[boost] Re: [Threads] Failures on msvc, msvc-stlport

From: Michael Glassford (glassfordm_at_[hidden])
Date: 2005-05-26 11:49:29

OK, I've done this. Thanks for the info.

Mike

Joaquín Mª López Muñoz wrote:
> 
> Michael Glassford ha escrito:
> 
> 
>>As far as I can see, these are all type_traits compile errors in the
>>Boost.Thread regression tests. Did something change fairly recently in
>>the type traits library or the way that Boost.Test uses it that might
>>cause this? Unfortunately, I don't have access to VC6 to help me debug
>>it, and don't know enough about changes in those or other libraries to
>>debug the problem "in a vacuum".
>>
> 
> 
> The problem is due to an internal typedef in the is_convertible implementation
> called "binder", which clashes with your "binder" in libs\thread\testutil.inl.
> The "binder" in is_convertible was introduced about two months ago,
> hence the regression.
> I don't know the exact circumstances where this compiler bug shows, but I
> know I've been bitten by it in the past. Simply renaming your binder to
> anything else (like in the attached patch) solves the problem.
> 
> HTH
> 
> Joaquín M López Muñoz
> Telefónica, Investigación y Desarrollo
> 
> 
> ------------------------------------------------------------------------
> 
> 143c143
> < class binder
> ---
> 
>>class binder_class
> 
> 146c146
> <     binder(const F& func, const T& param)
> ---
> 
>>    binder_class(const F& func, const T& param)
> 
> 156c156
> < binder<F, T> bind(const F& func, const T& param)
> ---
> 
>>binder_class<F, T> bind(const F& func, const T& param)
> 
> 158c158
> <     return binder<F, T>(func, param);
> ---
> 
>>    return binder_class<F, T>(func, param);
>>
>>
>>------------------------------------------------------------------------
>>
>>_______________________________________________
>>Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost