From: mentalg@geocities.com
  To  : andreas.beck@ggi-project.org
  Date: Sat, 12 Sep 1998 15:17:37 -0400 (EDT)

Re: Open list of issues before Degas goes out

On Sat, 12 Sep 1998 becka@rz.uni-duesseldorf.de wrote:

> > > > * Obviously, decide something on that synchronous rendering topic.
> > > 	The consensus seems to be that ASYNC mode has to be made the 
> > > default and SYNC mode be emulated with a target, right?
> > Eh, basically the SYNC stuff gets removed from libggi proper, and we make a
> > little helper library that basically spawns a thread and calls ggiFlush() a
> > lot.  
> 
> For the last time: That won't work. Please prove me wrong _before_ doing
> such a change.

spawning a thread and calling ggiFlush() a lot is _exactly_ what mansync
does... although it tries to fake it with signals when pthreads support
isn't availible, which is (IMO) bad unless we can make the entirety of
libggi async-signal-safe...

> Either we drop it totally allowing only for the stupid "sync after each
> drawing command", or more-or-less leave it as is.

> So far I like Marcus' idea of making ASYNC default and enabling old mansync
> code, if someone turns on SYNC mode best.

I could probably live with that.  But I'd rather not see things breaking
when someone turned mansync on...

> It encourages using ASYNC, while not killing a working and useful system
> which cannot be replaced easily.

I'm basically just suggesting moving the existing implementation into an
extension library, instead of the helper library it currently resides in.
And maybe dropping the signal implementation if we can, somehow.  Well, and
not making SYNC the default.

-=MenTaLguY=-