Re: Bug#954075: busybox: provide a low-priority alternative for vi, view, editor
On Tue, 17 Mar 2020 at 09:13:33 +0100, John Paul Adrian Glaubitz wrote:
> I think enabling vi in the busybox configuration is actually the best approach
> to address this problem as this way we continue to ship vi with debian-installer
> and at the same time get rid of the vim dependency which is regularly causing
> headaches when building debian-installer images for Debian Ports.
I think you're mixing up the mostly-udeb-based system that *runs* d-i
with the deb-based system that *is installed by* d-i, and I think you're
mixing up enabling the vi applet with making it available as the 'vi'
command in PATH.
The vi applet is already enabled in the ordinary and static .deb packages:
after "apt install busybox" or "apt install busybox-static", you can run
"busybox vi example.txt" to get a basic vi-compatible editor. It's a less
fully-featured vi than you're probably used to (in particular there is
no undo!) but it works.
However, it isn't automatically set up with symlinks
/usr/bin/vi -> /etc/alternatives/vi -> /bin/busybox etc. if a better vi
isn't installed, which is what I'm requesting in bug #954075.
If you want busybox vi to be available in the default system installed
by d-i or debootstrap, then that's a separate feature request, for which
you would need to ask the ftp team to raise the Priority of busybox from
optional to important.
If you want busybox vi to provide a vi command in the default system
installed by d-i or debootstrap, then you need the alternatives to be set
up (#954075), *and* you need the ftp team to raise the Priority of the
package (and probably lower the Priority of vim-tiny at the same time).
If you want busybox vi to be available in the shell while d-i is running,
then that's a different separate feature request, for enabling vi in
the busybox *udeb* (debian/config/pkg/udeb). At the moment the busybox
udeb doesn't enable CONFIG_VI, and the only visual text editor with a
udeb seems to be nano-udeb. (vim, nvi, elvis don't produce udebs either,
so there is no vi(1) in the d-i environment, and that isn't a regression.)
smcv
Reply to: