[arch-general] New vi/vim/gvim in testing requires intervention

[Jan de Groot]

On Wed, 2009-05-06 at 15:40 -0500, Aaron Griffin wrote:
> On Wed, May 6, 2009 at 3:15 PM, Pierre Chapuis <catwell at archlinux.us> wrote:
> > Le Wed, 6 May 2009 14:53:43 -0500,
> > Aaron Griffin <aaronmgriffin at gmail.com> a écrit :
> >
> >> Might be worth seeing if we can find a patch to fix the crash at least.
> >
> > Patching might be a good idea, but if you just patch the crash I think it will still be even more dangerous to put this version of vi in core because of the behavior described by André Ramaciotti.
> >
> > If any user whose language is not English writes a comment at the top of a config file (say rc.conf or /etc/sudoers if he runs visudo) and saves it, this config file will be emptied and he won't know it. Guess what will happen when he tries to reboot.
> >
> > There is a simple, easy to use and mostly bug-free editor in core, it's nano. Why the need for another one?
> 
> Not everyone likes or prefers nano. More to the point - a working vi
> is required to be POSIX compliant (or maybe it's SUSv3)

As for the working vi, we have two options:

- disable widechar support
- check debian sid for patches, they have some patches that cope with
broken widechar support