Release Notes

Release 0.3.4

Release:0.3.4
Date:September 15, 2015

New Features

  • Added support for Spark 1.4 on Python 2.7 and Python 3.4 (#294)

Experimental Features

Warning

Experimental features are subject to change.

  • Add support for specifying primary and foreign key relationships in the SQL backend (#274).

New Backends

None

Improved Backends

  • Dialect discovery on CSV files now samples a subset of the file. This allows S3(CSV) to have correct values for its dialect (#293).
  • Loading a set of files on a s3 bucket with a prefix into a redshift database now works (#293).

API Changes

None

Bug Fixes

  • Cope with Dask and bcolz API changes (#270).
  • Fixed a bug where columns in dshape were being ignored when converting a numpy array to a DataFrame (#273).
  • Fix appending into a sql table from chunks not returning the table. (#278).
  • Fix a bug where 'pytables://' wasn’t being properly stripped off the URI (#292)
  • Fix a bug where a non-existent header row was being removed from an S3(CSV) because the dialect was set incorrectly (#293)
  • Fix a bug where the SparkSQL backend wouldn’t work if we didn’t have paramiko installed (#300)
  • Fix a testing bug where the endlines were being compared and they shouldn’t have been (#312).
  • Fix a bug where sniffing multibyte encodings potentially chopped off part of the encoded string (#309, #311).

Miscellaneous

  • Adds copydoc() function to copy docstrings from one object onto another. This helps with the pattern of explicitly setting the __doc__ attribute to the __doc__ of another function or class. This function can be used as a decorator like: @copydoc(FromThisClass) or as a function like: copydoc(FromThisClass, to_this_function). (#277).

Release 0.3.3

Release:0.3.3
Date:July 7th, 2015

New Backends

None

Improved Backends

  • Implement SQL databases to CSV conversion using native the database dump (#174, #189, #191, #199).
  • Improve CSV header existence inference (#192).
  • Non-standard schemas can be passed into resource() with the schema argument (#223).

API Changes

  • unicode strings can be passed in as URIs to resource() (#212).

Bug Fixes

  • Fixed writing compressed CSVs in Python 3 and Windows (#188, #190).
  • Dask API changes (#226).
  • Fix some tests that would fail on binstar because they weren’t properly skipped (#216).
  • PyTables API compatibility when given a integer valued float (#236).
  • Default to None when plucking and a key isn’t found (#228).
  • Fix gzip dispatching on JSON discovery (#243).
  • ~odo.chunks.Chunks wrapping iterators can now be discovered without consuming the first element.