home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.unix.shell
- Path: sparky!uunet!mcsun!sunic!ericom!erix.ericsson.se!per
- From: per@erix.ericsson.se (Per Hedeland)
- Subject: Re: Action of `echo */'
- Message-ID: <1992Nov21.110404.4245@ericsson.se>
- Sender: news@ericsson.se
- Nntp-Posting-Host: super.eua.ericsson.se
- Organization: Ellemtel Telecom Systems Labs, Stockholm, Sweden
- References: <1eg04jINN9go@sprocket.csv.warwick.ac.uk> <1992Nov19.163705.15795@eua.ericsson.se> <By110F.1s8@root.co.uk>
- Date: Sat, 21 Nov 1992 11:04:04 GMT
- Lines: 54
-
- In article <By110F.1s8@root.co.uk> gwc@root.co.uk (Geoff Clare) writes:
- >per@erix.ericsson.se (Per Hedeland) writes:
- >>It's a bug, er... fix introduced in SunOS 4.1.3. It's in the release
- >>manual, section 6.6 "POSIX Bugs Fixed", "Kernel":
- >
- >>1039505 Trailing / in pathname should be ignored.
- >
- >This is a kernel bug fix, presumably relating to treatment of pathnames
- >passed as arguments to system calls.
-
- Yes of course, that's what '"Kernel"' implies...
-
- > It has nothing to do with the shell.
-
- ...but I find this assertion more than a little strange regarding
- something that demonstrably changes the behaviour of three different
- shells, including the primary interactive one shipped by the vendor...
-
- If you meant to say "It need not affect the shell", that is of course
- true, provided that the shell is aware that the kernel lies
- behaves this way - as tcsh is, when NAMEI_BUG is defined.
-
- But of course this isn't just about the shell (of which there are many)
- or globbing (which is sometimes done by non-shell programs), changes in
- the pathname lookup semantics tend to affect quite a few programs.
- E.g., does anyone think that
-
- % ls -l .login/
- -rw-r--r-- 1 per 878 Oct 16 1991 .login/
-
- is correct behaviour? Should all programs be changed to work around
- these semantics? Will they ever be? And if so, doesn't that tend to
- indicate that the semantics are wrong?
-
- [ Comments about the opinion of POSIX.2 on '*/' deleted ]
-
- >It might be interesting to submit this issue to the IEEE for an official
- >POSIX.2 interpretation. Any volunteers?
-
- I'd much rather see the issue (presumably in POSIX.1) about ignoring
- trailing slashes on pathnames that don't refer to directories
- "interpreted" - I haven't seen the text, but am told that it is
- similar to the one-liner above, i.e. there's no mention of directories
- or any other types of files - and preferrably amended to say that this
- should only be done for directories (dream on...).
-
- I can always choose my personal shell, and it will not match files with
- '*/' regardless of what POSIX.2 says - chosing my personal kernel is not
- as easy...
-
- --Per Hedeland
- per@erix.ericsson.se or
- per%erix.ericsson.se@sunic.sunet.se or
- ...uunet!erix.ericsson.se!per
-