Update: A little explanation. A glob (distinct from a typeglob) is a kind of simplified regex for filenames. All unix shells use them, as does DOS in a further simplified form. Perl glob() follows the csh style of filename globbing in its argument, returning a list of all matching file names. The diamond operator applied to a bare string, as in print <*.txt>;, is another way to use them, but glob() is a better choice when interpolation of variables (or "\n") must be done.

After Compline,
Zaxo

Thank you Zaxo. It worked. With the added bonus that the script deleted itself, of course (not the kind of script you want to leave lying around!)

I don't really understand why it worked or why

unlink "*\n"
[download]

or just unlink "*"didn't work. I read the glob section in the Camel book and I'm not much wiser.

This is because unlink is calling the underlying system call unlink which doesn't grok globs. You could do something like this:

opendir( CUR, "whateverdir" ) or die "opendir: $!\n";
do { unlink or warn "unlink " . quotemeta($_) . ": $!\n"} 
  for grep /\n$/, readdir( CUR );
closedir( CUR );
[download]

And just for reference, a non-perl solution would be to use something like this (which uses a find-ism (-print0) which may not work everywhere).

$ find -name "*\n" -print0 | xargs -0 rm
[download]

If you have account access to the machine, you can just use rm with the flags -ri. Namely:

rm -ri <dirname>

IIRC that's how i did it last time i had this problem :)

jynx

rm -r `ls -1 | grep "\\\n"`
[download]