[boost] Re: Workaround for is_better_conversion?

From: Eric Friedman (ebf_at_[hidden])
Date: 2003-11-01 17:47:31

Alexander Nasonov wrote:
> Eric Friedman wrote:
> 
> 
>>To all-
>>
>>Seeking to extend support of the variant< type-sequence > syntax to all
>>platforms, I've returned to the problem of simulating overload
>>resolution rules.
>>
>>In this direction, I have implemented the following type trait, which
>>works as follows:
>>
>>   is_better_conversion< From, To1, To2 >
>>     --> true iff conversion unambiguously prefers To1 over To2
>>
>>I have previously determined that is_better_conversion is sufficient to
>>implement the variant< type-sequence > syntax. Unfortunately, I cannot
>>get it to compile on MSVC6 or Borland.
>>
>>If anyone could help me in this regard (see attached files), it would be
>>of great benefit to users of Boost.Variant.
>>
>>Thanks,
>>Eric
> 
> 
> I see a reference to my (almost forgotten by me) trick in the code. At the 
> time of writing it (4 Jul 2002) nobody complained. I hope this trick is 
> standard conformant, isn't it?. May be I'll check this later. The reference 
> is: http://tinyurl.com/t8hp

I don't know the standard well enough to say whether the trick is 
standard conformant or not. FWIW though, the test I provided *does* 
correctly compile on MSVC7.1 and CWPro8 (and, after a workaround, on g++ 
2.95 and up).

Eric