ANNOUNCE: PySymbolic - Doing Symbolics in Python

[Alex Martelli]

"Janko Hauser" <jhauser at ifm.uni-kiel.de> wrote in message
news:87og0r39n5.fsf at ifm.uni-kiel.de...
> Perhaps I'm wrong, but I understood that GiNaC does not directly
> expose an API, but helps a C++ programmer to write expressions
> directly in C++, which would mean that the interface itself needs to
> be written and beforehand defined in C++. So SWIG would not be an
> option, perhaps CXX or something like that. Another system to look at
> is YACAS, which has an API.

My impression is that YACAS, like a typical CAS system,
implements and exposes its own little scripting language,
while GiNaC's novelty is that it *doesn't* -- it _just_
implements and exposes the functionality at the low
(i.e., C++) level.  So, if one's purpose is basically
to use it as an engine from _another_ scripting language
(namely, Python:-), one should be able to avoid having
to "work around" the intrinsic scripting language, &c.

As I said, it's an _impression_ -- I haven't made any
in-depth study...

SWIG, in theory, should be usable to interface Python
(and other scripting languages it supports) to C++ code
just as well as to C code.  CXX and py_cpp are, clearly,
also feasible alternatives.


Alex