[LLVMdev] optimization assumes malloc return is non-null

[David Vandevoorde]

On May 4, 2008, at 10:29 PM, Mike Stump wrote:
> On May 1, 2008, at 3:39 PM, David Vandevoorde wrote:
>>
>> Not quite.  Although there is a requirement there (and more precise
>> ones in Clause 3), there is no prohibition against doing additional,
>> observable stuff (e.g., log the calls) and hence allocations cannot
>> be elided.
>
>
> That's correct, there is no prohibition, but, if one does, there are
> no requirements placed upon the semantics of the program, none:

I don't read it that way.  I.e., I read it as saying that you can  
"add" to the "Required behavior".  But now I'm not sure -- so I'll ask  
the committee.

	Daveed