Changes in version 2.9.2 (2024-09-27)

New features or changed behaviour of old functions
    o  minsimRSE new method for optimalSpacing simulations

Bugs fixed
    o  run.scenarios() with fit="multifit" failed with default (shared) mask
    o  estimateSummary() lost scenario number in some grouped examples
    o  run.scenarios() with model2D = 'IHP' could fail with NA start value for D

Other
    o  internal function dfcast accepts character detectfn
    o  plot.optimalSpacing tweaked to allow pch other than 1 and new plottype option
    o  run.scenarios() with trapset functions outputs a message as each is tested
    o  added anchors to external documentation links

Changes in version 2.9.1 (2024-04-20)

New features or changed behaviour of old functions
    o  run.scenarios() default extractfn keeps raw counts (including RPSV) as attribute
    o  run.scenarios() acquires argument 'prefix' for saving intermediate output
    o  expand.arg() new utility function to generate lists of arguments
    o  scenarioSummary() outputs overlap index k

Bugs fixed
    o  estimateSummary() failed when some estimates NA

Other
    o  run.scenarios() tweak to allow missing detectpar in det.pars
    o  estimateArray() and estimateSummary() more robust to failed model fit

Changes in version 2.9.0 (2024-03-01)

New features or changed behaviour of old functions
    o  multi-model fits (nested fit.args for one fitindex)
    o  run.scenarios() acquires argument 'joinsessions'

Bugs fixed
    o  run.scenarios() could ignore maskset
    o  run.scenarios() failed with byscenario = TRUE
    o  Enrm() and related functions: E(m) sometimes NA due to near-zero hazard

Other
    o  outputtype<- in-place replacement (also reassigns class)
    o  transformOutput() function to recast output
    o  select.stats tweak to allow movingAverage
    o  estimateArray() and estimateSummary() more robust
    o  byscenario = TRUE uses cluster type = FORK on Unix-like machine
    o  tweaks to c.estimatetables and c.selectedstatistics()
    o  run.scenarios() fitarg start may be NULL (dodges binomial problem with "true") 
    o  onAttach() options (matprod = "internal")
    
Changes in version 2.8.2 (2023-03-11)

New features or changed behaviour of old functions
    o  none
    
Other
    o  example time reduced for CRAN

Changes in version 2.8.1 (2023-03-10)

New features or changed behaviour of old functions
    o  fit option 'design' (assumes detection model cf Fewster & Buckland 2004)
    o  estimateArray() and estimateSummary() new functions

Other
    o  summary field 'var'

Changes in version 2.8.0 (2023-01-19)

New features or changed behaviour of old functions
    o  run.scenarios() fit.function may be "ipsecr.fit"
    o  run.scenarios() acquires argument 'CH.function'
    o  functions to combine output from different executions 
       of run.scenarios() (methods c, rbind for estimatetables, selectedstatistics)

Other
    o  suggests ipsecr
    o  schematic figure in help page ?secrdesign
    o  revised help page ?run.scenarios
    o  run.scenarios() sets RCPP_PARALLEL_NUM_THREADS if ncores provided

Changes in version 2.7.0 (2022-12-04)

New features or changed behaviour of old functions
    o  GAoptim() optimization by genetic algorithm using min(n,r) criterion 
       (Durbach et al. 2021) and others 
    o  minnrRSE() accepts GAoptim() output
    o  En2() and Qpm() for additional optimization criteria
    o  run.scenarios() option to create trapset dynamically with function(s)
    o  run.scenarios() compatible with dynamic density function in pop.args

Bugs fixed
    o  run.scenarios() failed with grouped scenarios if maskindex not specified
    
Other
    o  on GitHub
    o  added tests
    o  exports internal functions GApenfn() and compactSample()
    o  make.scenarios allows single-level 'groups' 
    o  check for consistent traps etc. across groups within scenario
    o  parallel clusters not used by run.scenarios() if fit = FALSE
    o  shift to Rcpp, RcppArmadillo from C
    o  summaryScenarios() ncores defaults to 1 (usually faster than >1)

Changes in version 2.6.0 (2022-01-26)

New features or changed behaviour of old functions
    o  run.scenarios() 'fit' option uses multithreading capability of secr
    o  run.scenarios() 'ncores' defaults to NULL (then uses value of environment 
       variable RCPP_PARALLEL_NUM_THREADS set by setNumThreads())
    o  run.scenarios() 'byscenario' defaults to FALSE

Bugs fixed
    o  Enrm() expected movement could fail when some of mask far from detectors

Other
    o  removed openCR options and imports (deprecated from 2.5.8)
    o  initial value of RCPP_PARALLEL_NUM_THREADS (usually 2) is set by secr onload 
    o  depends R (>= 3.5.0), secr (>= 4.2.0)

Changes in version 2.5.11 (2020-04-10)

New features or changed behaviour of old functions
    o  none

Bugs fixed
    o  run.scenarios() default extractfn (no model fitted) mistake in RSEp approximation

Changes in version 2.5.10 (2020-01-31)

New features or changed behaviour of old functions
    o  run.scenarios() default extractfn (no model fitted) uses RSEp approximation of 
       Efford & Boulanger 2019

Bugs fixed
    o   optimalSpacing() with simulations, fit.function = "openCR.fit" and nrepeats>1
        miscalculated RB(D-hat)
    o   summary.selectedstatistics() failed when run.scenarios() pop.args included poly
    o   header() stumbled on non-numeric (covariate) specification of D
    o   run.scenarios() spurious warning for non-numeric D
    
Bug not fixed (see Other)
    o   With fit.function 'openCR.fit', the calculation of RB(D-hat) in 
        summarized simulations does not reflect nrepeats>1

Other
    o   fit.function 'openCR.fit' is deprecated, and will be removed in a later version; 
        secr.fit is now fast enough
    o   optimalSpacing() catches bad root finding when limits inappropriate
    o   optimalSpacing() only passes valid fit.args to run.scenarios
    o   force ncores = 1 when fitting model (internal function processCH)
    

Changes in version 2.5.7 (2019-02-04)

New features or changed behaviour of old functions
    o  none

Bugs fixed
    o   count.summary() failed with fit.function = "openCR.fit"

Changes in version 2.5.6 (2019-01-26)

New features or changed behaviour of old functions
    o   make.scenarios() allows uniform (detectfn = 4)
    o   count and predict methods for simulation objects inheriting from 'summary'
    o   select.stats() respects count statistics

Bugs fixed
    o   plot.optimalSpacing() failed if any values missing (NA)
    o   optimalSpacing() did not apply nrepeats when computing binomial rotRSE
    o   scenarioSummary() failed with single-detector arrays

Other
    o   vignette removed (had been re-introduced by mistake in 2.5.5) 
        (see www.otago.ac.nz/density/pdfs)

Changes in version 2.5.5 (2018-12-07)

New features or changed behaviour of old functions
    o   optimalSpacing() new argument 'fit.function' to select openCR.fit/secr.fit
    o   optimalSpacing() new argument 'distribution'
    o   optimalSpacing() adjusts rotRSE for binomial distribution of n
    o   plot.optimalSpacing() does not plot optimum
    o   summarySpacing() calculates detectors per home range

Bugs fixed
    o   run.scenarios() tripped up on multi-session simulation
    o   summaryScenarios() mostly ignored nrepeats

Other
    o   imports from openCR

Changes in version 2.5.4 (2017-12-05)

New features or changed behaviour of old functions
    o   none

Other
    o   vignette secrdesign-vignette.pdf removed (see www.otago.ac.nz/density/pdfs)

Changes in version 2.5.3 (2017-12-04)

New features or changed behaviour of old functions
    o   saturation() new function for expected detector saturation (trap success)
    o   scenarioSummary() reports average expected detector saturation
    o   derived.SL() and regionN.SL() replaced by methods for generic functions
        derived and region.N (see 'secr')

Bugs fixed
    o   Enrm() etc. detectfn 'HAN' and 'HCG' failed

Other
    o   tweak to select.stats to allow parameters without true value in scenarios
    o   depends on secr >= 3.1.2

Changes in version 2.5.2 (2017-10-10)

Other
    o   vignettes secrdesign-Enrm.pdf and secrdesign-tools.pdf removed (see www.otago.ac.nz/density/pdfs)

Changes in version 2.5.0 (2017-10-06)

New features or changed behaviour of old functions
    o   functions Lambda(), Enrm(), minnrRSE(), costing(), getdetectpar() and optimalSpacing()
    o   new function scenarioSummary() reports deterministic summary of scenarios (e.g., expected number 
        of individuals, costings)
    o   run.scenarios() and fit.models() acquire argument byscenario (parallel scenarios or replicates?)
    o   fit.models() mask may be passed as component of fit.args, when it has priority over maskset
    o   run.scenarios() user may specify different masks as core in pop.args
    
Bugs fixed
    o   run.scenarios() ignored Nmark detection argument
    o   chatnsim lost if multiple scenarios
    o   nrepeats was ignored when method = 'none'

Other
    o   more robust outputtype when no results
    o   packageType message if pre-release
    o   persistent problems in header() for non-scalar arguments

Changes in version 2.4.1 (2016-09-28)

New features or changed behaviour of old functions
    o   rule-of-thumb RSE added to default extractfn when fit = F
    o   validate() acquires argument 'quietly'
    o   detectpar list may be specified in det.args (overrides scenario)

Bugs fixed
    o   summary.selectedstatistics() failed when type = 'array'

Other
    o   all array dimensions now named in output of summary.selectedstatistics()

Changes in version 2.4.0 (2016-04-23)

New features or changed behaviour of old functions
    o   mark-resight data are simulated with secr::sim.resight if traps object
    	has markocc attribute
    o   run.scenarios() new argument chatnsim for simulating overdispersion of sightings
    o   header() allows for complex 'details' and 'start' components
    o   validate reports number of problem rows
    	
Bugs fixed
    o   header() failed with model2D="IHP" (problem in argdf - temporary fix)
    o   run.scenarios() respects detectpar component of det.args (see makeCH)

Other
    o   number of occasions in markocc overrides noccasions in scenario
    o   internally streamlined use of multiple cpu (no unnecessary export)
    o   select.stats() failed with parameter not 'D'
    o   run.scenarios() uses message instead of cat
    
Changes in version 2.3.1 (2015-05-14)

New features or changed behaviour of old functions
    o   scenariosFromStatistics() new function to generate scenarios with specified
         number of first captures and recaptures.
    o   improved output header for fit.args etc.


Changes in version 2.3.0 (2015-03-25)

New features or changed behaviour of old functions

    o   fit.models() acquires arguments 'scen' and 'repl' to select a subset of rawdata
    o   run.scenarios() and fit.models() when predict.secr returns a list the default 
        extractfn selects first session or group, with a warning (previously crashed)
    o   summary.selectedstatistics() has improved labelling of output
    o   select.stats() acquires argument 'true' for user override of 'true' values

Bugs fixed
    o   fit.models() did not work with grouped populations
    o   default extractfn failed when no detections and fit = FALSE
    o   run.scenarios() dots argument (...) not passed to worker processes

Other
    o   more robust handling of start values for secr.fit()

Changes in version 2.2.2 (2015-01-11)

Other
    o   CITATION file uses package metadata

Changes in version 2.2.1 (2014-11-27)

New features or changed behaviour of old functions

    o   run.scenarios() allows value "linear" for model2D argument of sim.popn 
    o   details 'userdist' argument suppressed when printing summaries
    o   make.scenarios() optionally generates scenarios with group structure (e.g. males and females)
    o   run.scenarios() simulates composite (grouped) populations

Bugs fixed
    o   validate() did not catch NaN values

Other
    o   attribute 'regionarea' renamed 'regionsize' to cover linear case
    o   workflow figure dropped from secrdesign-package.Rd (see vignette)
    o   argument order changed in summary.selectedstatistics
    o   default extractfn checks for non-null 'fit' of returned secr object

Changes in version 2.1.3 (2014-06-04)

    o   vignette renamed and edited
    o   vignette source code included

Changes in version 2.1.2 (2014-04-29)

New features or changed behaviour of old functions

    o   validate() function to set out-of-range values to NA
    o   min, max as fields in summary.selectedstatistics
    o   secrdesign.pdf vignette
    o   header in summary output now includes pop.args
    o   run.scenarios() allows IHP population simulation as 
        alternative to specifying uniform density

Bugs fixed

    o   maskset was not saved when specified in run.scenarios()