config test errors on gcc-4.1.2

[troy d. straszheim]

on a debian-testing machine with gcc-4.1.2 limits_test dies with a link error, and has_pthread_yield_fail and no_ded_typename_fail both compile, though the jamfile thinks they shouldn't. I'm looking in to this.

[troy d. straszheim]

it would seem that no_ded_typename_fail.cpp should check for

BOOST_NO_DEDUCED_TYPENAME

rather than

BOOST_DEDUCED_TYPENAME

looking at the others...

[troy d. straszheim]

it isn't clear to me how boost_has_pthread_yield_fail could ever have passed on machines that do have it... the only place BOOST_HAS_PTHREAD_YIELD is mentioned in config is commented out in aix.hpp. Maybe I'm missing something. Maybe such plaforms haven't come up yet.

[troy d. straszheim]

Somebody tell me if I'm generating too much spam by updating this thing too often.

The mac does have nanosleep and pthread_mutexattr_settype. These need adding to macos.hpp

# define BOOST_HAS_NANOSLEEP # define BOOST_HAS_PTHREAD_MUTEXATTR_SETTYPE

[troy d. straszheim]

I've been told that these tests aren't intended to be run by users but are rather a mechanism used to generate configurations. Sorry about the noise... it is confusing to have broken tests laying around in released, tarballed software.