Done, simply assuming that you are right and that it is unnecessary.

Thanks!

The [CCode (instance_pos = -1)] is actually needed when using Gtk.Builder.connect_signals. Even if it won't segfault in this example, trying to access either 'window' or 'this' will probably do. So it needs to be there when using GtkBuilder.

Ah, I see, that was the reason. I will revert the change and add an explanation comment.

Created attachment 208320 [details] [review]
project-wizard: only add [CCode (instance_pos = -1)] when using GtkBuilder

Review of attachment 208320 [details] [review]:

Thanks!

If GtkBuilder is using GModule to find a version (ie: for signal handler names specified in the builder file) then this CCode attribute is necessary.

For signals that are connected manually in the code like:

  widget.signal.connect(func);

then you don't need the attribute.  The compiler is clever enough to figure it out.

In the case of the sniplet, the on_destroy function is used like so:

		window.destroy.connect(on_destroy);

and therefore never needs the attribute.

btw: The commit to revert this is broken in any case because you added back the attribute to main(), not on_destroy().

Created attachment 208341 [details] [review]
project-wizard: [CCode (instance_pos = -1)] should be in the signal callback

@Johannes: you shouldn't be accepting patches blindly, even coming from me :-p

@Ryan: Depending on whether the user chooses to use GtkBuilder or not, they get either:

		try 
		{
			var builder = new Builder ();
			builder.add_from_file (UI_FILE);
			builder.connect_signals (this);

			var window = builder.get_object ("window") as Window;
			/* ANJUTA: Widgets initialization for [+NameHLower+].ui - DO NOT REMOVE */
			window.show_all ();
		} 
		catch (Error e) {
			stderr.printf ("Could not load UI: %s\n", e.message);
		} 

or

		Window window = new Window();
		window.set_title ("Hello World");
		window.show_all();
		window.destroy.connect(on_destroy);

and in the first case, the signal connection is done using GtkBuilder

@Abdrerrahim: Well, I am wrong sometimes, I have too many patches to review and I need to trust on some people's opinion.