------- Comment #3 from longb at cray dot com 2010-02-05 23:49 ------- This bug was filed because it appeared that the test worked with the Cray, PGI, and Intel compilers. However, based on the notes in Comments 1 and 2, more tests were written which uncovered that the Cray compiler seems to do the right thing for an F03 compiler (dynamically allocating and deallocating the thread private versions of the allocatable array), whereas PGI and Intel were effectively sharing the array contrary to the threadprivate directive. Based on this, the OpenMP testers agreed that the interpretation (at least for a pre-F03 compiler) given for gfortran was valid and the Bug can be closed.
-- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42865
