[boost] Re: Re[mpl & serialization] Can't compile serialization......

From: Fredrik Blomqvist (fredrik_blomqvist_at_[hidden])
Date: 2003-11-11 17:08:54

Robert Ramey wrote:
> Firingme wrote:
>
>> Platform: Win XP SP1
>> Compiler: VS2003
>
>> TestCode : "demo.cpp" with serialization draft #13 release
>
>> Error Address:
>
>> file: greater.hpp
>
>> #if !defined(__BORLANDC__)
>>    typedef bool_<value> type;
>> #else
>>    typedef bool_<(
>>          (BOOST_MPL_AUX_VALUE_WKND(T1)::value)
>>            > (BOOST_MPL_AUX_VALUE_WKND(T2)::value)
>>        )> type;
>> #endif
>
>> Error 2975
>
> I don't have VC 7.1 installed so I can't verify/fix/track this down
> this directly. This gets invoked by a higher level STATIC_ASSERT.
> Other users
> have just commented out the STATIC_ASSERT and moved on.  If
> anyone can get to bottom of this and fix it please let me know so
> I can roll in the change.
>
>
> I suspect a couple of anomolies with VC 7.0 and VC 7.1 regarding
> mpl bool_ and if_  . Also I have some doubts about type traits
> is_fundamental<size_t>. I don't have VC 7.1 installed.  I looked at
> the compiler status page and much
> to my disappointment I don't see any run of the mpl tests.  mpl
> does have a complete test suite.  Is there any reason that this
> test can't be included in the regression test run?
>
> Robert Ramey
> _______________________________________________

FYI; I believe I have a working patch for VC7.1 that should allow a full
compile.
I'll try to run the complete tests tomorrow and isolate changes before I
post it.

(To the MPL experts: IIUC the core-problems are found in level.hpp &
tracking.hpp.
I'd say a 'clean' solution needs to do similar workarounds here as MPL does
in it's integral-constant code)

Regards
//Fredrik Blomqvist