Marcus von Appen wrote:
That totally defeats the purpose of Cython. One maintains the Cython code. The C code is just an intermediary step in compiling an extension module. The intermediate C is easier to follow than that generated by other high level languages, but one would not want to alter it. Have a look at pypm.c for an example generated with Pyrex.On, Fri Mar 13, 2009, Toni Alatalo wrote:Marcus von Appen kirjoitti:I wonder if porting them from Python to Cython (the new pyrex) to get the good and robust C implementation would be a good solution.On, Thu Mar 12, 2009, Iuri wrote:In the wiki we have 2DVectorClass and 3DVectorClass classes. Why theseThey are good, but in Python, making them slower than a good and robust C implementation.I do not know anything about Cython, but if the code it generates is actually maintainable (means: well-structured, readable and easy to modify), it might be worth a try.
But I do have my concerns. Cython is not a stable language. It may change to remain compatible with Pyrex, which is still in alpha development. This just means keeping ones Cython compiler current. Also Cython doen't support variable length types, tp_itemsize not zero. This might be of concern if vector size is an attribute.
Lenard -- Lenard Lindstrom <len-l@xxxxxxxxx>