yeah, but the test.cpp is the simplest test and can reproduce it : push message to activemq in thread (test it using main function) would result memory leak; and the test.cpp is so simple that you can test it ; I think the test.cpp is so easy to read; I think it is activemq-cpp who leads to memory leak, not the other codes in the test.cpp , because if you comment out the "sup->PushByteMessage" line in the sendtomq_once function , then everything is ok. But if you do not push message in thread (test it using main1 function), it has no memory leak.
So this problem is so strange. And the same problem in the current release 3.9.5. So I think if we do not find out the problem, it would greatly limit the usage of activemq-cpp, or the activemq-cpp will be not useful, because it is the common usage that concurrent threads send message to Activemq. And I also use the valgrind to test it , and valgrind also report some memory leak in activemq-cpp self. but I am not the author of activemq-cpp, so I can not understand the activemq-cpp codes and can not find what leads to memory leak. The test.cpp is so easy to read it, and is so easy to compile it ; So, I want to let you test the test.cpp , and find out what code of activemq-cpp leads to memory leak, Because you are the author of activemq-cpp , or you can find the author of activemq-cpp. test.cpp <http://activemq.2283324.n4.nabble.com/file/t379406/test.cpp> -- Sent from: http://activemq.2283324.n4.nabble.com/ActiveMQ-Dev-f2368404.html
