get re-used, but if I invalidate my cache when I update my list, I can be assured that there are no issues. My idea is as follows:

Move on to the next file if it is in my cache

If not, check to see if it meets my criteria

If yes, move it off the directory and do nothing with cache

If no, add the file to my cache and move on to the next file

Once a minute, clear my cache

next if (exists $cache{$_});
or
next if (grep /\b$_\b/ , @cache);

and later on .....

push @cache , $_;
or
$cache{$_}++;
[download]

%cache = ();
or
@cache = ();
[download]