On Thu, 20 Jul 2006 17:09:19 +0200 Martin Schaffstall wrote:
> On 7/19/06, Glenn Fowler <gsf at research.att.com> wrote:
> > isn't that "usefull to have ONE configuration file for these shells" 
> > /etc/profile?

> /etc/profile and /etc/ksh.kshrc are different things. ...

ok

as it stands now /etc/ksh.kshrc is enabled if either it or /etc/bash.bashrc
exist on the build system

> > also, the patch to force /etc/ksh.kshrc checks for a vendor name rather than
> > a feature -- we prefer to do that only as a last resort
> > a more appropriate place for vendor specific settings would be in the vendor
> > specific makefile(s) that set up compile time options
> > in this case -DSHOPT_SYSRC should do what you want
> > if it doesn't then let me know so I can make it work

> I filed he patch because there is no way for a normal, unprivileged
> user to enable the feature without hacking the code. Both
> /etc/bash.bashrc and /etc/ksh.kshrc can only be created by root and no
> one else.

my comment was about the patch placement, not the patch itself
arrange for -DSHOPT_SYSRC in the solaris makefiles
(some other -D... are already thrown in the proposed solaris ksh93 makefiles)
this separates vendor specific code from feature specific code

the reason for ast avoidance of tests for vendor specific strings is
suppose one day solaris decides to change its name to sunos
or what if boonix and farnix also want this solaris feature
it could take full time position to keep track of vendor names
spit out by uname(1) etc. or what feature menu defaults correspond
to each new vendor release

-- Glenn Fowler -- AT&T Research, Florham Park NJ --


Reply via email to