NEWS

---

DTSg 2.1.0.9000

• Added capply() method exclusively to the R6 interface, which acts as an alias for the colapply() method. Unfortunately, it is not possible to also supply S3 methods for all of these R6 only aliases (capply(), names(), raggregate(), rbind(), set() and setnames()) due to incompatible S3 generics.
• Converted all TALFs to S3 generics and added respective S3 methods (*.POSIXct()). This does not change the package's API, but improves its extensibility.
• Converted the interpolateLinear() and rollback() functions to S3 generics and added respective S3 methods (interpolateLinear.numeric() and rollback.POSIXct()). This as well does not change the package's API, but improves its extensibility.
• Fixed a possible loss of precision in the interpolateLinear.numeric() method when applied to an integer column.
• Slightly improved the documentation.
• Minor internal code improvements.

DTSg 2.1.0 (2026-03-28)

• The day saving time shift for the ignoreDST argument of the aggregate(), colapply() and subset() methods is now attempted to be estimated and not invariably assumed to be one hour long.
• More time zones are now considered UTC or equivalent (execute grep("^(Etc/)?(UTC|UCT|Universal|Zulu)$|^(Etc/)?(GMT(\\+|-)?0?|Greenwich)$", OlsonNames(), ignore.case = TRUE, value = TRUE) for a full list).
• Updated the package's pkgdown website to Bootstrap 5.
• Slightly improved the documentation.
• Minor internal code improvements, which require R ≥ 4.0.0.

DTSg 2.0.0 (2025-01-20)

• Added "timechange" funbyApproach utilising timechange::time_floor() as the main function for transforming timestamps. This new approach generally is way faster than the existing ones for both families of TALFs and all time zones, hence it is the new package's default.
• Sped up the byY_____() and byYm____() TALFs of the "base" funbyApproach.
• The print() method does not print the values' key anymore, as it used to be data.table's default before v1.15.0.
• Fixed a bug in the aggregate() method, which could lead to incorrect results under the following conditions:
  • The time zone of the DTSg object is not UTC or equivalent.
  • The ignoreDST argument is TRUE.
  • A single column with missing values is aggregated.
• Related to the above fix are the following breaking changes when aggregating a single column with missing values:
  • A possible .n column now contains 0 instead of NA in case all values of a certain temporal aggregation level are missing.
  • The stripping of missing values within summary functions now depends on the value of a possible na.rm argument.
• Fixed the support for data.table's GForce optimisation when the fun argument of the aggregate() method is provided with a character vector specifying summary functions. Please note that the column order of the resulting DTSg object is now different due to this fix.
• Fixed that the setCols() method and its R6 only alias set() was too anxious about removing all value columns.
• It is no longer possible to use the deprecated byFasttime*() TALFs. Please use the "fasttime" or even better the new "timechange" funbyApproach from now on.
• Removed magrittr from the suggested packages list (vignettes now use R's native pipe operator).
• Slightly improved the documentation.
• Improved tests.
• Major internal code improvements.

DTSg 1.1.3 (2023-09-27)

• Ceased the use of legacy time zone symlinks in tests.
• Bumped the minimum tested R version from 4.0.2 to 4.2.3 using the corresponding Posit public package manager snapshot.
• Slightly improved the documentation.
• Minor internal code improvements.

DTSg 1.1.1 (2022-06-08)

• Column names can often now be additionally specified by a character string containing either comma separated column names, for example, "x,y,z", or the start and end column separated by a colon, for example, "x:z".
• Fixed a bug in the interpolateLinear() function causing partial last observation carried forward behaviour when its roll argument was specified smaller than the size of the gap to be interpolated.
• Fixed a bug in the interpolateLinear() function causing partial interpolation in certain cases when its roll argument was specified smaller than the size of the gap to be interpolated.
• Slightly improved the documentation.

DTSg 1.0.0 (2022-05-02)

• Added funbyApproach argument to the new(), aggregate(), colapply() and subset() methods: allows for specifying the flavour of the applied temporal aggregation level functions (either "base" utilising as.POSIXct() or "fasttime" utilising fasttime::fastPOSIXct() or "RcppCCTZ" utilising RcppCCTZ::parseDatetime() as the main function for transforming timestamps). Custom approaches for user defined TALFs are also possible. Please note that the byFasttime* versions of the TALFs are now deprecated. Use this argument from now on instead.
• Added funbyApproach also to the list of helper data (funbyHelpers argument) passed on to TALFs.
• Added funbyApproach field reflecting the individual funbyApproach of a DTSg object (can also be actively set in order to change the utilised approach).
• Added undocumented names() method exclusively to the R6 interface, which acts as an alias for the cols() method.
• Added mode and typeof arguments to the cols() method: allows for getting column names with a certain mode() and/or typeof(). This can be especially handy when making use of the units package.
• Added DTSgFast, DTSgFunbyApproach and DTSgNA.status options providing default values for the fast, funbyApproach and na.status arguments of the new() method.
• Added DTSgDeprecatedWarnings option: allows for disabling warnings from deprecated features.
• Greatly sped up the by______() and byFasttime______() TALFs.
• It is no longer possible to use the deprecated value "all" with the class argument of the cols() method in order to get all column names. Use the default value NULL instead for this. "all" is treated as a filter for classes of type all from now on.
• Added an example to the setCols() method showing how to set measurement units with the help of the units package.
• Added an example to the documentation of the colapply() method showing how to calculate running correlations with the help of the runner package.
• Bumped the minimum tested R version from 3.5.2 to 4.0.2 using the corresponding MRAN repository snapshot.
• Slightly improved the documentation.
• Minor internal code improvements.

DTSg 0.8.1 (2021-10-27)

• Added a “special class” called ".numerary" available to the class argument of the cols() method: allows for querying the names of integer and numeric columns in one go.
• Added undocumented raggregate(), rbind(), set() and setnames() methods exclusively to the R6 interface, which act as aliases for the rowaggregate(), rowbind(), setCols() and setColNames() methods.
• The print() method now truncates the number of printed rows of the values more aggressively.
• Created a pkgdown website.
• Improved the vignettes and documentation.

DTSg 0.7.1 (2021-05-30)

• Fixed error "Error in prettyNum(.Internal(format(x, trim, digits, nsmall, width, 3L, : invalid value 0 for 'digits' argument" related to tests in upcoming R 4.2.0.
• Slightly improved the documentation.
• Minor internal code improvements.

DTSg 0.7.0 (2020-09-05)

• Added rowaggregate() method: allows for applying summary functions row-wise to a DTSg object.
• Added rowbind() method: allows for combining the rows of a DTSg object.
• Added setColNames() method: allows for renaming columns of a DTSg object.
• Added multiplier and funbyHelpers arguments to the aggregate(), colapply() and subset() methods: allows for adjusting the temporal aggregation level of TALFs and for passing on user defined helper data to TALFs.
• Added multiplier also to the list of helper data passed on to TALFs.
• Added helpers argument to the colapply() and rollapply() methods: controls if helper data is passed on to an applied function (makes occasionally needed anonymous function wrappers obsolete, e.g. x$colapply(fun = function(x, ...) {cumsum(x)}, funby = byYm____) can now be written as x$colapply(fun = cumsum, helpers = FALSE, funby = byYm____)).
• TALFs based on the fasttime package now also support time zones equivalent to UTC (execute grep("^(Etc/)?(UCT|UTC)$|^(Etc/)?GMT(\\+|-)?0?$", OlsonNames(), ignore.case = TRUE, value = TRUE) for a full list of supported time zones).
• Fixed that a DTSg object with only one timestamp did not set the name of its .dateTime column as expected.
• Improved the vignettes and documentation.
• Minor internal code improvements.

DTSg 0.6.0 (2020-07-12)

• Added subset() method: allows for filtering rows and/or selecting columns of a DTSg object.
• Added setCols() method: allows for setting the values of columns of, adding columns to and/or removing columns from a DTSg object.
• Added [ extract operator: acts as a shortcut for the getCol() method.
• Added examples to the documentation of the colapply() method showing how to calculate moving averages with the help of the runner package instead of the rollapply() method.
• The aggregate() method can benefit from data.table's GForce optimisation now when its fun argument is provided with a character vector specifying summary functions.
• Greatly sped up the nas() method.
• Temporal aggregation level functions supplied to the funby argument of the colapply() method are not forced to return a POSIXct timestamp any longer. They are, however, forced to return an atomic mode (the same goes for the subset() method).
• The getCol() method now is capable of also querying the .dateTime column.
• The R6Method argument of the S3WrapperGenerator() function now also takes a public method of an R6ClassGenerator as a function and not only as an expression.
• Fixed that not all missing values were made explicit after a call to the merge() method despite an "explicit" na.status in some cases.
• Fixed that the getCol() method tried to query all numeric columns instead of only the first one by default.
• Improved the vignettes and documentation.
• Minor internal code improvements.

DTSg 0.5.0 (2020-06-09)

• Added getCol() method: allows for querying the values of a single column of a DTSg object.
• Added funby and ignoreDST arguments to the colapply() method: allows for applying functions like cumsum() to a certain temporal level.
• Added na.status argument to the new() and alter() methods: allows for making missing values either "explicit" (default) or "implicit" or leaving them alone via "undecided".
• Added na.status field reflecting the status of missing values (can also be actively set).
• Added na.status also to the list of helper data passed on to temporal aggregation level functions.
• The funby argument of the aggregate() method now also accepts a named list of functions: allows for calculating several summary statistics at once.
• The periodicity field can now be actively set in order to change the periodicity of the time series.
• The timezone field can now be actively set in order to convert the time zone of the series.
• Fixed that a DTSg object with only one timestamp accepted a missing value in its .dateTime column.
• Improved the vignettes and documentation.
• Minor internal code improvements.

DTSg 0.4.1 (2020-05-08)

• Fixed error "Error in as.POSIXct.numeric(e) : 'origin' must be supplied" related to the rollback() function in upcoming R 4.1.0.
• Slightly improved the vignettes.
• Minor internal code improvements.

DTSg 0.4.0 (2020-03-28)

• Added memoryOverCPU argument to the rollapply() method: allows for preferring CPU over memory usage which makes the method more flexible in terms of resource consumption.
• Settings of the option DTSgClone in e.g. .RProfile are now respected and not overwritten by TRUE when the package is loaded.
• Fixed useless coercion of POSIXct .dateTime columns to POSIXct upon object creation in case they contained at least one missing value.
• Switched to the unit testing framework of the tinytest instead of the testthat package.
• Slightly improved the vignettes and documentation.
• Minor internal code improvements.

DTSg 0.3.0 (2020-01-08)

• Added resultCols and suffix arguments to the colapply() and rollapply() methods: allows for adding return values of applied functions as new columns instead of replacing existing ones.
• class argument of the cols() method now accepts a character vector of class names.
• Changed the default value of the class argument of the cols() method from "all" to NULL, however, for backward compatibility "all" can still be used for the same result, but will eventually be treated as a filter for classes of type all.
• Added pattern and ... arguments to the cols() method: allows for searching column names.
• Switched to the argument check functions of the checkmate package instead of the assertive.base, assertive.numbers, assertive.sets and assertive.types packages.
• Slightly improved the vignettes and documentation.
• Minor internal code improvements.

DTSg 0.2.1 (2019-10-20)

• Fixed error "Unsupported type passed to argument 'data'." related to the dygraphs package in at least some releases of R ≤\ 3.5.3.

DTSg 0.2.0 (2019-10-14)

• Added swallow argument to the new() method: allows for a more resource efficient object creation.
• Added drop argument to the values() method: allows for a resource efficient destruction of a DTSg object while preserving its values.
• Added class argument to the values() method: can be used to return the values of a DTSg object as a data.frame instead of a data.table.
• The alter() method and linked with it the new() and other methods are now way more resource efficient in some cases.
• Added timestamps field providing the total number of timestamps.
• The print() method now omits empty metadata fields.
• Added means to measure code coverage with the help of the covr package.
• Removed xts from the suggested packages list (already comes along with the dygraphs package).
• Slightly improved the vignettes and documentation.
• Minor internal code improvements.

DTSg 0.1.3 (2019-08-29)

• Slightly improved the vignettes and documentation.

DTSg 0.1.2 (2019-03-13)

• First release.