HISTORY.rst

Changelog

• master

• v0.2.28 (2015-07-03)

  • implement RDD.sortBy() and RDD.sortByKey()
  • additional unit tests

• v0.2.24 (2015-06-16)

  • replace dill with cloudpickle in docs and test
  • add tests with pypy and pypy3

• v0.2.23 (2015-06-15)

  • added RDD.randomSplit()
  • saveAsTextFile() saves single file if there is only one partition (and does not break it out into partitions)

• v0.2.22 (2015-06-12)

  • added Context.wholeTextFiles()
  • improved RDD.first() and RDD.take(n)
  • added fileio.TextFile

• v0.2.21 (2015-06-07)

  • added doc strings and created Sphinx documentation
  • implemented allowLocal in Context.runJob()

• v0.2.19 (2015-06-04)

  • new IPython demo notebook at docs/demo.ipynb at https://github.com/svenkreiss/pysparkling/blob/master/docs/demo.ipynb
  • parallelize() can take an iterator (used in zip() now for lazy loading)

• v0.2.16 (2015-05-31)

  • add values(), union(), zip(), zipWithUniqueId(), toLocalIterator()
  • improve aggregate() and fold()
  • add stats(), sampleStdev(), sampleVariance(), stdev(), variance()
  • make cache() and persist() do something useful
  • better partitioning in parallelize()
  • logo
  • fix foreach()

• v0.2.10 (2015-05-27)

  • fix fileio.codec import
  • support http://

• v0.2.8 (2015-05-26)

  • parallelized text file reading (and made it lazy)
  • parallelized take() and takeSample() that only computes required data partitions
  • add example: access Human Microbiome Project

• v0.2.6 (2015-05-21)

  • factor out fileio.fs and fileio.codec modules
  • merge WholeFile into File
  • improved handling of compressed files (backwards incompatible)
  • fileio interface changed to dump() and load() methods. Added make_public() for S3.
  • factor file related operations into fileio submodule

• v0.2.2 (2015-05-18)

  • compressions: .gz, .bz2

• v0.2.0 (2015-05-17)

  • proper handling of partitions
  • custom serializers, deserializers (for functions and data separately)
  • more tests for parallelization options
  • execution of distributed jobs is such that a chain of map() operations gets executed on workers without sending intermediate results back to the master
  • a few more methods for RDDs implemented

• v0.1.1 (2015-05-12)

  • implemented a few more RDD methods
  • changed handling of context in RDD

• v0.1.0 (2015-05-09)