On Wed, Mar 14, 2018 at 03:08:58PM +0000, Alan Third wrote:
> On Tue, Mar 13, 2018 at 08:34:10AM -0700, Aaron Jensen wrote:
> > 
> > I haven't seen any glitches using this patch yet, fwiw.
> 
> I haven’t come across anything yet, but I’ve not been using this patch
> much.
> 
> Watch out for macOS re‐enabling screen updates after ~1 second. It
> does this automatically; I guess to stop buggy code from freezing the
> application completely.

It turns out that even with screen updates disabled some calls still
force the screen to update. I’m not sure which ones, but it does mean
we’re unlikely to completely lock up Emacs with this.

I’ve attached a new version of the patch.

Eli, I left the global variable as static since as far as I can tell
static global variables are still global to all threads, which is the
behaviour I believe we want. These functions should only be called
from the main thread anyway, Cocoa will kill the thread if GUI calls
are made from non‐main threads.

If it definitely shouldn’t be static, what should I do with it?

I moved the call to ns_enable_screen_updates in unwind_redisplay as
that’s where the double buffering code appears to unblock updates.
-- 
Alan Third