On 2007-08-21 13:23:53 -0400, Michael Stone wrote: > On Tue, Aug 21, 2007 at 06:32:53PM +0200, Vincent Lefevre wrote: >> On 2007-08-21 11:12:11 -0400, Michael Stone wrote: >>> You're arguing for enforcing a consistency that doesn't exist. No >>> matter how strongly you argue, you're not going to change that fact. >> >> What fact? > > That your starting premise is incorrect. (As described in the previous > email.)
Well, this is a bug in zsh. In any case, this makes things non obvious for the user. >>> that the behavior of touch is standardized by posix, and that the >>> handling of symlinks is *also* standardized by posix. >> >> Even if this is standardized by POSIX, the behavior should be >> documented. > > No. It's fundamental to how unix systems behave, and is not useful to > document on every man page. If it's so fundamental, why don't the ls and stat command dereference symlinks by default? Why are there bugs in various commands (the chmod command from the coreutils, chgrp from zsh)? >>> (Unless otherwise specified, the target of the symlink is operated >>> upon, rather than the symlink itself.) >> >> Where is this documented? > > http://www.opengroup.org/onlinepubs/009695399/xrat/xbd_chap03.html#tag_01_03_00_57 The POSIX standard is not the man pages. The man pages should be self-contained. Or for the same reason, you could remove all what is defined by POSIX (such as the description of the -a option for 'touch'). -- Vincent Lefèvre <[EMAIL PROTECTED]> - Web: <http://www.vinc17.org/> 100% accessible validated (X)HTML - Blog: <http://www.vinc17.org/blog/> Work: CR INRIA - computer arithmetic / Arenaire project (LIP, ENS-Lyon)
