January 2009

Surprising! as of now, in my experiences of software engineer, I met more people that read/know about TAOCP than people who read Code Complete.

It may be a probability accident (I kind of remember that probabilities and small numbers are not good friends). It may also be influenced by my fields of experiences. But assuming this is real (i.e. programmers know more about TOACP), it’s doesn’t seems right.

I consider TAOCP as a reference to write efficient code; Code Complete is a reference to write good software (eventually being a product maintained by a team). Both are important, but the first one being used way less often!


This blog is not completely abandoned … I’m just buzy lazy those days.

Just to relate my brief come back to roots: doing physics simulation in 3D ;).

I’ve been playing with Irrlicht as 3D framework. It’s good, not to complex. But the API is not always easy (you’re forced to use the irrlicht not-automatic counting ref). I haven’t found doc about coordinate systems (orientation and angle’s units) but it’s using common practices in game engine. It’s pretty fast, and the simple included demos help a lot to get started (Ogre3D don’t come with simple examples).

For the physics engine, I’ve used Bullet, and it’s really pleasant to use! and result are great! I only quickly went through some part of the code, and it’s great code, I think I will learn quite a few tricks of this code (e.g. how to avoid conditional branch (‘if’) to select between 2 unsigned integers … and it even can be usefull for efficiency on some processor like Cell (maybe related to parallelism and pipelining?). No really, this is beautiful code, and if you just want to play a little bit with physics, using Bullet directly is more fun than PAL (Physic Abstraction Layer).

I’ll post the code when/if my Domino Rally simu is done 🙂 … but without sound it’s not good.