Boost Testing :
From: Niko Demmel (niko.demmel_at_[hidden])
Date: 2007-10-28 12:10:03
On 26/10/2007 18:24, Douglas Gregor wrote:
> On Fri, 2007-10-26 at 08:13 +0100, John Maddock wrote:
>> Niko Demmel wrote:
>>> The latest regression.py from the svn trunk (with both tarball and
>>> svn) failes when building process_jam_log.
>>> From the output of regression.py:
>>> darwin.compile.c++ bin/darwin/release/process_jam_log.o
>>> ../../../boost/test/impl/execution_monitor.ipp: In constructor
>>> 'boost::detail::signal_action::signal_action(int, bool, bool, char*)':
>>> ../../../boost/test/impl/execution_monitor.ipp:541: error: expected
>>> id-expression before '(' token
>>> "g++" -ftemplate-depth-128 -O3 -finline-functions -Wno-inline
>>> -fPIC -dynamic -Wno-long-double -no-cpp-precomp -DBOOST_ALL_NO_LIB=1
>>> -DBOOST_FILESYSTEM_STATIC_LINK=1 -DBOOST_SYSTEM_STATIC_LINK=1
>>> -DBOOST_THREAD_POSIX -DNDEBUG -I"../../.." -c -o
>>> "bin/darwin/release/process_jam_log.o" "../src/process_jam_log.cpp"
>>> ...failed darwin.compile.c++ bin/darwin/release/process_jam_log.o...
>>> Any ideas?
>> Doug Gregor has reported this already:
>> http://article.gmane.org/gmane.comp.lib.boost.devel/166790 but he's also
>> reported it fixed? However, since his results aren't showing up either
>> maybe there is still a problem somewhere... Doug?
> I believe I've fixed this. My results aren't showing up because even
> though process_jam_log gets built, regression.py isn't finding the
> resulting binary.
Have you submitted the fix? If so, I have a different issue. With the a
svn checkout made yesterday I still had trouble building process_jam_log
../../../boost/test/impl/execution_monitor.ipp: In constructor
'boost::detail::signal_action::signal_action(int, bool, bool, char*)':
../../../boost/test/impl/execution_monitor.ipp:541: error: expected
id-expression before '(' token
Line 541 is an assertion (which shouldn't be essential)
BOOST_TEST_SYS_ASSERT( ::sigemptyset( &m_new_action.sa_mask ) != -1 );
I commented it out and it built just fine. Not sure whats going on...
The regression run now works and there were no other complications. (I use Rene's run.py with svn)