[cfe-commits] r91599 - /cfe/trunk/docs/UsersManual.html

[Nuno Lopes]

> Nuno Lopes wrote:
>> Author: nlopes
>> Date: Thu Dec 17 04:00:52 2009
>> New Revision: 91599
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=91599&view=rev
>> Log:
>> document -fno-assume-sane-operator-new, per Chris request.
>> please review for English grammar mistakes
>>
>>  +<dt 
>> id="opt_fno-assume-sane-operator-new"><b>-fno-assume-sane-operator-new</b>:
>> +Don't assume that the C++'s new operator is sane.</dt>
>> +<dd>This option tells the compiler to do not assume that C++'s global - 
>> and any
>> +possibly overloaded - new operator will always return a pointer that do 
>> not
>> +alias any other pointer when the function returns.</dd>
>>
> This is too strong an assumption. We can assume this for the normal 
> operator new, but placement new already violates this as specified in the 
> standard:
>
> void *memory = get_some_mem();
> T *pt = new (memory) T(); // pt aliases memory
>
> Sebastian

Yes, sorry, I already noticed that my intial patch was buggy. I've already 
commited a fix in r91600.
Nuno