On Mon, 19 Jul 2004 06:53:50 -0700, Bert Sierra
Caveat: I am Python biased, but...
While it is interpreted it is not normally as slow as your
figures suggest and optimisers like Psycho and Pyrex are normally
very good at speeding up CPU intensive jobs like number
crunching. Also there are several math specific libraries
written in C.
From my experience I find, with C as a base, that
performance ranges like this:
C - 1
C++ - 2-5 (assuming using OOP features)
Java native code - 3-5
Java Bytecode - 5-10
Python bytecode - 10-20
Python optimised - 3-15 (depends heavily on app)
For Python to be 100 times slower than Java I suspect some
major inefficiency in the Python implementation of the algorithm!
But the general point is valid that Python is interpreted and
should probably be compared to Applescript rather than
Objective C in a Mac programming context.
And Python has the wonderful PyGame application kit that provides
most of the CPU hitting bits in C wrappers...
But I admit I'm biased :-)
Author of the Learn to Program website