=========================================
 evgam v0.1.2 (Release date: 2020-04-18)
=========================================

Changes:

* The Fremantle data from package ismev have been added, and are used for examples. Usage is data(fremantle), as in ismev.

* colplot() adds the option to add a legend, which defaults to FALSE.

* logLik.evgam() now returns an object of class `logLik', allowing, e.g., AIC() and BIC() to be used.

* extremal0() has gone, as extremal() can now do the same.

* evgam()'s trace argument now allows -1, which suppresses any information on the console.

Bug fixes:

* Negative response data now work okay with family = "ald".

* evgams()'s formula argument may have smooths and parametric-only terms in any order. (Previously, smooths had
    to come first, so formula = list(response ~ s(), ~ 1, ~ s()) broke.)
    
* predict.evgam(object) with missing(newdata) only gave one set predictions for object$data. It now gives predictions
   for all rows of object$data (as it should).

=========================================
 evgam v0.1.1 (Release date: 2020-03-11)
=========================================

Changes:

* Compilation flag with clang++ in gradHess.cpp addressed.

* plot.evgam() now has informative y-axis labels for one-dimensional smooths.

* simulate.evgam() correctly labels variables for family = "response".

=========================================
 evgam v0.1.0 (Release date: 2020-03-08)
=========================================

Initial release.
