Changes in version 0.1.6 (2025-11-30)                  

No change - version increment due to submission mistake.

                        Changes in version 0.1.5                        

Fixed a lazily-written unit test.

                 Changes in version 0.1.4 (2025-09-01)                  

Custom boundary weight functions enabled through tape_bdryw(), the
results of which can be passed to tape_smd().

                        Changes in version 0.1.3                        

Bug fixes: export of evaltape_range() and keeprange(). Added a new tape
builder that averages range avgrange().

                        Changes in version 0.1.2                        

Exposing new functions for modifying the arguments of Rcpp_adfun
objects: fixdynamic() and fixindependent().

                 Changes in version 0.1.1 (2025-01-08)                  

No longer using RcppXPtrUtils. Small fix to creation of XPtr to ADFun in
movetoXPtr()

                        Changes in version 0.1.0                        

Major version number to reflect that functionality of the package is now
pretty good.

                       Changes in version 0.0.71                        

Cleaning up of help.

                       Changes in version 0.0.70                        

  - Looking for cmake in the /Applications/CMake.app/Contents/bin means
    installation works on mac.
  - Help for Rcpp_ADFun objects.
  - Removed old customll() associated functions/code
  - More stable test of cppad_search() due to differing convergence of
    optimx::Rcgmin() on differing OS.

                       Changes in version 0.0.69                        

Major improvement and breaking change: taping of custom log-density
functions occurs at the same time as compilation. This works across all
OS (previous customll() only worked on some cases of unix) and means the
major object passed between functions is a 'tape' object. Use tape_uld()
for this functionality.

  - New function tape_uld()
  - New class wrapping ADFun is from an Rcpp module and contains a
    special object (pADFun class) that points to an ADFun.
      - This new class has many of the properties available of ADFun.
      - It took a lot of experimentation to work out how to get expose
        this, and then further expose it to external R packages.

Converted all cases of std::exit to Rcpp::stop. See cout2Rcout.sh for
script.

                       Changes in version 0.0.68                        

CppAD errors handled when compiled in debug mode

                 Changes in version 0.0.67 (2024-07-25)                 

changed source code to use && or || for boolean objects

                       Changes in version 0.0.66                        

  - Deleted many unneeded files from original CppAD source.
  - Rearranged calling of wrapas.hpp. Now much of the types are declared
    in scorematchingad_forward.h. Then, following RcppEigen,
    scorematchingad.h includes scorematchingad_forward.h, then Rcpp.h
    then wrapas.hpp

                       Changes in version 0.0.65                        

Moved to a recent version of CppAD (version 20240000.5), which contains
some macros defined in ./inst/cppad/include/cppad/configure.hpp via a
call to cmake. Thus this package builds via a configure script for
installation. The CppAD source code is now in ./inst/cppad. After cmake
the ./inst/cppad/include/cppad directory is moved to
./inst/include/cppad/. One cpp file is also moved to ./src/

Also moved to using rlang instead of ellipsis to check_dots_used()

                 Changes in version 0.0.64 (2024-06-27)                 

URLs to CppAD and Eigen help pages, data locations and even paper
locations are breaking frequently enough for CRAN testing to pick up. So
I've removed a few urls in this version.

                       Changes in version 0.0.63                        

No change.

                       Changes in version 0.0.62                        

  - Bug for M1Mac test not floating point but actually because sign
    eigenvectors of Bingham A matrix are undetermined.

                       Changes in version 0.0.61                        

  - Bug fix to passing paramvec_start to ppi_robust()
  - More generous limit for equality for a M1Mac equality error
  - Faster examples

                 Changes in version 0.0.60 (2024-05-21)                 

Again, tiny polish of description.

                       Changes in version 0.0.59                        

Small polish of DESCRIPTION and examples in response to CRAN review.

                       Changes in version 0.0.58                        

Small clean up of package version number and spelling in help.