[PATCH weston-ivi-shell v4 3/9] ivi-shell supports a type of shell for In-Vehicle Infotainment system.

[Nobuhiko Tanibata]

Hi pq,

I applied your comments except for several ones,

>> +    struct ivi_shell *shell = wl_resource_get_user_data(resource);
>> +    struct ivi_shell_surface *ivisurf = NULL;
>> +    struct weston_layout_surface *layout_surface = NULL;
>> +    struct weston_surface *es = 
>> wl_resource_get_user_data(surface_resource);
>> +    struct wl_resource *res;
>> +    int32_t warn_idx = -1;
>> +
>> +    if (es != NULL) {
>> +        layout_surface = weston_layout_surfaceCreate(es, id_surface);
>> +        if (layout_surface == NULL)
>> +            warn_idx = 1;
>> +    } else {
>> +        warn_idx = 0;
> 
> I think this case would be a compositor internal error, not a client
> error. Userdata of a surface resource should never be NULL. I.e.
> assert() kind of thing would be more approriate.
> 

I expect it would send warning to client. So I implement it like the 
above.


>> +    init_ivi_shell(ec, shell);
>> +    weston_layout_initWithCompositor(ec);
>> +
>> +    shell->destroy_listener.notify = shell_destroy;
>> +    wl_signal_add(&ec->destroy_signal, &shell->destroy_listener);
>> +
>> +    if (wl_global_create(ec->wl_display, &ivi_application_interface, 
>> 1,
>> +                         shell, bind_ivi_application) == NULL) {
>> +        return -1;
>> +    }
> 
> You are not setting compositor->shell_interface, but OTOH those are
> only used for Xwayland, which probably doesn't make sense here.

I was advised by kritian before it shall not support other shell in 
ivi-shell at the same time.

BR,
Nobuhiko

> 
>> +
>> +    return 0;
>> +}
> 
> 
> Thanks,
> pq