In article <441D4EFD.1010507@omicron-persei-8.net> you write:
>YAMAMOTO Takashi wrote:
>>what i don't understand here is, why you think NULL initialization is
>>better wrt future bugs.
>>  
>>
>Would a dereference of the NULL ptr not cause immediate failure?

Not necessarily.  On older ARM systems, the kernel has a page mapped at 
address zero containing the CPU vectors.  Reading through null pointers 
works fine, and writing corrupts the vector table, which you'll only 
find out about when you next take a page fault, interrupt or whatever.

-- 
Ben Harris