• They'll video tape

• Cancelations, so some one could move to tomorrow

• Python3, Guido's Talk

  • Python 2.5, some improvements

  • Python 3000

    • Language update
    • Stroustroup compatibility
      • C++ backwords, add a new keyword, was a major incompatibility
    • Perl 6 other extreme

  • -Q warn about integer errors, // <- integer division

  • Pep 3100 for info, python.org/dev/peps

  • guidos blog, artima.com/weblogs/

  • big changes

    • kill classic classes
    • Exception derive from single root
    • absolute import ( can be turned on in 2.5 )
    • replacement for d.has_key if x in d
    • interating over file is better than xreadlines

  • smaller changes

    • exec, function again

    • x in favor repr(x)

    • except x,y,z as err instead except e,err

    • generator replaces list comprehension * list comprensions currently leak their name into the surrounding

      System Message: ERROR/3 (<string>, line 38)

      Unexpected indentation.

      scope

    • kill raise E, use raise E(arg) (work away from string exception)

    • xrange has a problem with long ints

      • will change range to do xrange things with lon support

  • string type updates

    • string becomes unicode
    • also introduce a byte type, all conversions from byte to strings require specifying conversion

  • new standard i/o stack

    • replace the c stdio with something a bit more inspired by java * unbuffered io * buffered io built on top
    • see sandbox/sio/sio.py for early prototype

  • print becomes a function

    • easier to change print to something
    • 2.x/3 compatibility requires only printing single string

  • dict views instead of lists

  • drop default <, <=, >, >= for classes (currently does memory)

  • Google python sprint next week

• Understanding NumPy, Travis Oliphant

  • numeric compatibility layer, numpy.oldnumeric
  • dtype object ( defines what the elements of the NumPy array )
    • array scalar ( wraps element returned from cell )
    • dtype code '<i4' (byte order, type, size)
    • isbuiltin, 21 static objects
    • dt.itemsize
    • fields, since array is just chunks of memory, because of alignment issues
    • a.flags ( properties on arrays )
    • a.empty((10,),dtype='int,float,a10') * a.dtype * a.dtype.fields
    • arbitrary structure needed to memmap complicated files

• Extending NumPy, swig, f2py, pyrex, ctypes,

  • swig, first
  • weave, embed C code in python string, kinda fortran like in variable names
    • compiles on the fly
  • pyrex
    • from c_numpy cimport, ... import_array()
  • ctype, calls shared libraries
    • .argtypes allow specifying type checking for c calls

• Python for modern scientific algorithm development, Fernando Perez

• Building a Distributed Component Framework (pyre), Michael Aivazis

  • distributing apps on many machines is hard
  • user space, ssh, scp, pyre services
  • pyGridWare (Kieth jackson)
  • gsl
  • http://www.cacr.caltech.edu/projects/pyre
  • http://pyre.caltech.edu (soon)
  • relase early 2007

• Enthought tool suite

  • What are traits?
    • notification (considered important)
    • visualization

• Synthetic Programming with Python

  • provide low level access to cpu without doing string processing
  • http://www.synthetic-programming.org

• 3D visualization with TVTK and MayaVi2, Prabhu Ramachandran

  • Traited VTK
  • MayaVi2 uses enthought exposes lots of objects and gui tools for playing with them

• Realtime Computing with Python, Andrew Straw

  • introduction to relatime simulation of fly
  • http://dickinson.caltech.edu/Research/Grand_Unified_Fly
  • comedi control measurement device interface

• Lightening Talks

  • Dude from JPL, Mike Ressler
  • Future of IPython, Brian Granger, Tech-X corp
    • planning on extending to ipython to do distributed programming
    • control remote supercomputers from laptop
    • support multiple parallel types
    • focus on interactive work
    • distributed programming in python
      • multiple processes, threads are limited by GIL
      • non-blocking sockets
      • asynchronous error/fault
  • Travis Oliphant, about PEP
    • using trac pages for NumPy http://projects.scipy.org/scipy/numpy
    • table @ dinner about arrays
  • Travis Vaught, Enstaller
    • egg package installation manger
    • has post-install scripts
    • build egg from pointing to a directory
    • code.enthought.com/enthon/enstaller.shtml
    • enthought.com/enthought/wiki/Enstaller
  • The Current State of Vision, Michel Sanner
    • Visual programming tools
    • supports nested graph nodes
    • looks written in tk
    • www.scripts.edu/~sanner/software
  • Quick overview of Chaco, Peter Wang
    • chaco visualization tool
    • http://code.enthought.com/chacho

• SAGE, Alex Clemesha

  • good environment for algebra, number theory, geometry
  • what is it
    • Distribution of math software
    • python library
    • new interface to math software
  • where is it: modular.math.washington.edu/sage
  • their ajax based web notebook looks really impressive

• Mathematica like plotting for SAGE, Alex Clemesha

• BioHub, Diane Trout

• Software Carpentry, Greg Wilson

  • he's a good speaker.
  • debugger for make
  • debugger than can trace back and forth between python & c code
  • Guido pronounced django the web framework.