lots of methods return a gboolean that is redundant with their GError parameter; the return value is FALSE if the error is set, and TRUE if the error is clear.

In bindings that turn GErrors into exceptions, these return values are pointless, but mostly harmless, because you can just pretend the function is void. (Though some languages might warn about you ignoring the return value?)

However, in bindings that also turn out args into extra return values, the "real" return value becomes annoying. Eg:

gboolean g_shell_parse_argv (const gchar   *command_line,
                             gint          *argcp,
                             gchar       ***argvp,
                             GError       **error);

in gjs, you would call this by doing something like:

    try {
        let [success, argc, argv] = GLib.shell_parse_argv(command_line);
        ...;
    } catch (e) {
        ...;
    }

But the "success" variable is completely useless, because it will always be true.

So it would be nice if you could say:

 * Return value: (no-error): %TRUE on success, %FALSE if error set

and then the binding would know that the return value was redundant with the GError, and could strip it out. Presumably this would only be allowed when the return value type was gboolean.