Re: Mouse a11y tab

[Jens Granseuer <jensgr gmx net>]

On 14.11.2007 16:21, Gerd Kohlberger wrote:
> Jens Granseuer wrote:
> > On 12.11.2007 17:17, Gerd Kohlberger wrote:
> >> Transforming the mt daemon in a settings module isn't going to work,
> >> because
> >> they are all compiled into one binary. This, I guess, makes it
> >> impossible to use
> >> mousetweaks in GDM.  If the daemon stays in its own binary then there
> >> is no
> >> reason to move it inside the gnome-cc tree.
> >
> > Well, yes and no. I agree that the GDM case may well be a problem.
> > I don't know the a11y stuff very well, though. The a11y daemon is
> > also started only after logging in, isn't it? How do they solve the
> > GDM issue? (Do they just ignore it? If so, should we, too?)
>
> Not sure if I understand the question, but the at-spi registry daemon is
> available in gdm. Most if not all gnome a11y apps use it. A description
> of how the mt is currently used with gdm can be found here [1].

My thinking was: the at-spi-daemon is usually started by gnome-session,
just like g-s-d. However, it is apparently possible to have those a11y
features turned on in gdm regardless, so maybve we could use the same
mechanism for the mt stuff. Then again, this would probvably mean we'd
havbe to start g-s-d in gdm already, and I'm not sure whether that
would cause problems.

> > * MT daemon as a g-s-d module. If (!) GDM turns out to be a problem
> >   serious enough to keep a separate daemon, you're probably going to
> >   need some g-s-d integration anyway (starting/stopping the daemon
> >   when the options are changed). But yes, in that case the daemon
> >   (and only the daemon) might make sense as a separate package, too.
> >   I must admit I'm not too fond of the prospect of yet another daemon,
> >   though.
>
> Assuming g-s-d is not available in gdm, it is a serious problem.

Another idea I'd like to throw into the ring is a dual approach: mt
is in g-s-d bvy default, but we either make it possible to compile a
stand-alone binary of the mt code that could be used in gdm, or a
stand-alone mt-daemon is simply provided as a separate package for
those who need it. Of course, that would be an additional
maintenance burden, but maybe it's not that bad? Just a thought.

> I've
> however created a first patch for the mouse module [2] that checks for
> the mt features and toggles the daemon.

Please open a bug in bugzilla regarding mt integration and attach
any patches there. As long as you favour the separate module
approach we'll probably want to hold off applying the g-s-d parts
until we know what the final decision on the module proposal is,
but we could start with the mouse capplet parts right away.

Jens