wscons for FreeBSD?
John Baldwin
jhb at freebsd.org
Fri Oct 28 08:45:39 PDT 2005
On Thursday 27 October 2005 02:47 pm, Peter Wemm wrote:
> On Thursday 27 October 2005 10:07 am, Gordon Bergling wrote:
> > * Thus spake M. Warner Losh (imp at bsdimp.com):
> > > In message: <20051027134919.GA858 at node26.0xfce3.net>
> > >
> > > Gordon Bergling <gbergling at 0xfce3.net> writes:
> > > : * Thus spake Poul-Henning Kamp (phk at phk.freebsd.dk):
> > > : > In message <20051027133336.GA740 at node26.0xfce3.net>, Gordon
>
> Bergling writes:
> > > : > >I am currently working on project, which refactors a lot of
> > > : > > FreeBSDs console code.
> > > : >
> > > : > Sorry for nitpicking, but just to make sure we're clear about
> > > : > what we talk about here:
> > > : >
> > > : > When you say "console", do you mean
> > > : > "vga+keyboard" ?
> > > : > or
> > > : > "printf(9) destination" ?
> > > : > or
> > > : > "/dev/console" ?
> > > : > or all of the above ?
> > > :
> > > : What I mean was 'all of the above'. My projects aims to modernize
> > > : FreeBSDs console in a big picture. That whould range from
> > > : "simple" vga + keyboard to a possible machine indepent
> > > : framebuffer API.
> > >
> > > wscons in NetBSD can be a console (as in /dev/console) destination.
> > > Much like sio(4) can be a console in FreeBSD. However, it is
> > > really a framework for integrating one or more frame buffers (vga),
> > > with zero or more keyboards and mice into a coherent input system.
> > >
> > > It is more of a syscons replacement than a solution to the
> > > 'console' issues that you've been keen on working out.
> >
> > With other words, its something desireable to have at FreeBSD?
>
> A syscons replacement isn't such an enticing deal if it still leaves all
> the 'console' issues. Changing something for the sake of change
> without fixing the problems just causes transition pain for no benefit.
>
> Don't forget, we have to deal with the Xservers as well. Also things
> like libvgl. syscons, with all its warts, runs on i386, amd64, alpha,
> sparc64 and there are patches to make it work on ia64 around somewhere.
> It works with multiple mice and (supposedly) multiple keyboards, but
> that is more a function of external drivers (sysmouse and kbdmux etc).
> I think a wscons port was attempted before, early in the sparc64 days,
> but I don't remember that clearly.
It is _really_ grotty on all but i386 and amd64 though. It would be nice to
have a syscons replacement that truly does work well with framebuffers and
isn't VGA centric. I think phk's console concerns are rather orthogonal to
syscons vs. wscons.
> Things that are personally crucial to me:
> * Speed. (syscons is lightning fast)
> * text mode mouse cursor with cut/paste that Just Works.
> * cursor movement control code compatibility. No less than perfect
> compatibility is good enough! The cons25 termcap entry has to work.
* truly supports framebuffers as well as displays that can render text
directly (i.e. VGA, EGA, etc.)
* allows displays, keyboards, and mice to be tied together in arbitrary
combinations such that each "console" can consist of zero-or-more
displays, keyboards, and mice with well-defined behavior for what
state is shared vs. private when multiple objects of the same class
are connected to a single console.
--
John Baldwin <jhb at FreeBSD.org> <>< http://www.FreeBSD.org/~jhb/
"Power Users Use the Power to Serve" = http://www.FreeBSD.org
More information about the freebsd-arch
mailing list