Subject: Re: should we replace the shell?
To: None <mycroft@gnu.ai.mit.edu>
From: David S. Miller <davem@nadzieja.rutgers.edu>
List: tech-userlevel
Date: 12/07/1994 17:08:02
   Date: Wed, 7 Dec 1994 16:48:56 -0500
   From: "Charles M. Hannum" <mycroft@gnu.ai.mit.edu>
   Cc: jtc@cygnus.com, tech-userlevel@netbsd.org

   It *has* been, and was quickly discarded.

   1) It's buggy, WAY too large, and WAY too slow.

Yes, it does have problems talking to sockets :-) ugh! This can really
be seen by tape-backup scripts that rsh around different machines and
try to reliably acquire the exit status from the other end :>

   3) It doesn't have arrays.

??? is this is the POSIX spec BTW? I personally don't see this as
system-critical but I can understand that others may.

      It can't be strictly because it is GPL'd
      code because gcc is in netbsd [...]

   Your reasoning is flawed.  GCC is a critical component for which no
   reasonable alternative exists.  ash is a critical component for which
   many alternatives exist, bash being only one of them.

   An old version that *works*, I might add.

GCC works very well on platforms which have people willing to organize
and contribute some time to give the FSF a hand with the large task
that maintaining GCC on so many platforms encompasses. Ala the Linux
GCC team, perhaps it would be a good idea for a netbsd effort of a
similar style to emerge? I would be more than willing to give a hand
in such an endeavor. Please don't take me as a FSF-head, I have an
open mind and respect the work and beliefs of netbsd. It just so
happens that I am very used to having a GNUish environment on the
machines that I do work.

   That is only true if a configuration option was added to remove all
   non-POSIX builtins and shell variables by default.  Last I knew there
   was no such option.

You may be right here, I will have to check the latest snapshot and
find out if such a thing has been added lately. But I do know there is
a "compile me as a stripped down version", however I believe this one
is not completely POSIX.

Later,
David S. Miller
davem@nadzieja.rutgers.edu