Re: [Planner Dev] Windows version



Jani tiainen wrote:

Richard Hult wrote:

On tis, 2004-07-27 at 22:15 +0300, Jani tiainen wrote:

Richard Hult wrote:


On tis, 2004-07-27 at 21:51 +0300, Jani tiainen wrote:


Well, finally I got windows version up and running. Still there is lot of work to do.



Great! How did you solve the type registration stuff?


At some point I changed loading routines a little - that involved setting when doin g_module_open from lazy binding to bind all. Something that creates problems...


Well, that was too early to say.. =)

Still same problem exists. I think that codesharing with gantt-view, and fact that Windows can't have undefined references is the real problem.

I hope that I can overcome that problem soon.


Well, regarding this registration stuff I found following mail in Gnome archives
<http://mail.gnome.org/archives/gconf-list/2003-July/msg00006.html>

    * From: Tim Janik <timj gtk org>
    * To: Michael Meeks <michael ximian com>
* Cc: Botond Botyanszki <botond opt cs kobe-u ac jp>,gconf <gconf-list gnome org>
    * Subject: Re: cannot register existing type `LINCConnection'
    * Date: Tue, 8 Jul 2003 15:00:19 +0200 (CEST)

On 7 Jul 2003, Michael Meeks wrote:

> Hi Botond,
>
> On Thu, 2003-07-03 at 14:37, Botond Botyanszki wrote:
> > I'm working on a gtk2 input module that uses GConf to store its settings.
> > If the input module is loaded twice I get a segfault caused by GConf.
>
> 	Ok - so this is most likely because the static variable used to store
> the GType in the foo_get_type() method is re-initialized to 0 - so we
> try to re-register the type.
>
> 	You walk into a whole can of worms here. I don't recall what the
> official solution is; AFAIR there was some design I saw somewhere once
> to make things like this work - Tim ?

gtk input modules usually:
a) don't conditionalize type registration to a static variable
b) use GTypeModule which takes care of your type registration.
so if you're implementing an input module like the rest
of input modules in gtk, you shouldn't get a warning about
type reregistration.

----end----

I think that what happens is case a) - type is registered to static variable, which seems be something that shouldn't be done...

--

Jani Tiainen




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]