------- Comment #5 from 3dw4rd at verizon dot net 2009-08-14 22:10 ------- (In reply to comment #4) > Can you extract the tests that were failing and add them to the testsuite? >
I will do that. Also, I just found out something interesting as I was searching for a nice way to test this. I failed to read through the Concept requirements correctly. We are supposed to have operator== for the distributions (and for the param_type in each distribution). I assume that unconceptified random will still require operator==. (I'll wait and see) For now, I will have compile only tests on operator<< and operator>>. In a separate patch that adds operator== we'll actually test that serialization/deserialization works as expected. That is if things go as I think... I'm glad we had this bug ;-). -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41061
