Release note for v0.2.3

[heronshoes]

RedAmber v0.2.3 has released:tada:

This release will support Arrow 10.0.0 especially related to to_arrow and to_arrow_array responsibility such as initialization with Numo::Narray. I introduced benchmarks for key features. I expect v0.2.3 to be the slowest version and hope RedAmber will be much faster after this. I also introduced test coverage measurement by simplecov. Coverage is 99.1% and about 30 lines are not covered now. I will pull up the coverage to 100%! Your help is welcome!

[0.2.3] - 2022-11-16

• Bug fixes

  • Fix DataFrame#to_s when DataFrame.size == 0 (Introduce the features to combine DataFrames #125)
  • Remove unused lines in funcs (Add test coverage measurement #128)
  • Remove unused methods in helper (Add test coverage measurement #128)
  • Add test for invalid arg in DataFrame.new (Add test coverage measurement #128)
  • Add test for Vector#shift(0) (Add test coverage measurement #128)
  • Fix bugs for DataFrame#[], #pick and #drop with Range of Symbols and Symbol (Fix #[], #pick and #drop by Range and Symbol #135)

• New features and improvements

  • Upgrade dependency to Arrow 10.0.0 (Upgrade to Arrow 10.0.0 #132)

    It is possible to initialize by the objects responsible to to_arrow since 0.2.3 .
    Arrays in Numo::NArray is responsible to to_arrow with Red Arrow Numo::NArray 0.0.6 .
    This feature is proposed by the Red Data Tools member @kojix2 and implemented by @kou.
    I made also Vector to be responsible to to_arrow and to_arrow_array.
    It becomes a member of ducks ('quack quack'). Thanks!

    • Change dev dependency to red-dataset-arrow (Bump version up to 0.2.3-HEAD #117)
    • Add dev dependency for red-arrow-numo-narray (Upgrade to Arrow 10.0.0 #132)
    • Support Numo::NArray in Vector.new (Upgrade to Arrow 10.0.0 #132)
    • Support Vector#to_arrow_array (Upgrade to Arrow 10.0.0 #132)

  • Update group (Update group operations #118)

    • Introduce new DataFrame group support (experimental)

      This additional API will treat a grouped DataFrame as a list of DataFrames.
      I think this API has pros such as:

      • API is easy to understand and flexible.
      • It has good compatibility with Ruby's primitive Enumerables.
      • We can only use non hash-ed aggregation functions.
      • Do not need grouped DataFrame state, nor #ungroup method.
      • May be useful for concurrent operations.

      This feature is implemented by Ruby, so it is pretty slow and experimental.
      Use original Group API for practical purpose.

    • include Enumerable to Group (experimental)

    • Add Group#each, #inspect

    • Refactor Group to align with Arrow

  • Introduce DataFrame combining methods (Introduce the features to combine DataFrames #125)

    • Introduce DataFrame#concatenate method
    • Add DataFrame#merge method
    • Add DataFrame#inner_join method
    • Add DataFrame#full_join method
    • Add DataFrame#left_join method
    • Add DataFrame#right_join method
    • Add DataFrame#semi_join method
    • Add DataFrame#anti_join method
    • Add DataFrame#intersect method
    • Add DataFrame#union method
    • Add DataFrame#setdiff method
      • Rename #setdiff to #difference
    • Support natural join in DataFrame#join
    • Support partial join_key and renaming
    • Fix DataFrame#join to merge key columns
    • Add DataFrame#set_operable? method
    • Add join/set/bind image to DataFrame.md
    • Fix DataFrame#join, #right_semi, #right_anti (Add test coverage measurement #128)

  • Miscellaneous

    • Return Vector in DataFrame#indices (Update group operations #118)

• Improve tests/ci

  • Improve CI

    • Add CI test on macOS (Update CI test #133)

    • Enable bundler-cache on macOS (Add test coverage measurement #128)

    • Add install gobject introspection prior to glib in CI (Update CI test #133)
      This will stabilize CI system installation especially with cache.

    • Rename workflows/test.yml to ci.yml (Update CI test #133)

      • Fix link in CI badge of README.md (Update group operations #118)

    • Add github action for coverage (Add test coverage measurement #128)

  • Add benchmark

    • Add benchmarks with Rover (Update group operations #118)
    • Introduce benchmark suite (Add benchmark #134)
    • Add benchmark for combining operations (Add benchmark #134)

  • Measuring test coverage

    • Add test coverage measurement (Add test coverage measurement #128)

• Refactoring

  • Remove redundant string escape in test_vector_function (Upgrade to Arrow 10.0.0 #132)
  • Refine tests to use assert_equal_array (Add test coverage measurement #128)
  • Rewrite Vector#replace (Add test coverage measurement #128)

• Documentation

  • Update README.md for installation (Update documents #126)
  • Add clause that keys must be unique in doc. (Update documents #126)
  • Rows should be called as 'records' (Update documents #126)
  • Update Jupyter Notebook 83 examples of RedAmber (Fix #[], #pick and #drop by Range and Symbol #135)

• GitHub site

  • Update Jupyter notebooks in Binder
  • Change default branch name from 'master' to 'main' (Changed default branch from master to main #127)

• Thanks

  Ruby Association Grant committee
  It is a great honor for selecting RedAmber as a project of Ruby Association Grant 2022.