Re: GTK+ Modules and GDK
- From: Owen Taylor <otaylor redhat com>
- To: Joel Becker <jlbec evilplan org>
- Cc: gtk-devel-list gnome org, timj gtk org
- Subject: Re: GTK+ Modules and GDK
- Date: Wed, 19 Jun 2002 17:03:05 -0400 (EDT)
Joel Becker <jlbec evilplan org> writes:
> On Tue, Jun 18, 2002 at 02:12:06PM -0400, Owen Taylor wrote:
> > E) Make modules export a special entry point indicating
> > that they are "multihead aware". If they don't export
> > it, then don't initializate them until a default
> > display has been set up.
> >
> > (This entry point could be 'gtk_module_init2()', a cleaner
> > looking way to do it would be to say that if a module exports
> > gtk_module_init_display(), then the gtk_module_init()
> > will be called initially, and gtk_module_init_display()
> > for each display.)
>
> I must admit I can't quickly come up with a situation where I'd
> want to start a GUI toolkit without a GUI, but the ability for GTK+ and
> modules to do per-display work is obviously needed. I like your E)
> solution with the init()/init_display() calls. It covers all the bases
> and cleanly maps to the problem space, whereas some of the other
> solutions are roundabout or unintuitive to the module programmer.
Just to kill speculation :-), the reasons for initialization
without opening a display are:
1) Programs that are structured as:
a) Initialize GTK+
b) Take requests for opening a connection to display X
2) Introspection by documentation or language-binding generation
tools. (gtk-doc doesn't need GTK+ to run any more, but running
gtk-doc for GTK+ does require GTK+.)
Mostly 1), since 2) basically works now ... none of our class_init()
functions require gtk_init() to have been called at the moment.
Regards,
Owen
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]