On Thursday 19 October 2006 04:14, Bob Ippolito wrote: > There's always ways to optimize things later. Write the code. If it's > not fast enough, *profile* the code to see what's not fast, and take > steps to optimize that part of the code. > > Sometimes using psyco is going to be enough, depending on which > platforms you care about. Otherwise you could change the rendering > strategy to use OpenGL (which is still the fastest way to do anything, > even in 2D), or writing some of the tight loops in C or C++ (possibly > via Pyrex).
This is very good advice and bears repeating through just quoting it again :) I should also note that I've written quite a number of action-oriented games in Python now and have *never* had to resort to coding parts of them in C or C++. If you find your game is slow and you can't figure out how to improve performance, I recommend asking for advice on this list. That almost always results in improvements. For example: I went down the re-coding-in-C path once a long time ago when I was writing a 3D game. I coded a number of parts of the low-level rendering in C in response to poor performance. Then someone clued me in to OpenGL display lists. I ditched the C bits because they were *slower* than using display lists. Richard ps. there's a wiki page somewhere out there that lists python game speed tips. I know there is. I can't find it though :(