Index: [thread] [date] [subject] [author]
  From: Jon M. Taylor <taylorj@gaia.ecs.csus.edu>
  To  : GGI mailing list <ggi-develop@eskimo.com>
  Date: Thu, 10 Jun 1999 21:36:03 -0700 (PDT)

Even yet still more GGIMesa updates

	New stuff:

* GenKGI helper autodetects the presence of an accelerated KGI driver 
(Savage4 only for now) and loads the appropriate helper.

* The fbdev now target piggybacks its accel-handling on top of the 
underlying LibGGI fbdev target's accel handling.

* Huge directory-layout reorg.  GGIMesa's directory structure now mimics that
of LibGGI almost 100%.  The only major difference is that LibGGI installs its
sublib .conf files into the top-level etc/ggi/display directory, while
GGIMesa puts everything but ggimesa.conf into etc/ggi/mesa/[display].  
Everything should be nice and clean and in its proper place now.

* Tons of misc bugfixes.

	This will probably be the last major GGIMesa structural update for a
while.  I have what I need to continue my Savage4 work without needing any
ugly hacks that would affect other GGIMesa users/developers, the whole thing
should be LibGGI 2.0 compliant, and all major features are in place.  The 
only loose ends I see right now are:

* The sublibs which use a .conf file should use LibGG to parse it 
themselves instead of .include'ing the .conf file in the top-level 
extension library's .conf file.  I'm going to put this one on the shelf 
until Andy reimplements the vendor/ subsystem.

* The GGIMesa genkgi helper is hardwired to autodetect the presence of a
Savage4 only, no other KGI 3D accelerated drivers are probed for.  If a
Savage4 is not detected, the genkgi helper does nothing and generic-* libs
are used.  The KGI acceleration detection is an ugly hack and will need to be
replaced with a general-purpose suggest-string parser at which point there
are some other 3D accelerated KGI drivers in existence.  Until then, this one
goes on the shelf as well.

* The GGIGLUT situation is a real mess.  GLUT is (C) by Mark Kilgard, and
although it is open-source, it is NOT redistributable with modifications.  It
isn't even in Mesa's CVS tree right now |-<.  Based on this, I have doubts
about the legality of GGIGLUT, because it must be linked to the rest of GLUT
in order to function.  Some talk about forming a free-GLUT project has been
seen on the mesa-dev list, and of course GGI would be the best platform to
base it upon, but since that would be a huge amount of work and QuakeIII
doesn't use GLUT anyway, someone else will have to take this one on.  In the
meantime, building of ggiglut.c has been disabled in the Mesa automake system
and will remain disabled until the situation is resolved. 

* A good chunk of Uwe's original GGIMesa code remains in some form, and that
code is LGPLed.  Since Mesa 3.1 is moving to a X-style license, this is
technically a violation of the LPGL.  I haven't had any luck in contacting
Uwe, so I guess I will just not worry about it unless and until he complains.

	Enjoy the new GGIMesa, folx.  Up next: The first Savage4 KGI/Mesa
driver beta!  I can't say exactly when this will be released, but it
shouldn't be too much longer, now that all the necessary GGIMesa cleanups
have been taken care of and I have a killer development environment to work
within.  Soon, all the world will know the true power of GGI.... 

Jon

---
'Cloning and the reprogramming of DNA is the first serious step in 
becoming one with God.'
	- Scientist G. Richard Seed

Index: [thread] [date] [subject] [author]