Hmmmmm....
The more we know, the less we know. We now know it wasn't a Safari flaw. So... was it *actually* a Quicktime flaw or (possibly more likely) a Java flaw....
The zero-day vulnerability that allowed a hacker to commandeer a brand new MacBook Pro late last week resides in a flaw in Apple's QuickTime media player, the exploit's author says. The revelation corrects descriptions given last Friday that the exploit targeted Safari. Dino Dai Zovi set the record straight in a blog posting …
I think it's more likely that the flaw is a combination of the two programs, probably Java generating some data, or doing something that it isn't supposed to do, since web java is supposed to play in a sandbox, and then quicktime, when presented with data that is totally unexpected, after all, you can't test for every possible case, and hence overflowing with some executable code causing a remote shell to pop up.
Not checking bounds is always bad. It's one of those things almost everybody does (in particular thanks to the C/C++ languages which leave this task explicitly to the programmers) but it is a very bad practise nevetheless.
Considering the speed of modern hardware, there is no reason to omit bounds checking. It is about time that programmers are getting their butts kicked to always do bounds checking, no exceptions ever allowed. Better still, compilers should be upgraded to apply bounds checking by default.
Until that time comes, we will have to put up with software ridden with security holes and bugs like a Swiss cheese.