0.2-1  Fixed bug where pp and psum were int instead of double in potts.c

0.3    Changed niter argument to three: nbatch, blen, nspac
       Now works just like metrop in the mcmc package

0.3-1  Fixed bug where |A| in |A| * theta_color was incorrect (cc instead of gg)
          Bad bug.  Simulations all wrong before.

0.3-2  Fixed bug where no error was reported for theta-star < 0, for
           which Swendsen-Wang algorithm doesn't work.
       Also introduced debug argument to potts, and "expose the innards"
           validation (tests/debug.R).  So far only boundary == "torus"
           mode validated, but looks o. k. except for ...
       One weird open bug.  The saving of .Random.seed as out$initial.seed
           doesn't work right.  Two runs are not the same as continuation
           of one run.

0.3-3  Fixed the "weird" bug.  .Random.seed must be assigned in the
           R global environment to have effect, see use of assign function
           in mcmc/R/potts.R

0.3-4  Finished validation.  Now all boundary specifications validated.
           Also greatly enlarged details section of help page for potts
           function.

0.4    Added outfun

0.5    Added composite likelihood stuff

0.5-2  moved vignette to vignettes directory

0.5-3  As per instructions from CRAN, redid *.Rout.save files in
           tests directory that used ks.test because print.htest
           changed in core R

0.5-4 Explicitly import stuff from stats and graphics, required by R-3.3.0

0.5-5 Modernize registration of native routines as per Sections 5.4 and 6.15
      of Writing R Extensions

0.5-6 Put DOI in DESCRIPTION

0.5-7 Fix horrible bug (cmpfun inside function called many times)

0.5-8 Eliminate calls to cmpfun.  Just rely on jit.

0.5-9 Add calls to R_CheckUserInterrupt(); in MCMC main loop

0.5-11 Change \textbackslash to \setminus