From: David Abrahams (dave_at_[hidden])
Date: 2004-01-03 15:35:55
I don't know how or why this nastiness comes about, but it's true.
Just think of how many checks which look for __GNUC__ are foiled by
this one!
I am inclined towards a new policy for Boost which requires compiler
identification only via BOOST_-prefixed macros.
If we do that, we can also ensure that BOOST_CONFIG #defines all of
them (mostly to zero) and we can eliminate the issues with warnings
from BOOST_WORKAROUND doing comparisons on macros which haven't been
defined.
Thoughts?
-- Dave Abrahams Boost Consulting www.boost-consulting.com