"Brian L. Matthews" wrote: > > That isn't correct behavior for /name/.., because if name is a symlink, > /name/.. may not be /. > Chris Nandor mentioned it too. It really depends on the user's point of view. They may well expect it to be /, and be unhappy when they end up /somewhere/else/entirely. Cancelling /symlink/.. may break when dealing with paths in a physical sense, but it works in the logical sense. The approach I'm taking deals with them in a logical sense. Both are valid and have their uses. There's always File::Spec::cat_dir() for dealing with them in a physical sense. In many cases, symlinks are used to add convenience or hide complexity, and are meant to be 'transparent'. Certainly the average user doesn't consider symlinks when cding around a directory hierarchy. Under Unix some (usually more recent vintage) shells are dual-mode, and default to logical mode, presumably for this very reason. Having worked a lot on one of the most heavy (ab)users of symbolic links in the known universe, SCO OpenServer, I can attest to the fact that working with systems where symbolic links / aliases whisk you away to parts unknown unexpectedly is extremely annoying. ***** Want to unsubscribe from this list? ***** Send mail with body "unsubscribe" to mac-perl-request@iis.ee.ethz.ch