Re: user_data for callbacks

[Simon Schampijer <simon schampijer de>]

On 09/25/2012 07:24 PM, Daniel Drake wrote:
> Hi,
>
> Just checking the introspection of g_volume_mount():
>
> /**
>   * g_volume_mount:
>   * @volume: a #GVolume.
>   * @flags: flags affecting the operation
>   * @mount_operation: (allow-none): a #GMountOperation or %NULL to
> avoid user interaction.
>   * @cancellable: (allow-none): optional #GCancellable object, %NULL to ignore.
>   * @callback: (allow-none): a #GAsyncReadyCallback, or %NULL.
>   * @user_data: user data that gets passed to @callback
>   *
>   * Mounts a volume. This is an asynchronous operation, and is
>   * finished by calling g_volume_mount_finish() with the @volume
>   * and #GAsyncResult returned in the @callback.
>   *
>   * Virtual: mount_fn
>   **/
>
> The user_data parameter above doesn't have a (closure) annotation, but
> looking at the gir file, it doesn't seem necessary: the callback
> parameter has closure=4 already set.
>
> Python wants me to pass 5 arguments, i.e.
> volume.mount(flags, mount_op, cancellable, callback, user_data)
>
> I was expecting the user_data parameter to be optional, as it is with
> e.g. mybutton.connect('clicked', clicked_cb)
>
> Similarly, I am forced to provide a user_data parameter in my callback
> function to receive the value that was passed, otherwise python throws
> an exception about the callback receiving the wrong number of
> arguments.
>
> Is this the intended behaviour, or is there a bug somewhere?
>
> Thanks
> Daniel

Hmm, this sounds similar to:

"Using gtk_container_foreach or gtk_container_forall [1] does expect the
callback data to be passed in Python, it does work in a subclassed
container due to Carlos fix in the GtkCallback annotations [2]."

[1] 
http://developer.gnome.org/gtk3/3.4/GtkContainer.html#gtk-container-foreach
[2] 
http://git.gnome.org/browse/gtk+/commit/?id=db569cbee7e3842d802c5f1d53e28d0dde98ffeb

Regards,
   Simon