Re: [Boost-testing] request for help with serialization tests on someplatforms.

From: K. Noel Belcourt (kbelco_at_[hidden])
Date: 2008-03-21 14:38:43

On Mar 21, 2008, at 1:07 PM, Robert Ramey wrote:

> K. Noel Belcourt wrote:
>>> pgi- 7.0
>>>
>>> Still hoping for a fix for BOOST_STATIC_WARNING for this
>>> platform.
>>
>> FYI, I'm going to drop testing for pgi-7.0 and will be moving to the
>> pgi-7.1.x compilers shortly (a couple of weeks).  I'm hopeful this
>> will clean up some problems.  Perhaps you can hold off on pgi related
>> fixes until I get 7.1 installed?
>

Hi Robert,

> I wouldn't call it a "fix".  BOOST_STATIC_WARNING basically
> has to be "configured" for each compiler.  Given that this hasn't  
> every
> been done for the pgi compiler - its not surprising that it has  
> come up.
> I would like to get this in "once and for all" - Until this is one,  
> we won't
> see any results for any pgi compiler.

The pgi-7.1 compilers work out of the box (i.e. they emit the warning  
as they should).

> It should only take a couple of minutes.  Just compile
> test_static_warning.cpp
> and tweak static_warning.hpp until you can get a warning message as  
> opposed
> to silence or synax error.  I woule be grateful if you could do this.

The new pgi-7.1 compilers do emit a warning:

[kbelco_at_sahp7645 test]$ pgCC -I../../.. test_static_warning.cpp
"test_static_warning.cpp", line 19: warning: delete of pointer to  
incomplete
           class
   BOOST_STATIC_WARNING(sizeof(a1) == sizeof(a2)); // Warn.
   ^

What I failed to convey to you was that I am uninterested in fixing  
the pgi-7.0 compilers at all.  To the best of my knowledge, we are  
the only organization interested in using Boost with Pgi and we have  
(or will shortly) stopped using pgi-7.0, that's why I'm less  
interested in investing time to make Boost work with pgi-7.0.

-- Noel