[RFH] Puzzling behaviour with a wstring constructor

[Nathan Myers]

On Tue, Mar 18, 2003 at 09:56:27PM +0100, Paolo Carlini wrote:
> with current mainline (vs 3.2.2, 3.3 pre) the following doesn't link:
> 
>  // template<typename _InputIter>
>  //   basic_string(_InputIter begin, _InputIter end, const allocator& a)
>  // where _InputIter is integral [21.3.1 para 15]
>  std::wstring s(10, 0);
> 
> On the other hand the string version is ok.
> Also, FWIW, are both ok:
>    std::wstring s((unsigned)10, 0);
> and
>    std::wstring s(10, (unsigned)0);
> 
> I'm really puzzled: is the testcase above (which I blindly adapted from 
> the string version) correct or must be amended?
> ... 
> P.S. EDG + Dinkum also compile it just fine.

Is this a place for another of those "do the right thing" hacks?

Nathan Myers
ncm-nospam@cantrip.org