I guess most Virtools users already wondered over the precision problems when using vectors. For example: you enter a number and it transforms to a different value (or changed after reloading), or you iterate through some partitioned dimension and the accumulation of all the steps is not equal to the original dimension size.
Tom Forsyth is a programmer at Rad Game Tools, he is also very present on several GameDev related mailing lists. He published a couple of interesting articles – for example about shadow maps (for Startopia) and he also has an interesting blog. There he explains the precision problem with floats and it’s also understandable for non-hardcore programmers like me 😉 He mainly refers to large-scale worlds but I guess it has a much broader effect. I wonder if other projects do consider this? I think, I mostly see floats everywhere …
Darn dom, that’s one hell of a nice blog there ! Thanks for pointing it out 🙂