[pandas] annotated tag v0.4.2 created (now 4be6155)

[Andreas Tille]

This is an automated email from the git hooks/post-receive script.

tille pushed a change to annotated tag v0.4.2
in repository pandas.

        at  4be6155   (tag)
   tagging  35c6b68facf00788e9092d1fcedd53acbf99be89 (commit)
  replaces  v0.4.1
 tagged by  Wes McKinney
        on  Mon Oct 3 00:46:47 2011 -0400

- Log -----------------------------------------------------------------
Version 0.4.2

Wes McKinney (50):
      DOC: update release notes
      RLS: update release notes
      ENH: playing around with the AST. not sure about using magic
      ENH: starting work on faster merge infrastructure
      ENH: reindex functions no longer return masks. related refactoring
      ENH: cython code generation for faster take functions, some integration
      ENH: work on faster merging function that does not produce intermediate copies
      ENH: got new merging function working. many more test cases required however. bumped dev version to 0.4.2
      ENH: major speed optimization in Index.intersection
      ENH: significantly speeding up Index.union
      ENH: change to buffered datetime check, is_monotonic function
      BUG: fix outer_join_indexer type
      MSC: sampling script update
      ENH: more templated Cython code coming, renamed module
      ENH: Int64Index class, Cython codegen, refactoring to enable transparent use
      STY: got rid of superfluous autogen docs
      ENH: join refactoring to enable subclasses (e.g. Int64) to implement faster join operations
      ENH: shizam! int64 axes with super fast join performance
      ENH: bug fixes and fixed up benchmark script
      ENH: speed optimization / tweaks in merging
      BUG: group_index being computed incorrectly in unstacker, only presents itself if nlevels > 3
      BUG: groupby cython bug in _aggregate_group, unit test and raise up exceptions
      ENH: major Cython code cleanup, left_join_indexer bugfix and speed opt, inner_join_indexer speed opt
      ENH: implemented BlockManager.take for axis >=1 and seriously sped up DataFrame.take
      DOC: note about requiring Cython for dev build, address GH #186
      ENH: groupby speed optimization for Int64Index
      BUG: address GH #185, prevent use of step in label-based slicing
      ENH: Int64Index.union optimization in the monotonic case
      ENH: Series.align method to leverage faster joins, faster binary ops
      ENH: Series.sort_index perf enhancement
      ENH: notnull/isnull perf enhancement regression from 0.3.0. address GH #187
      RLS: update release notes
      ENH: groupby speed enhancement due to less aggressive integrity checking in BlockManager. Implemented fast GroupBy.size function
      BUG: isnull works with float32, address GH #182
      BUG: implement remaining as_index=False handling in DataFrameGroupBy.aggregate. address GH #181
      ENH: fast DateRange.intersection in some cases and sped up DateRange.union in some cases. address GH #178
      ENH: add get_level_values method and tests, address GH #188
      ENH: implement shallow copy on BlockManager, address GH #125
      TST: test DataFrame.take for mixed-dtype
      TST: groupby test coverage
      TST: test coverage for take functions and added out keyword to take_1d
      BUG: fix left_join_indexer bug
      TST: test coverage for index.py, minor refactoring
      TST: finish index.py test coverage
      BUG: join_managers bug using wrong indexer
      TST: finish up pandas.core test coverage, test Series.align
      BUG: fixed incorrect check for integer dtype on 32-bit platforms
      REF: changed index types to Py_ssize_t
      TST: raise SkipTest for overflow in pre-epoch dates for now
      RLS: Version 0.4.2

-----------------------------------------------------------------------

No new revisions were added by this update.

-- 
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/debian-science/packages/pandas.git