v2.0.0

Known issues: https://github.com/PredictiveEcology/SpaDES/issues

version 2.0.0

Dependency changes

• SpaDES package split into several (#198):

  • caching moved out of SpaDES and into new package reproducible. reproducible added to Imports.
  • plotting moved out of SpaDES and into new package quickPlot. quickPlot added to Imports.
  • all Rstudio addins moved out of SpaDES and into new package SpaDES.addins.
  • core DES components moved out of SpaDES and into new package SpaDES.core. SpaDES.core added to Imports.
  • additional modelling utilities (non-core components) moved out of SpaDES and into new package SpaDES.tools. SpaDES.tools added to Imports.
  • shiny-related components moved out of SpaDES and into new package SpaDES.shiny.

• minimum R version increased from 3.2.2 to 3.3.3 as required by several dependencies.

• added bit, fastmatch, fastdigest and Rcpp packages to Imports.

• removed stringr package from Imports.

Defunct

• getFileName. Wasn't used.
• p. Use params or P instead.
• shine moved out of SpaDES into the SpaDES.shiny package.
• versionWarning. Wasn't used.

New vignette

• iii-cache which shows many of the ways to use Cache and build it into a larger reproducible workflow.

New functionality

• namespaced module functions have now been implemented via a nested environment in the [email protected] slot. This means that module functions can be called by their name only, without a sim$ prefix. Also, there should not be any name clashes between modules, so each module can have its own init function, say. This has been implemented in a backwards compatible way, but the old way may be deprecated down the road.
• new functions saveSimList which saves all environments recursively and file-backed objects, such as Raster objets (currently only one implemented).
• Created a generic for Copy, moved it to reproducible, and here added a method for simList objects that deep copies all environments recursively.
• change default value for speedup in gaussMap; now 1.
• new function getPaths() to return the list of working dirs from the options
• new function setPaths() as wrapper for setting the options for working dirs; uses ~/SpaDES as default base path
• downloadModule checks for and uses environment variable GITHUB_PAT if it exists (alleviates 403 download errors caused by GitHub download limits)
• spread2: new function that is more robust than spread, slightly slower under some situations, but faster in many others. The function is designed to be used as a building block for more complex spreading, where the user can wrap spread2 inside a custom function that iteratively calls spread2, while optionally changing any of the input arguments.
• adj - has a new argument returnDT, which slightly improves speed in cases where the user wants the output to become a data.table. This simply prevents a call to as.matrix in the return(), which was occurring if the return value was a data.table.
• new function copyModule for creating a copy of an existing module

Performance

• add C++ internal functions for speed
• event queues (event, completed, current) now do shallow copies of pre-existing data.tables, rather than rblindlist( ). This improved DES speed by > 40%. Benchmarking of 1400 events in 1.6 seconds now; previously 3.0 seconds.
• add Copy (capital C) as replacement for copy which conflicts with data.table::copy, and add queues argument to do a deep copy of the event queues.
• spread: minor speedups, plus 3 new parameters relativeSpreadProb, numNeighs & exactSizes
• distanceFromEachPoint: uses faster Rcpp code for some use cases & can now take arbitrary columns in either from or to arguments
• use %fin% as faster replacement for %in%

Bug fixes

• workaround issues with RStudio graphics on Linux (with #116): use dev.noRSGD() to bypass the Rstudio graphics device for your current session (sets the device option for your platform).
• checksums(..., write = TRUE) ignores the contents of CHECKSUMS.txt, overwriting that file with the checksums of all files in the module's data/ directory. This makes it easier to update the checksum file, e.g., when adding new data (#332).
• improved module versioning (#321)
• minor bugfixes for unusual cases
• some broken examples were fixed

Other updates

• improved documentation

• modules are now run with their required packages (reqdPkgs in metadata) temporarily bumped to the top of the search path (i.e., search()) during each event. search path is restored on.exit from spades or simInit call

• change order of parsing of modules: defineModule is last, so can use objects defined within module for parameters

• improved simInit debug mechanism -- passing character string of module name will enter into a browser call inside doEvent

• improvements to caching --> these are moved to reproducible package:

  • improved caching for Raster* objects & S4 methods - now it normally persists across sessions;
  • add Cache (upper case) which derives cacheRepo arg automatically from either the cachePath(sim), if used within a module, or getPath()$cachePath if not within a module. Also, the upper case removes the name conflict with archivist::cache;
  • uses fastdigest::fastdigest for RAM objects and digest::digest for disk-backed objects;
  • add caching mechanisms at the module-level and event-level (via new .useCache parameter, which can be logical indicating whole module or character indicating individual events);
  • add caching for .inputObjects function in simInit, via .useCache parameter in module
  • detailed caching overview now in cache help: ?Cache for details.
  • strips dirname for outputs and inputs, i.e., only keeps the filename, not absolute paths. This may not be stringent enough in some cases.
  • New cache-related function: keepCache

• implemented checkModuleLocal() to check for presence of module files in the module dir before attempting download from remote module repository

• improve module template to auto fill module author info using devtools.desc.author option if set.

• zipModule now has data argument, allowing data to be omitted from zipped module.

• improved moduleDiagram to show _INPUT_ node in different colour from the other modules

v1.3.1

Known issues: https://github.com/PredictiveEcology/SpaDES/issues

version 1.3.1

• Default module path (set via spades.modulePath option) is now set to a temporary location to avoid unintentionally writing to the user's home directory.
• add options for cache, input, output dirs (defaults to a temp dir until changed by user)
• rename spades.modulesPath and spades.modulesRepo options to spades.modulePath and spades.moduleRepo
• New cache-related functions: showCache and clearCache

v1.3.0

Known issues: https://github.com/PredictiveEcology/SpaDES/issues

version 1.3.0

• stricter package version dependencies in Imports and Suggests
• debug argument in spades() can now take any expression, character strings. TRUE/FALSE changed behavior to show only current event.
• timeunits: when there are parent and grandparent modules, if timeunit is defined, it overrides the "smallest unit" rule. Thus, a parent module can force a timeunit.
• Plot - enhancements and fixes:
  • more robust base plotting and many visual tests on Windows;
  • re-add col arg to Plot (mimicks cols). Was lost from version 1.1.2;
  • allow any arbitrary function to be used internally to Plot (e.g., barplot, plot, etc.);
  • add arr argument to Plot(), allowing passing of arrangement;
  • allow title arg in Plot() to accept character for plot title;
  • Plot can use character passed to title as a title;
  • some additional functionality for plotting of factor rasters, incl. clickValues, legends correct for wide variety of types;
  • change new arg in Plot(). Now it does one plot at a time, not whole device. Use clearPlot() to wipe whole device.
• Add modulesGraph, showing parent and child module relationships.
• Add filesOnly arg to shine(). This can be in preparation for publishing to www.shinyapps.io or other pages. Currently still alpha.
• Add POM: Pattern Oriented Modeling (#269). A simple interface to a simList object, allowing fitting parameters to data.
• add .inputObjects functionality -- function that runs during simInit() to create default inputObjects
• add P as a namespaced shortcut for params: P(sim) would replace params(sim)$moduleName when called from within that module
• allow params(sim) & start(sim) & others in defineModule() by changing parse order in module metadata
• add explicit cl arg to parallel aware functions, for more control
• newModule gains new arguments type = c("child", "parent") and children = c(). See ?newModule (#300).
• module structure now includes an R subfolder for R scripts. These will be parsed during simInit.
• checksums updated to use faster hashing algorithm (xxhash64) and now only requires a single hash value per file (#295)
• new Rstudio addin for 'newModule' (#298); requires Import of shiny (>= 0.13), miniUI (>= 0.1.1), and rstudioapi (>= 0.5)
• fixed bugs in module template
• fixed bug in cir (#288, #311)
• improved use of package options; added new option spades.modulesPath.
• improved downloadModule to use option spades.modulesPath (#309)
• improved specification of module inputs and outputs (#189, #214, PR#310)
• remove module version warning (#312)
• other tweaks and fixes

v1.2.0.9011

Known issues: https://github.com/PredictiveEcology/SpaDES/issues

version 1.2.0.9011

• Add filesOnly arg to shine(). This can be in preparation for publishing to www.shinyapps.io or other pages. Currently still alpha.
• Add POM: Pattern Oriented Modeling (#269). A simple interface to a simList object, allowing fitting parameters to data.
• add col arg to Plot (mimicks cols). Was lost from version 1.1.2.
• add .inputObjects functionality -- function that runs during simInit() to create default inputObjects
• add p as a namespaced shortcut for params: p(sim) would replace params(sim)$moduleName
• allow params(sim) & start(sim) & others in defineModule() by changing parse order in module metadata
• allow any arbitrary function to be used internally to Plot (e.g., barplot, plot, etc.)
• add arr argument to Plot(), allowing passing of arrangement
• allow title arg in Plot() to accept character for plot title
• change new arg in Plot(). Now it does one plot at a time, not whole device. Use clearPlot() to wipe whole device.
• Plot can use character passed to title as a title.
• add RandomFieldsUtils to Imports
• some additional functionality for factor rasters, incl. clickValues, legends correct for wide variety of types
• add explicit cl arg to parallel aware functions, for more control
• newModule gains new arguments type = c("child", "parent") and children = c(). See ?newModule (#300).
• checksums updated to use faster hashing algorithm (xxhash64) and now only requires a single hash value per file (#295)
• new Rstudio addin for 'newModule' (#298); requires Import of shiny (>= 0.13), miniUI (>= 0.1.1), and rstudioapi (>= 0.5)

v1.2.0

Known issues: https://github.com/PredictiveEcology/SpaDES/issues

• fix bug associated with forthcoming dplyr update
• remove gtools and secr from Imports
• spread enhancements: circular spreading, landscape-based functions, allow overlapping events
• enhancements to cir and bug fix (#290)
• new function rings whose argument names closer match to cir
• performance enhancements in spread, rings, cir
• add cache and progress args to spades and experiment functions
• Plot now clips symbols (generally points) to plotting area. This will allow future "wiping" of plot area.
• dev() returns dev.cur invisibly, allowing for finer control of plotting devices
• add distanceFromEachPoint, a multipoint version of raster::distanceFromPoints
• improve splitRaster (#276, #284)
• new function mergeRaster (#282, #283)
• changed and performance boost to randomPolygons (uses spread internally now)
• minor performance enhancements
• many more and improved unit tests

v1.1.4

Known issues: https://github.com/PredictiveEcology/SpaDES/issues

• fix OSX CRAN check errors caused by inconsistent use of normalizePath

v1.1.3

Known issues: https://github.com/PredictiveEcology/SpaDES/issues

• add gtools to Imports
• bug fix in experiment that did not allow parallel spawning on some systems and crashed with empty outputs argument
• fix minor bugs in sampleModules
• lots of documentation enhancements, esp. ?inputs and ?outputs
• add .plotInitialTime and .saveInitialTime arguments to spades() to easily allow turning off plotting and saving
• more robust inputs and outputs, including extension-based automated outputs, using data.frame instead of data.table allowing lists to be passed for "arguments"
• allow vectorized n in setColors (#70), and partial n if named
• many more unit tests (#139)
• other bug fixes

v1.1.2

Known issues: https://github.com/PredictiveEcology/SpaDES/issues

• fix issues associated with upcoming dplyr updates
• performance enhancements: much faster simulation execution
• Plot accepts color column in SpatialPointDataFrame objects
• add col arg to Plot (mimicks cols)
• add experiment function, a wrapper around spades for running multiple simulations (#265)
• add shine function, with simList signature (#261)
• add copy function, which does a deep copy of a simList object
• add RColorBrewer to Imports
• allow Plot colours to be set in the Plot call using cols argument
• allow Plot colours to be set in the Plot call using RColorBrewer palettes
• fix bug when plotting certain real-numbered rasters
• fix bug in Plot legends
• fix bug in assigning vectors as default module param values
• fix bug that prevented printing simulation times
• fix bug in event list sorting by eventPriority
• new slot in simList object: current, to store the current event
• new accessors current and current<- to get and set the current event
• defineParameter now coerces the default value to match the type class
• export objectNames() for external use
• outputPath updates output(sim)$file file paths, in addition to just paths(sim)$outputPath
• other bug fixes

v1.1.1

Known issues: https://github.com/PredictiveEcology/SpaDES/issues

• require archivist version 2.0 or greater
• improved moduleCoverage testing and template (PR257)
• correct legends from rasters so that is.factor(raster) is TRUE
• user defined time units can be used in module metadata "timeunit".
• add module timeunits to simList show method (#260)

v1.1.0

Known issues: https://github.com/PredictiveEcology/SpaDES/issues

• require R version 3.2.2 or higher
• remove downloader from Imports (#203)
• add covr to Suggests and lazyeval to Imports
• require DiagrammeR version 0.8.2 or higher which fixes mermaid/knitr/pandoc error (rich-iannone/DiagrammeR#139)
• add data source info module metadata (#205)
• new function downloadData to download module data (with #205)
• new function checksums to verify MD5 hashes of data files (with #205)
• warn instead of error when using e.g., newModule with Rstudio on Windows (#209)
• more control of caching behaviour: added cachePath to paths slot (#227)
• only download data if file doesn't exist or checksum mismatch (#229)
• allow multiple checksums per file in checksum (#230)
• fix error caused by coercing NA to arbitrary class for which no suitable NA type exists (#231)
• use warnings instead of errors for missing modules metadata (#233)
• add event priorities (#236)
• enhanced functionality of spread() (#237)
• add unit tests and coverage to newModule (PR242, PR245)
• objects passed to simInit can be named differently from their objects (#247)
• downloadModule and downloadData now also download children modules/data
• new function divergentColors to generate divergent colour palettes for legends
• improve efficiency of loadPackages
• change .spatialObjects class to spatialObjects and export
• add .Rdata and .rds files to automatic loading.
• warn user if SpaDES package version doesn't match module version
• allow more signatures in simInit (modules can be character)
• other bug fixes