Re: mksh

[Adam Hoka <adam.hoka%gmail.com@localhost>]

On Thu, 13 May 2010 19:21:43 +0000 (UTC)
Thorsten Glaser <tg%mirbsd.org@localhost> wrote:

> Tobias Nygren <tnn <at> NetBSD.org> writes:
> 
> > I have no particular opinion except I'm aware that some of the addon
> > functionality such as the support for printf(1) style hexadecimal escape
> > sequences in the echo(1) builtin have the potential to break existing
> > scripts.
> 
> Not if the shell is built with -DMKSH_BINSHREDUCED when it is called
> as /bin/sh, -sh, etc. (for example via crunchgen).
> 
> Of course, while Iâd be delighted for both, you should probably distinguish
> between replacing /bin/ksh with mksh (with a hardlink from ksh to mksh) and
> replacing /bin/sh with mksh.
> 
> For what itâs worth, FreeWRT has done both, MidnightBSD too, Android-x86
> didnât even have a ksh and Android proper are currently considering it.
> Red Hat Enterprise Linux 6 ships with mksh as both ksh88 and pdksh repla-
> cement (and ksh93 for ksh93 scripts), and from what I gathered, their
> customers have tested their scripts with mksh and Iâm not aware of any
> bugs foundâ
> 

mksh would give us fixes for many pdksh bugs, {foo,bar} expansion,
UTF-8 fixes, and many more with very little cost.

Upstream is also very responsible.

-- 
NetBSD - Simplicity is prerequisite for reliability