Prev Next whats_new_10

Extensions, Bug Fixes, and Changes During 2010

12-12
The independent(x) and adfun(x,y) functions require the elements of x and y to all have the same type. If this was not the case, pycppad might crash with out a useful error message. This has been fixed.

12-05
First version that works (and installs properly) under MS windows using the cygwin system; see special pycppad instructions for installing under cygwin .

11-30
Upgraded from using cppad-20100101.2 to cppad-20100101.5. This fixes a problem when installing pycppad with version 1.44.0 of boost-python .

03-18
The upstream tarball for CppAD was missing from The install process was failing because the upstream tarball was missing from http://www.coin-or.org/download/source/CppAD/ . This has been fixed, and a newer upstream release has been used.

02-10
  1. Added --inplace to the optimized build instructions. Improved the install instructions and the discussion of the python path .
  2. Changes exit to sys.exit in test_example.py and test_more.py (required on some systems).
  3. Fixed the ad_unary.py example so it displays the correct source code and changed all to numpy.all in that example (required on some systems).


01-31
Add the abort_recording function.

01-30
Change setup.py do that it patches the CppAD distribution (this fixes a problem with the optimized build).

01-29
Added the python ODE solver runge_kutta_4 which can be used with `smart' floating point types (e.g., AD types).

01-28
Upgraded from using cppad-20090909.0 to cppad-20100101.0.
Input File: omh/whats_new_10.omh