The current implementation of the new_window signal makes writing language bindings for the gtkmozembed widget extremely cumbersome. A better way to handle this signal is to do a "proper" return of the GtkMozEmbed pointer, rather than stuffing it into space saved for it my the function call. The easiest way to do this change (and keep people who already use the current API) is to create another set of signals that instead of returning void, return a GtkMozEmbed widget.
This one doesn't seem to be fixed at all. Chris, we talked about this a while back. Would making a new_window2 signal be the sane thing to do here?
Created attachment 74975 [details] [diff] [review] Patch to add new_window2 signal patch to add new_window2 signal with different semantics. This version is much friendlier to language bindings.
The change doesn't look like it has any obvious problems. I would probably make the new marshal function static though -- no need to clutter the symbol table.
Also sounds important. Confirming.
Is this patch going into CVS anytime soon (propably before the 1.0 final release)?
we're not going to hold 1.0.1 for this.
Created attachment 87515 [details] [diff] [review] added NEW_WINDOW2 to gtkmozembedprivate.h This patch replaces the patch already in here. The given patch didn't compile with mozilla 1.0.0, mine does (at least with the debian source, but AFAIK the debinan patches don't change anything here.). The only difference to Marks patch is that NEW_WINDOW2 is added to the enum of possible signals in gtkmozembedprivate.h. It works fine on my machine, tried using debian pygtk 0.6.9-3 and current cvs pygme (in gnome CVS server).
Created attachment 109151 [details] [diff] [review] new_window2 implementation for gtk1 and gtk2 New new_window2 patch. This has been tested with gtk1 and gtk2. Hoping this can actually get included at some point for people working on gtk language bindings.
The current patch needs the patch from bug 121253 applied. I know that bug 121253 is already closed, but this is to make clear that the patch doesn't compile using mozilla 1.0 or 1.2 based mozillas. I also hope that this patch will be accepted soon, to help all pygme users (e. G. http://pykiosk.sourceforge.net/) because it is annoying to tell your users that not only they need a specific beta from mozilla, but they also have to patch and compile it themselves.
Hello? Anyone alive? ...