Re: [AD] About al_get_current_display and al_show_native_message_box

[Peter Wang <novalazy@xxxxxxxxxx>]

On 2010-11-22, David Capello <davidcapello@xxxxxxxxxx> wrote:
> On Mon, Nov 22, 2010 at 6:52 AM, Peter Wang <novalazy@xxxxxxxxxx> wrote:
> > On 2010-11-18, Peter Wang <novalazy@xxxxxxxxxx> wrote:
> >> On 2010-11-17, David Capello <davidcapello@xxxxxxxxxx> wrote:
> >> > Hello,
> >> >
> >> > One question, can we use "al_get_current_display" before calling "al_init"
> >> > or even if "al_init" fails? If we can, the following patch could be applied to
> >> > "examples/common.c" to avoid problems showing a message box
> >> > behinds the current display (e.g. in "ex_joystick_events.c" when we
> >> > don't have a joystick).
> >>
> >> I don't think it's officially supported, yet.  It may initialise some
> >> thread-local state, which could be somewhat complicated.
> >> We could make tls_get() return NULL if Allegro hasn't yet been
> >> initialised, then all of the al_get_* functions for thread-local state
> >> will return default values.
> >
> > Another alternative is to add al_is_system_installed.
> > This could be useful for other cases and arguably should be added
> > anyway for consistency.
> 
> Yes, that's a good idea. Indeed, I think it should be added
> for A5.0 so "al_get_current_display" can use it.
> 
> Anyway, could be "al_is_system_installed" implemented as follow?
> bool al_is_system_installed(void)
> {
>    return al_get_system_driver() != NULL;
> }

Yes.

We probably shouldn't have exported al_get_system_driver() actually,
so I have deleted its documentation.

Peter