From: Yohanes Santoso <ysantoso@...>
Date: 2002-04-03T06:34:04+09:00
Subject: Re: a note on the . and .. discussion

patrick-may@monmouth.com (Patrick May) writes:

> I just realized that unix commands do not recognize the "." and ".."

You're mistaking shell with unix command. The character * is expanded
by shell. The invoked unix command does not even know that the user
type the character *.

The shell does not include ., .., and all files that begins with a
.. in expansion of * because by policy, all files that starts with
.. is considered hidden. There is no preventing the user to make the
shell not to ignore these hidden files: 'grep something .*'

Since it is only a policy, there is no guarantee that my favourite
shell does not include . and .. in the * expansion. Luckily, no shell
has this behaviour.

> Adding to the apparent inconsistency -- even the root dir has "." and
> ".." entries.  I would think that if "." and ".." were to provide

.. of / is consistent. It refers to itself.

... of / is to make things convenient. Otherwise, you'd have to
put condition everytime you want to use ..: you'd have to check
whether the current dir is the / dir. This is inconvenient.

> - IMHO, I think that aDir.parent is cleaner than aDir[".."].  I also
> think that "." and ".." should continue to be valid "glob"

Exactly the argument why Ruby has $* when it already has ARGV. . and
... is such common thing now that it is adopted by most other
filesystems in various OSes. So, it's no longer a UNIX only convention.

YS.