Changes in robmixglm version 1.2-3 o fail gracefully if no starting values found. o fix e-mail address. Changes in robmixglm version 1.2-2 o changed starting values for fitting in bootstrap outlier test. This causes a slight change in results. Changes in robmixglm version 1.2-1 o changed gaussian so that outliers have variance sigma^2 outliers rather than sigma^2+tau^2. o improved speed of multiple cores o changed default number of multiple cores to number of cores divided by 2 (this has only slightly less speed) o tidied up vignette Changes in robmixglm version 1.2-0 o added negative binomial distribution o fixed bug causing too many displays of failure to find starting values message o increased number of random starting values to 50 to make finding the global maximum more likely o improved calculation of standard errors Changes in robmixglm version 1.1-1 o use multiple cores for determining starting values (if this causes problems then set cores=1) o tidied up code for testing outliers o improved error handling in outlier test o fixed bug in starting values for gaussian models Changes in robmixglm version 1.1-0 o added predict method o implemented bootstrap for testing for outliers using multiple cores (if this causes problems then set cores=1) o removed progress show from outlier test (this won't work properly with multiple cores) o set examples to donttest to satisfy CRAN checks Changes in robmixglm version 1.0-4 o fixed numerical problems o improved selection of starting values, especially with small number of observations Changes in robmixglm version 1.0-3 o fixed error due to changes in class() behaviour by replacing with inherits when appropriate Changes in robmixglm version 1.0-2 o added extra sentences in description o changed dontrun to donttest o changed several examples to allow running as part of check Changes in robmixglm version 1.0-1 o first release