Subject: Re: ksh [was: Re: miniroot for NetBSD/i386?]
To: Hubert Feyrer <feyrer@cs.stevens.edu>
From: Igor Sobrado <igor@string1.ciencias.uniovi.es>
List: port-i386
Date: 12/11/2005 22:41:02
In message <Pine.LNX.4.61.0512112148220.1950@m24s24.vlinux.de>, Hubert Feyrer writes:
> 
> The unfortunate thing is that NetBSD's ksh is pdksh which has a number of 
> features that are not available in other ksh implementations - if it was 
> the same everywhere, I'd immetiately switch to it.

I was not aware of the differences between pdksh and other ksh
implementations.  Indeed, it is much better a simple shell that strictly
follows a standard than a shell that has `features' not available in
other implementations of shells of the same class; in other words,
features *incompatible* with other implementations of shells of the
same class (either C or Bourne) or, as happens with pdksh, other
implementations of the same shell (Korn).

Hopefully all shell scripts I write are either sh(1) or awk(1) ones.
Since NetBSD 2.0 we have, in my humble opinion, the best awk(1)
flavour available, nawk.  Don't know a lot about the Bourne shell
implementation provided in NetBSD but it now supports the "set -o ..."
syntax, and it does not look as _strict_ Bourne shell either.  I do not
know when the "set -o [vi|emacs]" was introduced, but certainly it was
not here two years ago.

I like NetBSD because it is small, it follows standards as strictly
as possible (but certainly has a BSD heritage that should not be
ignored only to be more POSIX/SUS compatible, not all standards can
be implemented without breaking the right behaviour of a BSD system)
and has a nice lack of featurism.

I like calling these `featured' operating systems `electronic
swiss army knifes'.

Certainly, a standard and absolutely compatible ksh will be much
better.  But, sadly, it is not under the control of TNF.  I like
this shell, but certainly better compatibility with other ksh
implementations would improve it.  I fully agree with you.

Best regards,
Igor.