Round the parameter values in the test parameter sets rather than using values with many digits. Also include reference values for cdf obtained from Maple.

Have pdf return NA and warning rather than stop when encountering
wrong parameter values.

Use Diethelm's skewness and kurtosis from fBasics rather than mine.

For profile likelihood, look at maximum likelihood package which
creates wrappers to fix parameters.

Protect all calls to optim etc with try and an error message catch
Add silent argument to fit functions
Add errMessage to return list for all fit functions to be "" if no error
Add optOut to return list??
Fix documentation also: arguments and return list

Add more unit testing routines

Add data sets from Eric Morway as examples.

Add data set from Kevin Lin as an example.

Not really important, but for completeness, add lower.tail, log and
log.p arguments to dpqr functions for skewlap. This area is a bit of a
mess with the different packages having different arguments allowed of
this type.



