@Susurrus please check that this don't broke gtk-rs/gtk#469

Might make sense to add a simple test for this, looks good and correct to me apart from that

No idea how to test it :(

How did you get the bug?

See gtk-rs/examples#119 (comment)

@EPashkin You could manually create such a NULL-terminated array of e.g. strings with glib_sys::g_malloc() and ::g_strdup(). When passing that in there it should crash.

@sdroege, Thanks I don't even think to test it (forgot that it not Gtk and can be tested).

gtk-rs/gtk#469 is still fixed with this code, but this doesn't seem to fix gtk-rs/examples#119. I'm getting new errors now, like:

• (.:12663): Gtk-CRITICAL **: GtkBox 0x7f8893b191a0 returned a widget path for type (null), but child is GtkScrolledWindow
• (.:12663): Gtk-CRITICAL **: gtk_style_provider_get_style_property: assertion 'g_type_is_a (gtk_widget_path_get_object_type (path), pspec->owner_type)' failed
• (.:12809): GLib-GObject-WARNING **: g_object_weak_unref: couldn't find weak ref 0x7ff1eb45c5e0(0x7ff1ece22ed8)

But I don't get hard segfaults anymore with no warnings or errors, so I guess that's an improvement.

@GuillaumeGomez ready for merge

Thanks!