For latest release notes please visit the Releases page at GitHub.
Bleeding-edge development is reflected in the ChangeLog and in the list of commits:
The raw ProjectTemplate source code is always the place to go for ground truth.
The list of changes below covers all releases up to and including v0.8.2. It is not necessarily exhaustive.
- New for R 3.6, support for staged install (#291)
- Define default table types using config var
- Support for
feather files (#239)
- Migration from
rstudio.project files (#201)
- Link website documentation current release to News.md
- Added file types to
- Internal functions are now documenter (#243)
data_ignore applies to
cache vars (#290)
- More informative error messages if not in a ProjectTemplate Directory (#287)
migrate_project() creates missing
cache directory (#275)
- Uses the Tidyverse packages (#207)
- Added custom templates (#206)
- Added support for rds files (#224, #227)
- Documentation moved to
docs folder. Will automatically update on website
- Added tibble support (#218)
- Moved from
xls package to
readxl package (#159, #219)
- Allow overridng of globals (#222)
- Fixed failing test for
RSQLlite package (#228)
- Faster caching (#152, #155, #160, #166)
- New clear() function to manage global environment (#167, #172, #173, #183)
- New project.config() function and configuration help page (#164)
- Renamed doc to docs and added references to GitHub Pages configuration (#180)
- Added config option to treat the first row of a data file as header (current default) or not. (#168)
- Change csv2.reader() behavior to act as expected when running read.csv2() (#195)
- Add list.data and incorporate in load.project (#187)
- Moved logger initialisation to before loading helper functions, See issue #150
- Cleaner exit when running load.project() in the wrong directory (#157)
- Allow to call RSQLite’s initExtension() (#161)
- Refactor migrate.project() and fix version tests (#162)
- Clean exit for more functions when current directory is not valid Project template (#163)
- Enable dcf files to contain R code that can be evaluated upon read (#169)
- Changed execution environment of CODE chunks (#171)
- Remove unused function .get.template.tar.path in create.project and old commented out code in load.project (#181)
- Uniform reading/writing global.dcf (#184)
- Ignore files in data directory (#178)
- Remove warnings from Travis and devtools::test() checks (#188)
- De-deprecated devtools installation command (#193)
- Bug Fix: Missing hash file causes failure in subsequent cache() operation (#194)
README.md files in each subdirectory (#128). The README file in the main directory is now also formatted as Markdown, with the project name as title.
- Code that creates and runs tests now also allows non-R files in
- Libraries from the
libraries section are loaded before sourcing
lib/ files (#130).
- Missing packages are installed automatically (#146, @wilmeragsgh).
dplyr to the list of default libraries in
global.dcf (#143, @pavopax).
cache() gains a new
... argument which is passed to
save() (#148, @eribul).
inherits = FALSE in
get() calls within specific environments (#139, @famuvie)
port option is now used for
- Update to getting-started/mastering documentation (#142, @matt-jay).
- Fix typo in error msg (#141, @famuvie).
- Isolate tests by using temporary directories and properly undoing
setwd() calls (#132).
R CMD check shows no errors, warnings, or notes.
- Includes all modifications from v0.5-3 and v0.5-2.
- Packages required for loading data sources are attached to the search path (with a warning) only if the new compatibility setting
attach_internal_libraries is set to
TRUE. Attaching packages to the search path seems to be unnecessary to achieve proper functionality, but users might rely on this behavior, and so this is the default for migrated projects but turned off for new projects (#104).
- Fixed error message when
require.package is called from an anonymous
function and fails to load a package. (Using
deparse(nlines = 1) now.)
- Added functon
.add.extension(). This allows users to create custom readers for extensions, either locally in a project or as packages.
- The configuration now stores the version of ProjectTemplate in the
- New function
migrate.project() that allows upgrading a project to the
current version of ProjectTemplate (#90, #121).
- New parameter
allows substitution of individual configuration items (#76).
- Can use mustache style templating in SQL calls to access project data structures (#50).
- Support passwordless connection to postgresql databases (#115).
- Configuration entries that start with a hash (
#) are silently ignored (#74).
- New variables
new.config that store the default
configuration used for missing configuration items or for a new project (#76, #89).
- Missing but required directories are now created (with a warning) (#76).
- Missing entries in the configuration file, or a missing configuration file,
are substituted by defaults (with a warning). Extra entries are ignored
(with a warning) (#76).
- All logical configuration options are now stored as Boolean values instead of
'off'. Input values other than
'off' are converted using
as.logical, invalid values result in an error (#76).
- Dropped dependency on
Defaults package (#100).
- Suppress warnings in tests (#111).
- Fix CRAN check issues (#117).
- Improved presentation of available readers in documentation (#119).
- Restore compatibility to R 2.15.3 by avoiding use of the
list.files() in our
- New function
get.project() to access
- Attempting to load a missing package when reading data will lead to a
user-friendly error message (#26).
- Export existing
translate.dcf function, useful for implementing custom readers (#59).
- Add new function
.add.extension. This allows users to create custom readers
for extensions, either locally in a project or as packages (#59).
- Fix CRAN warnings concerning use of
- Fix CRAN warnings concerning assignments to
- Updated author information in DESCRIPTION file (#40).
- Store templates for empty projects in tar files instead of storing
the entire directory structure per CRAN request and to avoid having to build
--keep-empty-dirs switch (#41).
- New maintainers: Kirill Müller firstname.lastname@example.org, Kenton White email@example.com
- Allow string interpolation of R functions in sql queries
- Added a JDBC database wrapper for accessing PostgreSQL databases hosted by Heroku (#20)
- JDBC connection can now use path to jar stored in CLASSPATH
- project.info is stored in the global environment again (reverted change from
0.4-4); the active binding seemed to work for the tests but not from outside
- create.project now works if the directory exists. Merging with a non-empty
directory is supported by setting the new parameter merge.strategy.
- Fix CRAN checks (@krlmlr).
- project.info is now an active binding to avoid writing to the global
- Reenabled xlsx.reader (@krlmlr).
- Added JDBC support to sql.reader (@joshbode, #12).
- Various MySQL improvements (@cortex, #10).
- Fix “Argument port must be an integer value” when using port number for mysql
driver (@cortex, #7).
- Tentative CSV2 support.
- Added optional recursive data directory loading using a ‘recursive_loading’ setting.
- Added basic Oracle support (with tnsnames, no host/port) (@matteoredaelli, #6).
- Had to disable xlsx.reader temporarily because it does not build for 2.15.
- Added ‘eda.R’ example file in ‘src/’ directory.
- Allow loading from ‘cache/’ without loading from ‘data/’ through ‘cache_loading’ setting.
- Revised documentation.
- Added Postgres support through RPostgreSQL
- Very rough ODBC support. See example in
.dat files are now read as if they were
- Revised documentation.
- Added a
as_factors configuration option suppresses automatic character-to-factor conversion.
.zip support using temporary files.
.mp3 support through the tuneR package.
.ppm support through the pixmap package.
data_tables configuration option automatically translates data sets into data.tables.
- Added unit tests for SPSS, Stata and SAS file formats.
- Improved SQLite3 support:
- Load all tables from SQLite3 database using a
.sql file: set
table = "*"
- Load all tables from SQLite3 database automatically using
- Added a .file type that can load files outside of the project directory.
- Added a
stub.tests() function that autogenerates tests for helper functions.
- Added a
show.project() function that gives all known information about a project’s internal state.
- Added a
- Improved error handling.
- Fixed a bug in
- Added support for several new data formats.
- Changed the configuration system from YAML to DCF format.
- Moved the data loading helper functions into separate files. They are no longer nested inside of
- Started controlling which functions are exported using a NAMESPACE file.
- Added the ability to create minimal projects as well as the full projects that existing users will be familiar with.
- Switched all the
print() calls to
- Fixed a bug in cache().
- Version 0.2-1 adds the following directories:
- Inside the
config directory, a global configuration file called
config/global.yaml has been added that uses YAML syntax. This file provides the following configuration options:
data_loading: This can be set to ‘on’ or ‘off’. If
data_loading is on, the system will load data from both the
data directories with
cache taking precedence in case of name conflict.
munging: This can be set to ‘on’ or ‘off’. If
munging is on, the system will execute the files in the
munge directory sequentially. If
munging is off, none of the files in the
munge directory will be executed.
logging: This can be set to ‘on’ or ‘off’. If
logging is on, a logger object using the
log4r package is automatically created when you run
load.project(). This logger will write to the
logs directory. By default,
logging is off.
load_libraries: This can be set to ‘on’ or ‘off’. If
load_libraries is on, the system will load all of the R packages listed in the
libraries field below.
libraries: This is a YAML array listing all of the packages that the user wants to automatically load when
load.project() is called.