
CHANGES in 0.2.3 (2019-12-31)

  * Added 'secrFit' wrapper for 'secr::secr.fit' with 'fastproximity = FALSE'

  * References to 'sample' replaced with 'draws' in all MCMC-related locations.

  * Fixed: bug in 'dt2'; also error in argument names for '*t2' functions;
    new '*t3' functions added.

  * Fixed: bug in 'occ2sps' when Hessian not invertible.

  * Converted all data files to .RData instead of R. Needed Depends: R (>= 2.10).

  * 'diagPlot' revamped; accepts a range of classes, not just 'Bwiqid'; additional arguments.

CHANGES in 0.2.2

  * Fixed incorrect handling of class 'bugs' objects in 'diagPlot' and 'as.Bwiqid'.

  * 'as.Bwiqid' revamped; makes more effort to retain original Rhat and n.eff; parameter
    names conversion changed: a1[1,2] now becomes a1.1.2. instead of a11.2; original
    names retained as attribute 'BUGSnames'.

  * Changed 'diagPlot' display for integers to dynamite plot instead of histogram;
    separate means for each chain in 'diagPlot' density plots.

  * Added 'crosscorrPlot' function for a range of classes.

CHANGES in 0.2.1

  * 'diagPlot' no longer uses its own colour palette.

  * 'window.Bwiqid' now adds new diagnostics for Rhat and n.eff.

  * Rhat calculated with 'simpleRhat', n.eff with 'safeNeff' everywhere.

CHANGES in 0.2.0

  * Continuous covariates now standardised to mean=0, SD=1 (instead of SD=0.5).
    Default priors for 'BoccSS' and 'BsurvCJS' changed to match this.

  * Added 'standardize' and 'standardize2match' functions.

  * Added 'simpleRhat' and 'getMCerror' functions; used by 'as.Bwiqid'.

  * 'diagPlot' gains a 'howMany' argument to plot less than the whole chain.

  * 'as.Bwiqid.data.frame' gains arguments 'n.chains', 'Rhat' and 'n.eff' and is
    included in the help page.

  * If the variance-covariance matrix cannot be calculated, MLE functions
    now return the log(likelihood) but number of parameters, df = NA.

  * 'AICc' gains a 'df' argument which overrides the df value in the model
    object, so an AICc value can be got even if the covariance matrix cannot be
    calculated.

  * Added *beta2 and *gamma2, wrappers for *beta and *gamma with mean and sd
    parameters, and *beta3 with mode and concentration.

  * Added getBetaPar and getGammaPar to convert mean/sd parameters to shape/rate.

CHANGES in 0.1.5 (2018-04-20)

  * Fixed check for installation of package shiny in showShinyApp.

CHANGES in 0.1.4 (2018-04-20)

  * new function densityFolded, used in plotPost and plot.Bwiqid when showCurve=TRUE.

  * summary.Bwiqid method added, prints background and returns values invisibly.

  * diagPlot now displays n.eff

  * fix: in as.Bwiqid.mcmc.list, call to gelman.diag wrapped in try.

  * fix: histogram breaks in plotPost are equidistant (as always intended).

  * fix: the first argument to BoccSS0 can be a matrix or data frame with a detection history.

  * fix: occSSrn0 now accepts scalar n, as documented.

CHANGES in 0.1.3 (2017-11-18)

  * Calculation of SEs of fitted values now uses less memory, works for large data sets.

  * Functions more robust to underflow/overflow: closedCap*, occ2sps, occSS*

  * Changed parameterization in Bsecr0 from lam0 to g0; better default priors.

  * Bug fix in BoccSS0: burnin iterations are now discarded.

CHANGES in 0.1.2 (2017-09-10)

  * Added diagPlot function and improved other diagnostic plots.

  * Added function Bpoisson; Bbinom renamed Bbinomial

CHANGES in 0.1.1 (2017-06-09)

  * Added 'coef', 'vcov' and 'predict' methods for 'wiqid' objects.

  * 'survCJS' gains an 'interval' argument to deal with time intervals of unequal length.

  * dots argument added to MLE functions to pass parameters to 'nlm'.

  * Improved coding of several functions to prevent under/overflow issues.

