2.6
27/07/2015:
  NAMESPACE supplemented by imports from standard packages graphics, stats, utils

2.5
18/03/2015: 
  bug in bayessurvreg3 fixed which happened if 'inD' object was not calculated
  bayessurvreg3Para function added to allow for parallel simulations
   with bayessurvreg3 function

2.4
05/02/2015:
  CITATION file updated (desc changed to meta)

2.2
25/06/2013:
  bayessurvreg3 function calculates for misclassification models also DIC and pseudo-marginal likelihood,
  see attribute "fitStat" of the resulting object

2.1
06/2013:
  implementation of models involving possible misclassification of the event status

04/03/2013:
  some minor corrections and extensions invoked by icensKBL book

05/04/2012:
  basic maintanance required by CRAN

05/12/2008:
  bug in bayessurvreg2.priorBeta fixed which caused problems in bayessurvreg3 with fixed intercept as the only covariate

26/11/2008:
  bug in bayessurvreg2.cpp fixed which caused SEGFAULT in models with random intercept only (and no other covariates)
  bayessurvreg.design adjusted such that cluster(1:n) is no more needed in model formula
    when there are no random effects in the model included

27/03/2007:
  FILE: Gspline_update_a.cpp, FUNCTION: full_a_pars_uniCAR
  bug for order=3 when ia=1 or ija[0]=1 or ija[1]=1 fixed

06/02/2007:
  bug in bayessurvreg.design fixed. There was a problem when there was only 1 covariate in the model.

12/12/2006:
  function 'bayessurvreg3' allows for bivariate normal random intercepts acting in the onset
  and time-to-event part of the AFT model for doubly-interval-censored data

07/12/2006:
  function 'bayessurvreg3' allows for possible non-zero correlation between the random intercept
  in the onset model and the random intercept in the time-to-event model when analyzing 
  doubly-interval-censored data

06/12/2006:
  function 'rMVNorm' added

30/10/2006:
  FILE: Gspline_update_Scale.cpp, FUNCTION: Gspline::update_Scale
  gx+j changed to gx+j_ in
      full_sigma_logdens3(slice+j_, gx+j_, &dgx, &tmp, zeta_1, ipars, 3);
      solver_newton_raphson(slice+j_, gx+j_, &dgx, &horiz, zeta_1, ipars, full_sigma_logdens3, 
                            &iter_nr, &_maxiter_solver_nr, &_toler_solver_nr, &_epsilon, &err_nr);
  
  FILE: Gspline_update_sigma.cpp, FUNCTION: Gspline::update_sigma
  gx+j changed to gx+j_ in
      full_sigma_logdens3(slice+j_, gx+j_, &dgx, &tmp, zeta_1, ipars, 3);
      solver_newton_raphson(slice+j_, gx+j_, &dgx, &horiz, zeta_1, ipars, full_sigma_logdens3, 
                            &iter_nr, &_maxiter_solver_nr, &_toler_solver_nr, &_epsilon, &err_nr);

27/11/2005: 
  functions 'sampled.kendall.tau' and 'marginal.bayesGspline' added

18/11/2005: 
  function 'simult.pvalue' added

09/12/2004: 
  bug in in_output.cpp function 'writeToFiles' fixed.	If (!*storerP) 'writeAddToFile' function
  was incorrectly called with its second argument equal to 'nstore' and not 1.

08/12/2004: 
  bug in bayessurvreg.design fixed. There was a problem in creating the design matrix
  when random=~var and var was factor with two levels. I explicitely removed an intercept
  term from the design matrix and in this case it caused creation of a two-column matrix
  with one column=dummy for the first factor level and the second column=dummy for the 
  second factor level. However I wanted just one-column matrix with a dummy for one of
  the factor levels.




