Re: [Boost-testing] [pgi-7.0][iostreams] No placement new?

From: K. Noel Belcourt (kbelco_at_[hidden])
Date: 2008-01-24 10:47:13

On Jan 18, 2008, at 3:27 PM, Jonathan Turkanis wrote:

> K. Noel Belcourt wrote:
>> On Jan 16, 2008, at 4:01 PM, Jonathan Turkanis wrote:
>>
>>> The compiler output for several iostreams tests suggests this  
>>> platform
>>> is missing the placement new operator (e.g., http://tinyurl.com/
>>> 3c4krh).
>>> Is that correct?
>>>
>>> If so, what is the usual workaround? Do people define their own
>>> placement new?
>>
>> Apparently so.  As a quick workaround, you may be able to provide an
>> operator new (unsigned long, std::locale *) which calls new (unsigned
>> long,  void *).
>
> I think I'm missing something. If the user has not provided a  
> placement
> new, won't the operator new taking a locale* fail?

Hi Jonathan,

I got a patch from the PGI developers that should fix this problem.   
I just need to get our sysadmin to apply the patch to our  
distribution.  I'll let you know when the patch has been applied so  
we can see if it fixes the problem.

Thanks for your patience.

-- Noel