Review Results: Config system

From: Douglas Gregor (gregod_at_[hidden])
Date: 2001-09-02 10:43:37

The review of John Maddock's Config system ended on Wednesday, August 29th. 
The library was unanimously accepted into Boost.

Resolutions were reached on several aspects of the Config system:

Optimism vs. Pessimism about new compiler versions
--------------------------------------------------
 * Should the Config system assume that a new compiler version has all of the 
deficiencies of the prior versions (pessimism), or should it assume that a 
new compiler is fully compliant (optimism)?

Resolution: if the macro BOOST_STRICT is defined, all workarounds will be 
disabled. If BOOST_STRICT is not defined (the default), the Config system 
will be pessimistic about new compiler versions. Running regression tests 
with this macro defined will exercise new compiler versions so that the 
Config system can be updated without requiring users to deal with the Config 
system.

EDG refactoring
---------------
Configuration code for the EDG front-end (uSGI, Alpha cxx, Comeau, and Intel 
compilers -- Jens Maurer) will be refactored into a separate compiler 
configuration header (compiler/edg.hpp?).

BOOST_DISABLE_THREADS/BOOST_HAS_THREADS
----------------------------------------
There was some discussion about moving BOOST_HAS_THREADS into the Thread 
library instead of the Config library. The resolution was to leave the Config 
system as-is (includes BOOST_HAS_THREADS).

Thank you to all of the reviewers and to John Maddock.

        Doug