tv02 at c3sl.ufpr.br
Tue Jul 11 08:38:50 PDT 2006
Zephaniah E. Hull escreveu:
> Basicly, from the patches I've seen prior to now, I don't especially
> like the approach.
> More to the point though, there is some change in just what the DDX does
> and does not need to provide coming down the pipe which should, if done
> right, may make it fairly easy to move a good bit of the xfree86 input
> DDX that's not already part of mi into something that could be fairly
> easily reused by Xgl and the Kdrive servers without a ton more effort.
> That would actually allow the full xf86-input-evdev to be used in all
> cases, instead of a stripped down version that behaves very differently,
> and at least IMHO, that sounds like a big enough win to put the effort
> into once the other changes get into the git tree.
> But that's still going to be a bit in the future, as will some noticeable
> design changes to accommodate the server itself handling hotplug, so the
> current x-input-evdev is probably at least a good interim solution, but
> I am honestly not sure about it's place in the longer term.
You're totally right about the x-input-evdev code. Its ugly, however
functional. I wrote it quickly as a proof of concept that evdev code
inside X should be in a layer next to the kernel, hence others DDXs can
Firstly, what we gonna do is split the xf86-input-evdev in two:
x-input-evdev (mi and syscall functions) and xf86-input-evdev (DDX
only). You're even right right on the place where it all will go. IMHO
x-input-evdev will be a library instead a driver that xf86-input-evdev
(even some xgl-input-evdev, etc) will communicate to provide the X events.
Hotplug is cool and we must do it. Do you use IRC?
thanks in advance.
> Zephaniah E. Hull.
C3SL - Centro de Computação Científica e Software Livre
More information about the xorg