http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50288
--- Comment #6 from janus at gcc dot gnu.org 2011-09-05 12:26:14 UTC --- Sorry for the breakage, guys. Of course I *did* check the test case before committing, but for some reason the failures did not occur on my machine (I have no idea why). (In reply to comment #5) > A better way is the following patch, which adds "dg-do run-once", which should > then also be applied to cray_pointers_2.f90. That sounds reasonable (much better than the evil "double blank" trick). Ok with me.