Re: [g-a-devel]Re: [G2R] Re: gail-1.0.pc.in

[Bill Haneman <bill haneman sun com>]

Michael Meeks wrote:
> 
> Hi Bill,
> 
> On Wed, 12 Dec 2001, Bill Haneman wrote:
> > If we go this route we need to be sure that the GTK_MODULES loading
> > mechanism continues to work for GTK+ apps and that it does not clash
> > with gnome_program_foo code.
> 
>         Of course.
> 
> > I am dubious because I think that GTK+ apps (i.e. apps that don't use
> > gnome-program) can use widgets that we need to add support for, so our
> > extension mechanism probably should not be tied to g-p.
> 
>         In which case, either (again) you are implicitely relying on the
> user to type in a different GTK_MODULES line depending on whether he is
> using an app with the canvas or not - or you're pulling in redundant
> libraries ( the canvas, libxml, libxslt, gtkhtml2 etc. ) to a pure Gtk
> app.

I think that gtkhtml2's dependencies on libxml2 may be reason enough to
exclude gtkhtml2 support from libgail and accomplish the accessibility
support via another means.  The only extra dependency I see for
gnome-canvas is libart_lgpl (libxml is only used in the gnome-canvas
glade support).  We can leave it to the release team as to whether a
gnome-canvas dependency in libgail is a problem.

However splitting things out this way will require us to export many of
libgail's headers.

If we go this route (splitting things) then libgnomecanvas and gtkhtml2
will end up with explicit gail dependencies, as is the case for with
their libglade dependencies, so we need to get agreement about that.

If it is likely that "gtk+" apps will use gtkhtml2 or gnome-canvas
without using gnome-program-init then we will need to do something other
than the proposed g-p bootstrap for this support, for instance
overriding gtk_widget_get_accessible () for the widgets in these
libraries to load accessibility support on-demand.  We rejected this
approach (load on demand) for GTK+ because then any apps which
explicitly set accessibility properties will trigger loading of the
accessibility libraries even if there is no listening assistive
technology present.  This last disadvantage could of course be dealt
with via a separate, simple environment variable ENABLE_ACCESSIBILITY=1
or something).

Regards,

-Bill

>         Hmm.
> 
>                 Michael.
> 
> --
>  mmeeks gnu org  <><, Pseudo Engineer, itinerant idiot