-------------
For 2.4
-------------
- add JetDefinition constructor without R [why????]
- Sort out zero-particle case [avoid crash, check many places this enters]
  [+ add constructor without particles]; [Requests by Andy Buckley]
- sort out e+e- R across different algorithms
- enable dynamic libraries by default?
- add "version()" info to doc            [done: MC & GS: check you're happy!]
- document error meaningless on AAEG.    [done]

- default f=0.75 for cones:
   a. change th edefault f to 0.75 in ctor
   b. add a ctor where f is not specified that prints a warning
   c. add a note in NEWS
   d. add a note in the doc (where?)
   e. add a note on the webpage
   f. add a note at the end of ./configure
  this has to be done for all the cone algs + internally in SISCone

- add ghosts over asymmetric range of rapidity  [done]
  [for Roman Kogler roman.kogler@desy.de]

-------------
High priority
-------------

- sort out issue in tiled strategies for small R (large memory use)
- finish testing e+e- kt, and sort out interface
- implement jade class? [find out about license for CP software]
- get CMS and ATLAS cones as plugins
- sort out the constructors for ranges of algorithms
- DIS?
- auto FAQ?

---------------
Medium priority
---------------

- write Fortran wrapper for all algorithms?

- get stable-cone pass number from the SISCone extras variable.

- sort out error in passive area with jetclu (calling function that
  should not be called)?

- sort out error in active area with jetclu (valgrind shouts a lot,
  programs die)? [may be resolved -- check?]

- write a function that produces a list of citations?

- decide whether to copy jets for the diB step. If so, pay attention
  to
  * parents(...), child(), partner()
  * ??? area transfer ???
  * anything highlighted with RETHINK-IF-CHANGE-NUMBERING

- alternatively change the types of jet & cluster index to separate
  types to reduce the likelihood of problems arising.

- v3: Think about making ClusterSequence actually just point to an 
  implementation class that does all the work; then both
  ClusterSequence and all the PseudoJets could point to it through a
  reference counting pointer -- this would allow one (e.g.) to get the
  constituents of a jet without manually going via the
  ClusterSequence.


----
Done
----
2.3.x
- moved to autotools
- fix bug in reconstruction of areas from hosted sequences (* Done *)
- move doc over to new area approach
- fix build problem for plugin subdirectories on macs (copy SISCone 
  sub-dir structure)
- fix midpoint speed issue
- fix windows compilation
- understand and then fix segmentation violation in
  bug-examples/voronoi-square-problem related to Voronoi diagrams of
  certain sets of points on a square grid.

2.4-devel
- finish implementing gen-kt
- first version of kt
