Version 0.4.2

• Remove call to deprecated function.

Version 0.4.1

• Avoid potential deep copies in left_join with data.table as input (#22).
• Fix HTML documentation.

Version 0.4.0

• The logic of mutate with .sequential = TRUE has been changed to support scenarios where
  group_by or key_by were used as well.
• transmute gained the same .sequential argument with the same underlying logic.
• summarize and aliases gained parameter .assume_optimized in case table.express doesn't know
  about a function that data.table can optimize.
• summarize and aliases can support the sequential logic mentioned before if the summarizing
  functions are not assumed to be optimized.

Version 0.3.3

• More test adjustments to fix problems on CRAN for real.

Version 0.3.2

• Adjusted tests to avoid problems on CRAN.

Version 0.3.1

• Improved compatibility with packages that use dplyr without importing data.table (#21).
• Added summarize/summarise verbs.
• right_join gained parameter which.

Version 0.3.0

• Verbs can now be used without start_expr/chain/end_expr with some considerations, check the
  updated vignettes (#3). Importantly, this makes table.express conflict with dtplyr.
• Fixed a bug that prevented automatically built chains to work after a left_join.
• The *_sd verbs now support lambdas as formulas (#19).
• Added max_by and min_by verbs.
• mutate_sd gained parameter .pairwise.
• distinct gained parameters .keep and .n.
• transmute gained parameter .enlist (#20).
• filter_on now allows empty names for data tables that already have keys (#17).
• All filtering verbs now have a which parameter (#17).
• Added nest_expr for certain scenarios that need the captured data.table; check the vignette
  (#16).
• Simplified semi_join.

Version 0.2.0

• Added joining verbs and a corresponding vignette describing them (#1).
• Added the distinct verb.
• transmute is no longer an alias for select. This is to make the latter more flexible in some
  bare selection cases (like combining several tidyselect calls), and leave the former more simple
  and preferable for actual transmutation cases (#4).
• Both mutate_sd and transmute_sd now support a list of functions as input, but with slightly
  different semantics; check the vignette (#8 and #11).
• Now both mutate_sd and filter_sd support the .COL pronoun for .SDcols predicates, but they
  remain eager in this regard.
• select and filter_on gained a .negate parameter (#9 and #6).
• mutate gained a .sequential parameter to enable usage of newly created columns.
• Added some debugging statements when option table.express.verbose is TRUE, or
  .verbose = TRUE in start_expr/chain.
• Updates to main vignette.
• Semi-breaking change: *_sd verbs now have .SDcols before their ellipsis. This should only
  require code adjustment for transmute_sd when .SDcols was left missing.

Version 0.1.1

• Let transmute_sd build expressions with lapply when .how is a function and .SDcols is
  character or numeric.
• Support numeric input in select.
• Removed tracemem example from vignette to avoid CRAN problems.

Version 0.1.0

• Initial version with single-table verbs.