TODO.txt

Performance:

* maybe add a cached serialization block or cache serialized data
  within rabin-ben-or / cevallos encode+decode. This should reduce
  serialization calls from O(n*(n-1)) / O(n^2) to O(n)

* add a serializedHashData call? This could be implemented within
  Krawcywk, etc. and should remove the need for those

   for(Share s : VSSShare shares)
       collect s.inner_share

  calls

* Instrument test-cases for performance measurements. Currently our
  assertions take approx. 1/8th to 1/4th of the overall CPU time

Testability:

* remove fake random generator and replace it with a pseudo random
   generator and a fixed seed

Readability

 * implement standard java serialization API

 * write real acceptance tests for the test engine

 * overwork unit tests!

 * write "real" decoder tests, not just "call rabin-ids and compare
   output performance"

 * create an unified ICCryptoEngine as RabinBenOrEngine and
   CevallosEngine are almost the same

Notes for Thomas

 * CustomMatrix is the culprit #1: 66% of CPU time is spent in there

 * try Netbeans 8.0, almost everything needed is included out of
   the box

 * can i use rabin-ben-or without BW?

 * why is rabin-ben-or with BW 4x as fast (decoding) as cevallos?