RE: [boost] boost::config and covariant return types

From: Beman Dawes (bdawes_at_[hidden])
Date: 2001-09-30 20:10:07

At 02:15 PM 9/27/2001, Fernando Cacciola wrote:
 >
 >----- Original Message -----
 >From: Beman Dawes <bdawes_at_[hidden]>
 >To: <boost_at_[hidden]>; <boost_at_[hidden]>
 >Sent: Thursday, September 27, 2001 2:56 PM
 >Subject: Re: [boost] boost::config and covariant return types
 >
 >
 >> At 11:06 AM 9/27/2001, Markus Schoepflin wrote:
 >>
 >>  >I just noticed that there is no define for detecting the absence of
 >>  >covariant return types in the config system.
 >>  >
 >>  >Would it be possible to add something like
 >>  >BOOST_NO_COVARIANT_RETURN_TYPES to detect missing support for section
 >>  >10.3.5 of the standard?
 >>
 >> Keep in mind that the Boost config system is intended for internal use 
in
 >> Boost libraries.  It isn't intended to solve general broken compiler
 >> problems.
 >
 >Quite true.
 >However, I wonder: would it be possible to Standardize a header along the
 >lines of boost config? That might allow every piece of code, even outside
 >boost, to have a chance to identify and deal with broken compilers.
 >Perhaps boost\config.hpp can be considered by the LWG as a starting point
 >for a std <compiler_config>.

For the reasons given by Jens and others, I doubt you would find any 
support at all from the standards committee.

Remember too that C++ compilers are getting closer to being standards 
conforming.  Metrowerks shipped their latest September 10th.  GCC has put 
out two releases this summer and has committed publicly to an aggressive 
schedule of future releases.  Microsoft is well into a second beta for 
their upcoming release.  Comeau is claiming they will soon support 
"export".  In private communications I've had with several vendors, it 
seems there is increased interest in making sure compilers can handle at 
least the Boost libraries, if not more portable, standards conforming C++ 
code in general.

--Beman