On Jun 18, 2010, at 2:57 PM, Lisandro Dalcin wrote: > See the code below. As expected, you DO get an error (though the error > is ERR_INTERN, somewhat not informative). However, if you fist > create() % destroy a windows, you DO NOT get any error. This is VERY > strange, right?
It looks like it's a combination of two things: 1. A fix that you brought up a long time ago was applied to the trunk but not to the v1.4 branch (disallowing calling TYPE_FREE on predefined types). I just filed CMR #2448 to fix this. 2. It looks like MPI_WIN_CREATE is incrementing the refcount on MPI_TYPE_BYTE (assumedly by sending a message?), but that refcount is never getting decreased (perhaps the message is not received?). I do note that this does not happen on the trunk (the refcount stays at 1); it only happens on the v1.4 series. I'm not sure how high priority it'll be to fix this, but I filed ticket #2449 about the issue. Thanks for reporting these issues! -- Jeff Squyres jsquy...@cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
