------------------------------------------
Status symbols
------------------------------------------
- still to do
* completed

------------------------------------------
Next minor release (BUGFIX release)
------------------------------------------
Version : 2.0.2
Due : 200?-??-??

[Building issues]

- bring improved code in showaccel demo to showaccel2 demo
* The default value of the configure option '--with-gii' should always be
  the one of the '--prefix' option, independent if '--prefix' is specified by
  the user or not, unless '--with-gii' itself is specified by the user.
  This simplifies compiling libggi on a machine with no root-permissions.
  (Christoph)
* Lots of OpenBSD/libtool fixes (Todd T. Fries)
* libtool update let GGI compile again under *BSD (including Darwin)
  (Christoph)


[Documentation related stuff]

- documentate tele-target in libggi/doc/targets.txt and via XML.
- make sure all Get/Put function manpages make it easy to understand 
  contents of *buf, if only by mentioning ggiUnPackPixels/ggiPackColors


[Target specific issues]

* added missing physz option to memory target (Brian)
* use config file instead of static accel_strings array in fbdev (Brian)
- alter fbdev.conf appropriately for selected accel sublibs
* why is directfb drawing on only every other pixel column? (was DFB bug)
* enable DirectFB (Brian)
* disable libKGI target, because it's currently a "proof-of-concept"
  (Brian)
* BUGFIX: Exceed is a X-server for Win32, which uses the ABGR instead of
  the RGBA colorspace. This causes colors being wrongly handled. The
  result on the screen are colors with blue characteristics. The bug
  is reproduceable when running apps using tile-over-Xlib with 
  directbuffers respectively simply running apps on Xlib. The bug 
  disappears by disabling the directbuffer using tile-over-Xlib.
  (fixed in new X-target)
- shake down new X target until most users cannot find any bugs that didn't
  already exist in display-X or display-Xlib.


[Common issues]

* Fix up gamma map functions for non-24bpp visuals (Brian)
- Alter all gcchanged functions so that unknown flags are not set, as
  an indication that they are not supported by the target yet.
* Make sure all target flush functions clip appropriately.  Brian has
  a patch that makes stubs and mansync flush virt.x, virt.y which can
  be applied after this is done.


------------------------------------------
Overnext minor release (BUGFIX release)
------------------------------------------
Version : 2.0.3
Due : 200?-??-??

[Building issues]

- Verify and fix cross-compilation to other architectures
- cleanup demo building and installation. That means, that demos requring
  a special header/input device like ggicalibrate should only be built,
  if possible.
- Fix libtool to create shared libs instead of static ones
  on Linux/MIPS boxes. This misbehaviour lets compile and install, but don't
  run GGI on Linux/MIPS boxes.
  This should be fixed by libtool update. Please verify.


[Documentation related stuff]

- write a target developers guide


[Common issues]

- Alter all ggiSetPalette implementations as per mailinglist discussion
  of 1999-07-03 (http://marc.theaimsgroup.com/?l=ggi-develop&m=93097803729315&w=2).
- remove obseleted Xlib target


[Target specific issues]

- update targets after the common issues are done
- make fbdev use of gamma map functions



------------------------------------------
Next major release
------------------------------------------
Version : 2.1.0
Due : ????-??-??

- libKGI-target (compiling framework, probably only working with Matrox Cards)
- improve modelist.inc
- allow physz params to determine window size when visible == GGI_AUTO
- add display-auto to separate OS dependent auto-detect routines from
  libggi core
- formalize pixelformat strings (e.g. "r5b5g5p1") and provide
  parser/construction functions for them in the LibGGI internal interface.
  Some code for this is already in LibBuf CVS.


New X Target:

- Make display-xf86dga work with all pre-4.x XF86, move newer DGA support
  to a new dga helper.
- 1,2,4-bpp support for new display-X.
- Test/complete display-X with obscure bitdepths/endian reversal.
- Verify/correct interaction of mouse events with setorigin.
- Re-implement old display-X emulated setorigin for root windows and
  for buggy WMs with faulty colormap handling code.
- See if colormap can be force-installed under some buggy WMs.
- Turn off hidden virtual area flushing when no backing store is 
  available.
- Get upstream to fix stuff (Solid-fill ops into backing store, for 
  example.)
- Compile on crusty old libc5 machines; replace any new-style X* symbols
  with backwards-compatible old-style symbols.
- dbe helper for new display-X
- dga helper for new display-X
- vidmode helper for new display-X
- Investigate emulating lower fixed-palette bitdepths and find out how to
  get those neat 8-bpp-pseudocolor-window-on-24bpp-server visuals to work.



------------------------------------------
Over-Next major release
------------------------------------------
Version : 3.0.0
Due : ????-??-??

- A working full-featured KGI-target
- change way that pixela/pixel stubs are selected, integrate ggiFlush
  and ggiSetFlags with the changed() dl cascade, add renderers (in addition
  to extensions) to the cascade, add some infrastructure for next 
  cut of LibGGIMISC to make it easy for extensions to integrate with automatic
  frame flipping.  Make sure it's all fast and efficient.
- enhance config file system: aliases, in-file option strings, wildcards
- enhance dl system: allow single library to serve multiple suggest strings
- move libmmutil from libgalloc to libggi to allow normal ggi primitives
  to use batchops
- finish long-term plan
- implement below API additions
- implement below non-backward-compatible changes


------------------------------------------
Long-term plan
------------------------------------------

- improve fbdev mode negotiation.  Option to obey /etc/fb.modes or not.
- move display autodetection routines into a display-auto sublib, and
  create display-auto-linux, display-auto-bsd, etc, mapping them to
  display-auto in libggi.conf.

------------------------------------------
Things that will induce API additions
------------------------------------------

- make sure all bitfields (e.g. BM_TYPE) have associated _MASK/_SHIFT macros
- Figure out all the caveats for cloning a visual and add core support 
	and a ggiClone API function.
- Add a ggiFlushNeeded that checks to see if a ggiFlush should be called,
  but allows application to continue on its merry way if so.

------------------------------------------
Non backward-compatible changes
------------------------------------------

- Remove old ggi versions of event types, which were replaced by gii typedefs.
