Re: [moz-bonobo-list] a new mozbonobo plugin

[Jean Bréfort]

Luca Padovani wrote:

> Hi Christian,
>
> On Thu, 2003-07-31 at 23:34, Christian Glodt wrote:
>  
>
> > As I said, it would be worth trying the CVS version. It should at least
> > honor the "type" attribute of the <object> tag.
> >    
>
> I installed the CVS version, and after a few tweakings in my code it
> worked! So we have MathML displayed in mozilla/galeon using the external
> widget instead of the native support, yaeee! The reasons why I'm very
> much interested in this we'll be explained in the component's home page
> as soon as it will have one.
The MathML native support gives very strange result when mozilla is 
compiled with gtk-2 support ;-)

> A few notes:
>
> a) I had to provide an implementation of the PersistStream interface,
> even though PersistFile was supported as well. In fact, my PeristStream
> is broken, but apparently mozilla-bonobo needs at least to be able to
> sense it for working properly
>
> b) now that I have the widget inside the browser, I'd like to have
> interactivity. The question is, is it possible to map browser events
> back into the bonobo component?
They are mapped. I have written a control using mouse events in Gnome 
Chemistry Utils.

> c) My understanding of the differences between Control and Embeddable is
> that the former is usually complete with scrollbars, menus, and so on,
> whereas the latter is a naked view which is more suitable for inline
> embedding. However, if I look at the
>
> /usr/share/idl/bonobo-2.0/Bonobo_Embeddable.idl
>
> interface, I see no methods other than Unimplemented. What methods am I
> supposed to implement?
Bonobo_Embeddable is deprecated in bonobo-2.0. As far as I remember it 
is still possible to use BonoboCanvasComponent, and working samples 
should exist in most recent versions of libbonoboui.

> d) Is it possible to have the source document embedded directly into the
> HTML document, instead of in an external resource?
This is not a problem specific to mozilla-bonobo. It is much more 
general. The problem is that if you have MathML code in the (X)HTML 
document, how can you force Mozilla to use a plugin instead of its 
native support?

Best regards,

Jean