Following is a description of what is planned for Tablix in the next
year. Suggestions and comments are welcome as always.
Current Tablix kernel code is becoming increasingly hard to extend and
maintain. Large parts of the code should be refactored to allow for
greater modularity and readability. The Rottertablix project also
helped me found a few spots where code could be more general, which
would simplify modifications like this. Several important features are
still implemented as rough hacks in the code (for example
more_teachers.so module) and should be properly implemented with an
extended module API.
Experience with Tablix so far showed that the current genetic algorithm
is not very efficient with very constrained problems. Specially
timetables with a large number of "place capability" restrictions in
combination with different time constraints seem problematic. These
limitations could be solved with a slightly different genetic
algorithm. Changing the algorithm will most likely mean that most
modules will need to be rewritten.
All above modifications are planned to be included in the 0.2.0 release
of Tablix, which will probably be finished sometime before the end of
2005. A few weeks ago I tagged BRANCH_0_2_0 in the CVS, so anyone
interested in the Tablix development can it check out of repository. Be
warned though that code in this branch is currently a big mess and will
probably not even compile.
Branch 0.1.0 will still be maintained until the release of 0.2.0. I
expect two or three more stable releases of this branch, mostly to
include contributed modules and to fix any bugs that may be discovered,
but no new features will be added.
Received on Mon Dec 13 20:39:46 2004
This archive was generated by hypermail 2.1.8 : Tue Aug 16 2005 - 20:42:35 CEST