pyGiNaC Code

86 lines (74 with data), 3.8 kB

This project is licensed under the GNU General Public License, version 
2.0.  The contents of the bin/ directory have mostly been ripped 
wholesale from the checks/ directory of the GiNaC 1.3.0 tarball.  All
other components are Copyright (c) 2004 Jonathan Brandmeyer.
This version is patched to work with the current GiNaC by Vladimir V. Kisil.

--- Portability ---
PyGiNaC itself only uses ISO C++.  GiNaC and CLN, upon which GiNaC is
built, can be build on OS-X or MS Windows. Look the GiNaC web site
(http://www.ginac.de),  MoebInv Wiki
(https://sourceforge.net/p/moebinv/wiki/Mac%20installation/) and the Web
for further advises.


Each dependancy includes the name of the package in Debian Sid and Sarge that
satisfies the requirement.

Dependencies:
	Python 2.2 or higher, 2.7.11 recommended (python2.7-dev)
	The Boost libraries 1.31.0 or higher (libboost-python-dev)
	GiNaC 1.3.0 or higher, tested with 1.6.7-1.7.0 (libginac-dev)
	(pyGiNaC v.1.2 is the last version working with GiNaC version prior 1.5)

Build dependencies (only needed to build PyGiNaC):
	Scons (scons)
	GNU G++ 3.2 or higher (5.3.1 tested) (g++-5)
	Lotsa RAM.  The compiler takes a _lot_ of memory to build PyGiNaC, although
		you will not generally need nearly as much to run it.
	
Building:
	Once you have all of the dependencies and build-dependencies installed on
	your system, you only need to run one command:
	$ scons
	This will build the extension module in-place.  "scons install" is not
	yet supported, since this extension is very young.

Usage:
	You can use the "run" script to start an interactive session of Python that will be
	able to use PyGiNaC:
	$ ./run python
	or
	$ ./run python bin/some_script_file.py

	It is better to use some wrapper of Python, e.g. iPython, idle or pyzo. In the case
	of iPython the above example shall look like:
	$ ./run ipython

Files:
	src/*: C++ source files to bind Python to the C++ GiNaC library.
	src/check: C++ source files of tests included with GiNaC, to be ported
		to PyGiNaC and placed in bin/
	include/: C++ header files
	site-packages/ginac/*: Python source files
	site-packages/: Location for in-place construction of the target shared
		object.
	bin/*: Standalone Python scripts.  These scripts are all tests or demos
		of the PyGiNaC interface.
	SConstruct: The file which controls the build.  See scons(1).
	INSTALL: This file
	run: An executable Bourne shell script that sets up environment variables
		for running PyGiNaC-using scripts against a freshly built version of
		PyGiNaC in-place.
	linux-symbols.map: A linker version script for GNU ld on ELF.  By making 
		all	symbols other than the import function local, the dynamic symbol
		table is much smaller.  (Note that Boost.Python uses a lot of
		templates, which makes for a _large_ number of _really_long_ symbols).
		This makes loading the extension module much faster, and the size 
		of the shared object is reduced by roughly 50%.  This will probably be
		replaced in the future with GCC 4.0's -fvisibility option.

TODO: These were high-level projects of the original developers.  The
	current maintainer does not have resources and abilities to
	complete them. Feel free to send your patches for these and
	other improvements.

	* Write Python versions of the tests found in src/tests
	* Create a map_function system, probably based on python function
		objects
	* Wrap more of the GiNaC classes and objects.
	* Add extra member access functions for higher-level access of containerish
		types, like power.basis() and power.exponent(), for example.
	*  Make every func capable of taking more than one argument also take args
		in keyword form.
	* Port the documentation examples to talk about PyGiNaC's interface rather
		than ginsh or GiNaC C++.