These are chat archives for symengine/symengine

21st
Jul 2015
Abinash Meher
@abinashmeher999
Jul 21 2015 04:11
ok
Isuru Fernando
@isuruf
Jul 21 2015 11:09
@certik, @sushant-hiray, can you check http://trac.sagemath.org/ticket/18078 on OS X?
Sumith Kulal
@Sumith1896
Jul 21 2015 13:03
@certik Will do
Ondřej Čertík
@certik
Jul 21 2015 14:29
I'll try. My machine with OSX doesn't work properly (some header files are broken), but I'll try to find another one. It would be nice to write to the Sage mailinglist soon to ask people for feedback. @isuruf have you been following the "performance" thread on the sympy list? I posted some results yesterday using your MPC and other evaluators for very large expressions and they work correctly (checked against sympy) and are also extremely fast. I plan to do the same example using Sage. And create a nice file that does it using SymPy, Sage and SymEngine, all withing Sage.
Hopefully symengine will be the fastest.
Isuru Fernando
@isuruf
Jul 21 2015 14:31
Thanks.
Yes, I saw the performance thread. Let me know how I can help.
It'll be great if we can implement most of what PyDy people want.
Isuru Fernando
@isuruf
Jul 21 2015 16:58
@certik,
    # MSVC only supports __func__ in a C++ mode, not a C mode
    set(HAVE_C_FUNCTION_NOT_FUNC yes)

    # MSVC does not support =default for constructors
    set(HAVE_DEFAULT_CONSTRUCTORS no)
Is this true for VS 2015
Isuru Fernando
@isuruf
Jul 21 2015 18:48
I'm also looking at Python extensions using MSVC14. Two tests fail.
assert set((x**2 + 2*x*y + 5).args) == set((x**2, 2*x*y, 5))
Found the error. It should be ` assert set((x**2 + 2*x*y + 5).args) == set((x**2, 2*x*y, Integer(5))) since hashes of Integer(5) and 5 may not be equal.
With all the testing in Linux, OSX and Windows, it fails only with MSVC
Francesco Biscani
@bluescarni
Jul 21 2015 23:56
@certik that type of expression with symbolic exponents cannot be handled by Piranha