On 15/09/2011 21:28, mark Pashley wrote:
Hi all,

If any function called by a BOOST_GLOBAL_FIXTURE makes a BOOST_XXXX macro call my test case is terminated with a SIGSEGV. I assume that this is because some part of the boost test framework has not been initialised as the point where by call is made.

Is this a known problem or is this intentional behaviour?

If this is intentional what should I do instead of calling BOOST_MESSAGE / BOOST_FAIL / BOOST_CHECK (for now I have replaced the last 2 with assert but this is not ideal...).


There is actually a trac ticket [1] about this.
It should be pretty trivial to fix, however it seems for now Boost.Test regression tests are broken due to [2]...

[1] https://svn.boost.org/trac/boost/ticket/5563
[2] http://boost.2283326.n4.nabble.com/test-Running-regression-tests-td3568224.html