>> A consensus for ksh? Among what audience?
> Read the scope thread in 
> http://mail.opensolaris.org/pipermail/indiana-discuss/

The question still begs, among what audience.  The answer is that the
largest potential user-base, currently running linux and freebsd, are not
among those requesting ksh.

> Neither bash nor tcsh are POSIX shells. tcsh is even worse than bash
> as all system scripts will break, bash just breaks 90% of them.

You're confusing /bin/sh, the default script shell, with the
shell specified in /etc/passwd.  Nobody has proposed using tcsh, or any
csh variant, as a scripting shell.  Enough obfuscation already.

> RH, Suse, Ubuntu, Mandrake and Debian ship ksh93 as /usr/bin/ksh by default.

Maybe if you install everything by default, but that's not the standard.
`which ksh` under standard RH (f7) and Debian (4.0) installs returns:

   ksh: Command not found.

>> Reason for this is that the _most_ important feature of a programming
>> shell is compatibility.  People who want advanced features use scripting
>> languages like perl, python, and ruby, not shells.
>
> Ah I think you still live in the 80' since you would've noticed that
> it has floating point math, functions, multidimensional arrays and
> much more. This isn't ksh88 anymore you may remember, we're talking
> about ksh93.

The answer doesn't address the question, evidently to avoid discussion of
the primary theme of the prior post:

   compatibility

99.999% of systems programmers who need multidimensional arrays will use
perl, python, or ruby. Those using ksh will not expect their scripts to
work cross-platform.  It's exactly this mindset that has reduced Solaris'
market share to the single digits.

-- 
Roger Marquis
Roble Systems Consulting
http://www.roble.com/
_______________________________________________
indiana-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/indiana-discuss

Reply via email to