Ciaran McCreesh <ciaran.mccre...@googlemail.com> wrote:
> On Mon, 7 Jun 2010 23:05:27 +0400
> dev-ran...@mail.ru wrote:
> > On Mon, Jun 07, 2010 at 07:53:22PM +0100, Ciaran McCreesh wrote:
> > > It's there because if you break your X you probably want a usable
> > > editor to help you fix it.
> > 
> > vim, compiled with "vim-with-x" works correctly when X is broken. It
> > doesn't enable X11-based UI, like flag "X" suggests. It just enables
> > optional connection to x-server to use its clipboard, and vim still
> > works if that connection fails.
> 
> It does not, however, work if your X libraries are broken.

I certainly agree with you that removing the linkage to a handful of X
libraries makes vim more robust if those particular libraries fail.

However even with USE="-vim-with-x", there are a number of other
libraries that, if broken, will still render vim useless, such as
ncurses, perl, python, and ruby.

I suspect if one really wants a fail-proof editor, one would either
be building vim with USE="minimal" which will ignore the 'vim-with-x'
or 'X' USE flag (regardless of what we call it) and also ignore any
perl/python/ruby libraries, or one would want something more
trim, like busybox vi. Or even better yet, busybox vi with USE="static".

Of course changing the USE flag name to 'X' would still let users
decide to *not* link their app-editors/vim against any X libraries via
per-package USE flags.  The main difference in changing the name from
'vim-with-x' to 'X' is that instead of enforcing a default behaviour of
"Vim will not link against X unless explicitly told to do so", we will
be enforcing a policy of "Vim will link against X when USE='X'".

I suppose this is a bit like a transition from an opt-in policy to an
opt-out policy, with the caveat that by enabling USE=X globally, a user
has already declared their intent: opt-in to linking against X
in all packages where there is a choice to do so.

-- 
Jim Ramsay
Gentoo Developer (rox/fluxbox/gkrellm/vim)

Attachment: signature.asc
Description: PGP signature

Reply via email to