2do

o plot.MethComp is now a proper S3 method for MethComp objects.
  The former functions plot.MethComp and plot.VarComp have been renamed
  plotMethComp and plotVarComp and are called by plot.MethComp via the
  what= argument.

o A super-function check.MethComp added, which gives convergence diagnostics
  for a MethComp object, and optionally calls the trace.MethComp and post.MethComp
  functions.

Changes in 0.3.0
 
o New functions trace.MethComp and post.MethComp, giving the trace and 
  posterior densities for the parameters separately for the four chains.

o It is now possible to use BRugs(OpenBugs) for the simulations, by
  specifying program="OpenBugs", "OB" or "Brugs" (case insensitive) 
  in the call to MethComp. If nothing is specified MethComp will check 
  if BRugs is installed and if so use it, otherwise attempt to use WinBUGS. 

Changes in 0.2.2
 
o The BA.est function has been changed so that the LoA are based on
  p/m 2.00 instead of p/m 1.96.

Changes in 0.2.0

o The MethComp function revised and made somewhat modular. Default is
  now 4 chains, using different starting values for the variance components.

o The BA.est function is now used to generate sensible (and diverse) starting
  values for the variance components.

o A MethComp object is now of class mcmc.list, in order to allow use
  of the diagnostic tools from the coda package. Thus the posterior is now in
  a mcmc.list object (a list of matrices), transformed from the raw output
  mcmc.list object.
  
o R2WinBUGS and coda are not required any more, only suggested. Hence the
  BlandAltman plots etc. can be used without trouble loading the package even
  when R2WinBUGS and coda are not installed.

Changes in 0.1.19

o Typos in the documentation fixed (to some extent).

o plot.VarComp now takes range(grid) as the extent of the x-axes.

o Error in column naming in to.wide fixed.

Changes in 0.1.16

o BA.est has been expanded to handle different matrix effects (item by
  method) between methods if the number of methods exceeds 2. Thanks to
  advice from Peter Dalgaard.

o plot.meth now displays plots correctly. An argument to display the variable 
  names inside the panels have been added.

o BlandAltman (and hence BA.plot) has been enhanced to accept axis 
  specifications from limx, limy, eqax and ymax.

o BA.plot now does not give par() suggestions any more.

Changes in 0.1.13

o The default for the y-limit in the BlandAltman function now defaults to
  that the y-axis should have the same extent as the x-axis, symmetrically
  around 0, unless the differences or the limits of agreement are larger.