Numpy BoF at Scipy 2014

BoF logistics: Thursday July 10 1:30 p.m.–2:15 p.m. Room 204.

Notes from the discussion stored on a shared etherpad.

Fernando Perez will act as discussion facilitator.

Please use this page to collect ideas and topics for discussion for the Birds of a Feather session we'll hold at the Scipy 2014 conference. We'll use the content of this page as a rough agenda to keep the conversation on track.

This is my (Chuck's) first cut at listing topics for discussion.

• Status of Numpy and near future development

  1. Generalized ufuncs, not all ufuncs that can be generalized have been.

  2. Implementation of the @ operator.

  3. Datetime, Pandas vs Dynd vs Numpy, can we all agree on one implementation:

     • http://mail.scipy.org/pipermail/numpy-discussion/2013-April/066038.html
     • http://mail.scipy.org/pipermail/numpy-discussion/2014-March/069554.html

     A C++ standard proposal for datetime format: http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2012/n3344.pdf

  4. Problems: new types, missing values, inheritance from ndarray.

  5. Text array issues: (bytes, vs string vs unicode, py2 vs py3):

     • http://mail.scipy.org/pipermail/numpy-discussion/2014-January/068622.html
     • http://mail.scipy.org/pipermail/numpy-discussion/2014-January/068702.html
• Status of Dynd and how it differs from Numpy
  1. Mark Wiebe ?
• Avoiding the python2 -> python3 debacle
  1. Evolution of Numpy to Numpy++
  2. Straight replacement
  3. Can Numpy be evolved?
  4. Can Dynd be made backward compatible?
  5. Versioning?
  6. dtypes?
  7. Buffer protocol?
  8. Scalars?