...and of course also seahorse/seahorse-plugins/configure.in

Has something changed in libpanel-applet that it no longer requires Bonobo?

No - so we depend on gnome-panel here. Good to know.

gnome-panel has a bonobo-free branch now and will switch for 2.31.x.
See dependency bug.

http://live.gnome.org/GnomeGoals/AppletsDbusMigration if that helps anybody.

JFTR: Carlos Garcia did a good work with gnome-panel in bug #572131. It's very useful as it's doing the same.

Created attachment 163349 [details] [review]
Get rid of Bonobo

I have made a patch. It compiles well but I haven't tested it yet.
Comments are welcome.

Hi Pablo, thanks for trying to port the applet, some comments:

 - Menu xml file, service file and panel-applet file are missing in the patch, you probnably forgot to run git add

 - In order to hide the properties menu when the applet is locked down you can use something like:

if (panel_applet_get_locked_down (applet)) {
        GtkAction *action;

        action = gtk_action_group_get_action (action_group, "Props");
        gtk_action_set_visible (action, FALSE);
}

 - I wonder why there's a custom main instead of using the factory macro. I think we could use PANEL_APPLET_OUT_PROCESS_FACTORY macro and move seahorse_secure_memory_init() to seahorse_applet_factory(), but I'm not sure. Another solution might be adding a new macro PANEL_APPLET_OUT_PROCESS_FACTORY_WITH_CODE or something like that

Created attachment 163379 [details] [review]
Get rid of Bonobo (v2)

Thanks for your comments Carlos.
Changes:
* added missing files to the patch
* panel_applet_get_locked_down()

About the last change, I'm not sure if it would work. Quoted from the documentation: "seahorse_secure_memory_init (). Configures non-pageable (secure) memory. Must be called before the first memory allocation."
Maybe gtk_init() does some memory allocations, :?

Stef can you give us a clue here?

The way seahorse_secure_memory_init() is written, it needs to be called before any glib or gtk+ code, because it replaces the glib memory allocator.

The alternative is to rewrite all the secure memory handling stuff in seahorse so we explicitly, and stop using WITH_SECURE_MEM() and related macros. This would remove the need for seahorse_secure_memory_init().

Well, not using factory macro is not a problem anyway.

So, has anyone tested my patch?

*ping* Carlos

Created attachment 167091 [details] [review]
Fix minor issues of previous patch

Right, I knew I was going to forget it. During guadec Pablo and I tested his patch and we found some minor issues. This patch that goes on top of pablo's, fixes the problems.

Created attachment 170015 [details] [review]
Get rid of Bonobo (v3)

Please someone test this patch. I think it's done.

Review of attachment 170015 [details] [review]:

It works for me, there's just a minor issue.

::: plugins/applet/seahorse-applet.c
@@ +1029,3 @@
+
+    if (panel_applet_get_locked_down (applet)) {
+        action = gtk_action_group_get_action (action_group, "Props");

You are using action_group here that has been already unrefed, just move g_object_unref (action_group); after this block

Created attachment 170140 [details] [review]
Get rid of Bonobo (FINAL)

This is my proposed final patch. It would be good to commit it today, since tomorrow begins Hard Code Freeze.

* Fixed g_object_unref
* Fixed POTFILES.in

(In reply to comment #17)
> It would be good to commit it today, since tomorrow begins Hard Code Freeze.

It can be committed to master at any time after branching for gnome-2-32.

Pablo, unless Adam speaks up, I'm okay with you committing this before the 2.31.92 release. Especially if you work on releasing 2.31.92.

Pablo, Stef; could this be commited now?

This was applied, so I'm closing the bug.