Re: [AD] SF.net SVN: alleg:[13312] allegro/branches/4.9

[Thomas Fjellstrom <tfjellstrom@xxxxxxxxxx>]

On April 24, 2010, tjaden@xxxxxxxxxx wrote:
> Revision: 13312
>           http://alleg.svn.sourceforge.net/alleg/?rev=13312&view=rev
> Author:   tjaden
> Date:     2010-04-24 14:28:41 +0000 (Sat, 24 Apr 2010)
> 
> Log Message:
> -----------
> Init xinerama/randr/xf86vm interfaces lazily.

I'm not entirely sure this is a good idea. Especially once xrandr events are 
supported, xrandr will need to be inited on startup.

> Getting the information from randr can cause flickers and background
> sound to stutter, so delay doing so until actually necessary.

So why not do it at the beginning?
 
> Modified Paths:
> --------------
>     allegro/branches/4.9/include/allegro5/internal/aintern_xglx.h
>     allegro/branches/4.9/src/x/xfullscreen.c
>     allegro/branches/4.9/src/x/xsystem.c
> 
snip
> +static void init_mmon_interface(ALLEGRO_SYSTEM_XGLX *s)
> +{
> +   if (s->mmon_interface_inited)
> +      return;
> +
> +   /* Shouldn't we avoid initing any more of these than we need? */
> +

You could probably avoid initing xf86vm if you really want to. But Xinerama 
is tightly intertwined with both the xf86vm and XRandR code.

> +#ifdef ALLEGRO_XWINDOWS_WITH_XINERAMA
> +   _al_xsys_xinerama_init(s);
> +#endif
> +
> +#ifdef ALLEGRO_XWINDOWS_WITH_XF86VIDMODE
> +   _al_xsys_xfvm_init(s);
> +#endif
> +
> +#ifdef ALLEGRO_XWINDOWS_WITH_XRANDR
> +   _al_xsys_xrandr_init(s);
> +#endif
> +
> +   if (mmon_interface.store_mode)
> +      mmon_interface.store_mode(s);
> +
> +   s->mmon_interface_inited = true;
> +}
> +
snip


-- 
Thomas Fjellstrom
tfjellstrom@xxxxxxxxxx