home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!pipex!warwick!uknet!glasgow!sinclair
- From: sinclair@dcs.glasgow.ac.uk (Duncan Sinclair)
- Newsgroups: comp.sys.sun.admin
- Subject: Re: The case of the symbolic link traversal in 4.1.3
- Message-ID: <Bx8zGo.GG0@dcs.glasgow.ac.uk>
- Date: 5 Nov 92 14:41:11 GMT
- References: <398@ftms.UUCP> <Bx7G25.4C0@dcs.glasgow.ac.uk> <1992Nov5.000654.10013@cirrus.com>
- Organization: Glasgow University Computing Science Dept.
- Lines: 34
-
- dhesi@cirrus.com (Rahul Dhesi) writes:
-
- >In <Bx7G25.4C0@dcs.glasgow.ac.uk> sinclair@dcs.glasgow.ac.uk (Duncan
- >Sinclair) writes:
-
- >>Then when I saw this I ran a few tests, seems that it's not ls's fault,
- >>but the 4.1.3 kernel that does this. It has changed the path semantics,
- >>so that "foo/" == "foo", rather than "foo/" == "foo/."
-
- >But I don't see how "x/" can ever correctly refer to anything except a
- >directory. A kernel that simply ignores the trailing slash is broken,
- >in my opinion. At the very least, there should be a check to make sure
- >that x/ is a directory.
-
- Agreed, this is completely bogus. I hadn't even thought of this
- as a possibility before.
-
- I hope no shell will pattern match "*/" and include files as
- well as directories, I know zsh does it right. What is needed
- is for the shell to do the check that the kernel doesn't.
-
- >System V has been thus broken for some years. It has lately become
- >fashionable to incorporate System V's bugs into other OSs. POSIX
- >mandates this behavior too.
-
- This stinks.
-
-
- -- D*S
-
-
- --
- Duncan Sinclair | sinclair@dcs.gla.ac.uk | sinclair@uk.ac.gla.dcs
- --- Rave Culture: Don't techno for an answer. ---
-