v0.6.0

• Completely rewrote Parsers in a Pythonic, non-multiprocessed way
• Remove lxml from requirements. pronto now looks for any
  available xml parser among (in that order): lxml,
  xml.etree.cElementTree, xml.etree.ElementTree.
• Added a second OwlXML parser using a TargetCollector instead of
  building a tree, that is longer but less memory-hungry
• Added a 'parser' kwarg to Ontology to force usage of a parser.
• Removed Parser being imported with 'from pronto import *'
• Implemented partial Unicode sandwich, at least in Relationship.
  Complete use of unicode internals is a v1.0.0 requirement.
• Added a try/except block to use enums if available to represent
  the current section of the files parsers are exploring
• Ensured PyPy support
• Add full support in Python3 and partial support in Python2 of
  gzipped files parsing.
• Refactored tests. Now all tests are run through the unittest
  interface, even doctest.
• Cleaned dead code

v0.5.0

• Use six functions when possible
• Add try/except blocks to pronto/init.py (will stop raising ImportError if six is not installed)
• Add support for obo defined relationships (typedef)
• Add support for gzipped ontology files
• Revert to nested loops in Ontology.reference (may improve performance in versions other than 3.5)

v0.4.2

• Fix obo export with utf-8 chars

v0.4.1

• Fix owl parser
• Fix obo export on headerless ontology files

v0.4.0

• Changed Term obo export to be more compliant with the Obo File Format
• Added metadata extraction of Owl files (for now, only oboInOwl tags are supported, but more xml namespaces will be added
• Added obo export of metadata (Obo headers)
• Switched imports to absolute imports
• Reversed to the stdlib multiprocessing.queues.Queue implementation. This should fix both deadlocks and #4, but sometimes produces a BrokenPipeError in background processes after the end of the parsing...
• Enhanced API documentation
• Fixed occasional duplication of Terms in TermList

v0.3.3

• Attempted compatibility patch for multiprocessing to work with
  Mac OS X

v0.3.2

• Improved Python2/3 compatibility & performance using six

v0.3.1

• Added a stricter timeout to ontology opening to avoid waiting ages for each imports

v0.3.0

• Increased multiprocessing support
• Made all pronto types pickable (protocol >=2 required)
• Fix many performance issues

v0.2.1

• Fix Relationships other than predefined ones not extracting
• Moved Relationship class properties bottomup and topdown to class methods