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

------------------------------------------
Next release
------------------------------------------
Version : 0.8.2
Due : ???

[Main]

- Get rid of the AC_TRY_RUN warning when running autogen.sh
  Note: This is related to endian detection done by a compiled and
  executed program AC_C_BIGENDIAN.  It might be better to create a catalog
  of endianess that is indexed by $host_cpu set by AC_CANONICAL_HOST.

* BUGFIX: Make devinfo related structures exist per-input (Eric)

* Verify, if devinfo related structures per-input BUGFIX works. (Andreas Beck)

* BUGFIX: Make libgii working in a multi-threaded environment (Andreas Beck)

- BUGFIX: Input targets that callback to functions in a LibGGI .so should 
  be able to sense when the visual has been closed/unloaded, and not call any
  backhooks (thus causing segv's).  They should remain inertly loaded (because 
  they can now be detached via ggiDetachInput and may be closed separately.)  
  Mainly this pertains to input-x/xwin but there are likely others.

- ISSUE: the devinfo flaw has been fix, but the problem remains at the
  valuator info level. Valuator info arrays should be copied for each device.
  The problem is that gii has no way of knowing how many valuators there are
  (shouldn't it be added in the devinfo structure with num_buttons and
  num_axes?). Two solutions IMHO: add this info in dev info struct, and gii
  takes care of the copy, or assume it's the driver's responsability to
  copy the array. I like the first one better.

[Documentation]

* Document using filters in the docbook (May need double check - Eric)
* Write a device driver/filter developers guide. (On the way - Eric)

[Modules]

* Check that all drivers are giving a device name, not just name of driver.
  (Ok, except for input-null that has no device, inputX whose state is unknown,
   and pcjoy that should not exist - Eric)
* Enumerate all devices in input-linux-evdev (According to Brian, this
  is not relevant - Eric)

------------------------------------------
Next release
------------------------------------------
Version : 0.9.0
Due : ???


[Main]

- Implement a per-device key/value hash (which is sent across the
  communications channel, encapsulated in events, such that it is 
  available via the gii_input handle even when using a remote device
  e.g. via filter-tcp)  This will be general purpose, but in the short
  term, the reason for it is to give device information strings
  which are longer than what will fit in a gii_event structure.  This
  hash is stateful, and must be updated when inputs are opened/closed 
  or devices are added/removed from an input.  See 06/2001 thread on ml
  entitled "device info"

- Port a lot of programs to use LibGII as their event loop.  This is
  the best sort of demo.  GPM, X, Wine are obvious targets.

- Allow an input/filter to destroy/remove itself in response to an 
  event, leaving the rest of the inputs in a joined input intact.

[Modules]

- (wishlist) refactor the keytrans/mouse filters, make them configurable 
  both by *optional* configuration file and via some sort of runtime 
  control structure (using the abovementioned hash ?).  Add joystick axis 
  sensitivity curve.  Write a neat GGI app that allows tuning 
  keyboard/mouse/joystick.



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

- Update our userspace drivers from other sources (GPM, linux-console) thus
  porting the code from these projects to non-Linux platforms.
