Skip to contents

Changelog

Source: NEWS.md

squarebrackets 0.0.0.9011

  • Overhauled the _mod methods: they now use R’ native semantics for atomic objects, and shallow pass-by-value semantics for recursive objects.
  • Removed the _icom methods, as they’re no longer necessary.
  • Added quick way to filter rows by condition in a data.frame using double-tilded formulas.
  • Added more tests.

squarebrackets 0.0.0.9010

  • Giving non-existing names as indices now gives a proper error.
  • Added the dt_ functions.
  • Added the ci_df() function.
  • Using use = -2:-1 or use = 2:1 in the tt_ will now result in an error, to ensure consistent syntax.

squarebrackets 0.0.0.9009

  • Added arepl().
  • Replaced the use of do.call(...) in the ss_ methods with eval(quote(...)).
  • Replaced the strid_pv() stride function with the more generic stride_v() S3 method; Users can now create custom methods for it.
  • Moved the use argument from the long_ methods into the stride_ functions/methods.
  • Streamlined the internal code here and there.
  • Fixed some documentation errors.
  • Bug Fix: data.table gave an error when sub-setting it if one of its columns was a mutatomic object. This is because data.table does not directly call [.mutatomic. This issue is now fixed, by re-defining the mutatomic internal attributes. The mutatomic class is now compatible with data.table and other functions that internally do not call [ directly.

squarebrackets 0.0.0.9008

  • Completely overhauled the methods for index-less sub-set operations on long vectors; They are now the long_ methods, and use stride objects to perform index-less sub-set operations.
  • Replaced the sbt_ methods with the tt_ methods (same methods, different names)

squarebrackets 0.0.0.9007

  • Completely overhauled the arguments regarding inverting indices in almost all methods.
  • Removed imaginary numbers as a form of indices in favour of the much more flexible formula (~...) based indices.
  • Simplified the input for the slicev_ methods.
  • Restructured the unit tests for the slice_ and slicev_ methods.
  • Replaced idx method with the _icom methods.
  • Removed idx_r.
  • Replaced the obs, vars arguments with the row, col arguments in the sbt_ methods;
  • Fixed a few relatively small bugs

squarebrackets 0.0.0.9006

  • Overhauled the main methods: replaced {ii_, ii2_, ss_, ss2_} methods with { ii_, ss_, sbt_ }
  • Removed the dt_* functions.
  • Removed the idx_ord_* functions.
  • Renamed sub2ind and friends to ss2ii and so on.

squarebrackets 0.0.0.9005

  • 0L can now be used as an alias for NULL in the indexing arguments.
  • NULL or 0L can now be used in a list for the ss argument, for more convenience.
  • Added proper tests for the obs, vars arguments.
  • Improved speed of the array-related methods.

squarebrackets 0.0.0.9004

  • Improved speed for the _set methods.
  • Simplified the internal code used for matrices.
  • Bug fix in the ss, d argument pair, where something like 1:2, 1:2 behave differently then list(1:2), 1:2 for data.frames.

squarebrackets 0.0.0.9003

  • Improved speed for the slice_ and slicev_ methods.

squarebrackets 0.0.0.9002

  • Renamed the i_ and i2_ methods to ii_ and ii2_, to avoid confusion.
  • Renamed ci_flat() to ci_ii() and ci_sub() to ci_ss().

squarebrackets 0.0.0.9001

  • Placed the i2_rec() and i2_recin() in its own, separate set of methods: the lst_ methods.
  • Adjusted the documentation and tests in accordance with the above change.
  • The ss_/ss2_ methods will now give an error if dimensional sub-set operations are attempted on non-dimensional objects.

squarebrackets 0.0.0.9000

  • First “real” experimental version of ‘squarebrackets’ on GitHub.