Alas, I don't have any access to AIX or HP-UX either, so I can't try
to reproduce this. If you could recompile with debugging symbols, and
with -O0, I think that may help track down what's going on.
But based on this:
#3 0x1006aac0 in __ct__Q2_3std9allocatorXTQ3_3std9_Tree_nodXTQ2_3std12_Tmap_traitsXTQ2_3std12basic_stringXTcTQ2_3std11char_traitsXTc_TQ2_3std9allocatorXTc__TiTQ2_3std4lessXTQ2_3std12basic_stringXTcTQ2_3std11char_traitsXTc_TQ2_3std9allocatorXTc___TQ2_3std9allocatorXTQ2_3std4pairXTCQ2_3std12basic_stringXTcTQ2_3std11char_traitsXTc_TQ2_3std9allocatorXTc__Ti__SP0__5_Node_GQ2_3std4pairXTCQ2_3std12basic_stringXTcTQ2_3std11char_traitsXTc_TQ2_3std9allocatorXTc__Ti__RCQ2_3std9allocatorXTQ2_3std4pairXTCQ2_3std12basic_stringXTcTQ2_3std11char_traitsXTc_TQ2_3std9allocatorXTc__Ti__
I'm guessing the problem is in the "delete error_" line of
RE::Cleanup(). This seems perfectly safe, so I don't know why it's
dying. If the "this=0x3" claim is accurate, then it looks like it may
be a double-delete or something? It's hard to say.
