1.2

- Fix for bug in code used in the tutorial from the BDVAL
  documentation.  The ExecuteSplitsMode would fail if called
  directly from the command line.  Using the supplied ant scripts
  would not execute this section of code and worked properly.
  See mantis issue #1356 for more details.

- Added an option to not calculate p-value estimates in candidate
  model selection.

1.1

- Support for running BDVal on Sun/Oracle Grid Engine
  (http://www.sun.com/software/sge/).

- Upgraded commons-math to version 2.0. Commons-math provides the
  T-Test implementation used by the t-test mode.

- Make T-Test calculation dependent on --logged-array parameter.
  Please note that calibration models must be rebuilt for approaches
  which use fold-change and process logged array datasets.

- Change template to use "--logged-array" rather than the older
  "--log-10-array"

1.0.4

- Update template scripts and configuration files to include
  missing parameters that were referenced in the project
  documentation and prostate example ("use-feature-selection-fold"
  for example).

1.0.3

- "Survival" mode option was not being set properly and caused
  the ExecuteSplits mode to fail when generating statistics.

- The correct name for the prostate example test split is
  now set to GSE8402-FusionYesNo-TestSplit.cids.

- The "predict-validation" target now sets the required attribute
  "model-directory" properly.

1.0.2

- Set the properties for "use-feature-selection-fold" to true in
  the  prostate-example.xml file. When these are not defined, the
  prostate example doesn't produce any results.

- Improve error handling in the "compound file" code used to store
  results.

1.0.1

- Fixed a crash due to missing class "ProgressLogger".
  See http://icbtools.med.cornell.edu/mantis/view.php?id=1314

- Updated icb-utils.jar to version icb-utils-144.  Previously
  included version included duplicate but older CompoundFile code.

1.0

- First release.