Re: 'struct window' cleanup #3

[Eli Zaretskii]

> Date: Wed, 27 Jun 2012 21:59:58 +0400
> From: Dmitry Antipov <address@hidden>
> CC: address@hidden
> 
> I would like to propose alternate redesign for this: init_window_once
> sets Vframe_list to Qnil. After calling make_initial_frame, Vframe_list
> becomes non-nil, so !NILP (Vframe_list) means init_window_once was
> passed.

How is this better than the current use of window_initialized?  At
least that flag is clearly meant to indicate that window.c has been
initialized.  By contrast, if we use the fact that Vframe_list is
initialized in make_initial_frame that is called by init_window_once,
we are piggy-backing a mechanism that was never meant to be such an
indication.