Address Prof. Brian Ripley’s comment: “Packages in Suggests should be used conditionally: see ‘Writing R Extensions’. This needs to be corrected even if the missing package(s) become available. It can be tested by checking with R_CHECK_DEPENDS_ONLY=true.”
Add functions add_recipe() and reproduce()
check_worcs_installation() no longer throws error for rticles and prereg
Add more defensive programming to save_data() and load_data() to prevent errors.
Further improvement to how checksums are computed; instead of checking how Git treats the line endings (which breaks when a project is no longer version controlled, e.g., when downloading a ZIP from GitHub), we check if files are binary.
Add github action to reproduce a repo via github actions: github_action_reproduce()
Update Reproduce.Rmd vignette; the advice to download a ZIP from GitHub is bad practice, because this ZIP is no longer version controlled with Git.
Checksums are computed differently from preceding versions, to account for the fact that Git changes end of line characters which breaks consistency of checksums across operating systems.
Add endpoint functionality: add_endpoint(), snapshot_endpoints(), and check_endpoints()
Add github action for endpoints: github_action_check_endpoints()
Improve handling of relative filenames in save_data()
fix bug related to file path of codebook and metadata
add function report()
Fix breaking bug in check_worcs_installation() introduced in version 0.1.9
Update setup vignette to reflect new GitHub authentication protocols
closed_data() and open_data() now store factor and ordered value labels in a YAML file
Added functions data_unlabel() and data_label(), which respectively coerce factors to integer, and restore integers to factors using metadata stored in a YAML file
load_data() attempts to coerce variable classes using the codebook, and restores ordered levels from the value_labels YAML file
Update templates from rticles and prereg packages
Add check_worcs_installation() function to determine whether all non-R-package dependencies are installed correctly
Update references to WORCS paper
Replace GitHub version of papaja with CRAN version
Update rticles and prereg templates
Handle literal function names from rticles and prereg, such that users can use all available options from those packages without worcs explicitly referencing those functions
Add arguments ‘save_expression’ and ‘load_expression’ to closed_data() and open_data(). This allows users to save data in different formats. The ‘load_expression’ is stored in the .worcs file, and referenced by load_data().
Declared pandoc in SystemRequirements
Made all examples using pandoc conditional on availability of pandoc
Functions that wrap rmarkdown::render() now return NULL when pandoc is not available
Fixed failing tests as requested by dr. Brian Ripley
Workflow vignette updated to address reviewer comments from “Data Science”
New functions: add_synthetic() and notify_synthetic()
closed_data() no longer wraps errors from the call to synthetic(). Thus, the function call fails with an error if data synthesis fails.
closed_data() now passes formal arguments of synthetic() contained in ‘…’ on to synthetic().
synthetic() now allows ‘y’ to be referenced in ‘predict_expression’ when ‘model_expression’ is set to NULL, thereby enabling bootstrapping.
The internal function worcs:::col_message() now renders status messages in black print instead of colored print when rmarkdown::render() is on the callstack. This avoids errors when knitting to PDF. See https://github.com/rstudio/rmarkdown/issues/1951 for the error circumvented by this fix.
Necessary update due to breaking changes in dependency from the gert package
Add function add_preregistration() to add a preregistration at a later stage
Add preregistration templates “PSS” and “Secondary”
Fixed bugs in handling of filenames by open_data() and closed_data() codebook.Rmd from the name of the object supplied to the ‘data’ argument
By default, open_data() and closed_data() now construct the filenames of the .csv and codebook.Rmd from the name of the object supplied to the ‘data’ argument. This also means that the object will be loaded with the same name by load_data().
Add function add_manuscript() to add a manuscript at a later stage
Major bugfixes in worcs_badge(); this function did not work as intended. Fixed now
descriptives() now returns mean() instead of sd()
descriptives() now returns both n and missing
README.md did now correctly shows .Rproj file
rticles now declared as import
All documentation updated to ensure compatibility with the submitted version of the WORCS-paper.
Authors and references to WORCS-paper updated to include all co-authors.
Prevent git_user() from engaging in antisocial behavior, resetting users’ ‘Git’ credentials, by adding a default argument ‘overwrite = !has_git_user()’. By default, the function will do nothing when ‘Git’ user credentials already exist.
Bug fix to load_data() to ensure compatibility with the ‘RMarkdown’ behavior of running code chunks from the .Rmd file directory, instead of from the actual working directory.
Bug fix to cite_all() and cite_essential() to allow escaping literal double @.
Bug fix to export_project(), which now respects the folder structure within a project, and also supports relative paths to ensure portability, even when called from a different working than the working directory. Fixing this bug required changing the working directory before calling zip(). This change in working directory is immediately followed by a call to on.exit(setwd(oldwd)), in line with CRAN policy.
.worcs file is used to determine the existence of a worcs-project recursively.