On Feb 27, 2008, at 16:31 , Emmanuele Bassi wrote: > > On Wed, 2008-02-27 at 22:20 +0000, Ross Burton wrote: >> On Wed, 2008-02-27 at 22:32 +0100, Mikkel Kamstrup Erlandsen wrote: >>> Hmmm. This sounds like a very narrow minded decision to me. I have >>> been planning to write write a framework where one can send >>> GtkBuilder >>> XML snippets to a DBus service and have that service embed this >>> as out >>> of process plugins. >>> >>> The described GtkBuilder behavior forces me to write a separate >>> validator before I pass it to the GtkBuilder. Bugger. >> >> I don't believe that any data passed to a library should result in a >> fatal warning, surely a GError return would be a far better option >> here. > > this is data building your UI: it screams design tool, developer or > packager error, and I would expect GtkBuilder to error out very, very > quickly. if it were user input, or application data it would have > been a > completely different story - but the UI layout of an application *is* > the application, it's not user data.
But nobody can tell that the particular chunk of XML is "the UI layout of an application". It's not gtk_main_form(), it's gtk_parse_this_xml_and_build_something_of_it(). A plugin loading its UI from xml, is it user data, application data, or ...? (I don't quite understand how a glade file in /usr/share may or may not be The UI and not data, but then I also don't understand how people write applications which need a glade file on disk to start, so that's another story) Anyway, libglade doesn't crash, so it's a regression ;) Yevgen _______________________________________________ gtk-devel-list mailing list gtk-devel-list@gnome.org http://mail.gnome.org/mailman/listinfo/gtk-devel-list