On Thu, 18 Feb 2016 08:48:40 +0900
Wataru Natsume <[email protected]> wrote:

> From: Wataru Natsume <[email protected]>
> 
> Previous code cleaned up surfaces in layer once and then added
> surfaces to a layer in random. In this flow, two commitchanges are
> required.
> 
> Signed-off-by: Nobuhiko Tanibata <[email protected]>
> [[email protected]: Removes unnecessary check]
> Signed-off-by: Wataru Natsume <[email protected]>
> 
> ---
> Changes from v1 - Removes unnecessary check if the surface is on a layer.
> 
>  ivi-shell/hmi-controller.c |    9 ---------
>  1 file changed, 9 deletions(-)
> 
> diff --git a/ivi-shell/hmi-controller.c b/ivi-shell/hmi-controller.c
> index 8da3d3c..ace6555 100644
> --- a/ivi-shell/hmi-controller.c
> +++ b/ivi-shell/hmi-controller.c
> @@ -424,18 +424,9 @@ mode_random_replace(struct hmi_controller *hmi_ctrl,
>  
>       wl_list_for_each(application_layer, layer_list, link) {
>               layers[layer_idx] = application_layer;
> -             
> ivi_layout_interface->layer_set_render_order(layers[layer_idx]->ivilayer,
> -                                                     NULL, 0);
>               layer_idx++;
>       }
>  
> -     /*
> -      * This commit change is needed because ivisurface can not belongs to 
> several layers
> -      * at the same time. So ivisurfaces shall be removed from layers once 
> and then set them
> -      * to layers randomly.
> -      */
> -     ivi_layout_interface->commit_changes();
> -
>       for (i = 0; i < surface_length; i++) {
>               ivisurf = pp_surface[i];
>  

Hi Natsume-san,

this looks fine at first, but when testing it, mode_random_replace()
will trigger one "ivi_layout_layer_add_surface: addsurf is already
available" warning per existing surface.

ivi_layout_layer_add_surface() is checking if the surface is already
(current, not the pending state) on the given layer. This is likely
because in a previously intended future a surface might be in multiple
layers, and adding it multiple times to the same layer is considered a
mistake (given how surface positioning works in this ivi-layout API
design, that is understandable).

Maybe we should also just remove that check from
ivi_layout_layer_add_surface()? I don't see any value from it in the
current code base. If Emre adds views as a tying object in the
ivi-layout API, this code will get rewritten anyway.

Apart from the harmless log spew, this patch is:
Reviewed-by: Pekka Paalanen <[email protected]>

If you want to make a patch to remove the warning, I can push the both
patches at the same time.


Thanks,
pq

Attachment: pgp1mYcX6Jlwe.pgp
Description: OpenPGP digital signature

_______________________________________________
wayland-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/wayland-devel

Reply via email to