Hi ,
We recently upgraded from 0.7.0 to 0.11.0. With the latest version(0.11.0) we are observing crash while fetching AMQP headers.
Basically we have applications written in Java and C++. For c++ applications we are using rabbitmq-c 0.11.0 version. Java based applications are using spring-rabbit version is 2.3.10. There is an exchange of messages between c++ and java via RMQ.
Scenario :
When Java based applications are setting empty AMQP headers (key is set , but value is either not set or set to NULL) , C++ applications are crashing while fetching it.

BT:
Thread 1 (Thread 0x7f86ad7fa700 (LWP 27862)):
#0 0x00007f893d1b6387 in raise () from /lib64/libc.so.6
#1 0x00007f893d1b7a78 in abort () from /lib64/libc.so.6
#2 0x00007f893dac6a95 in __gnu_cxx::__verbose_terminate_handler() () from /lib64/libstdc++.so.6
#3 0x00007f893dac4a06 in ?? () from /lib64/libstdc++.so.6
#4 0x00007f893dac4a33 in std::terminate() () from /lib64/libstdc++.so.6
#5 0x00007f893dac4c53 in __cxa_throw () from /lib64/libstdc++.so.6
#6 0x00007f893dac51ed in operator new(unsigned long) () from /lib64/libstdc++.so.6
#7 0x00007f893db23cd9 in std::string::_Rep::_S_create(unsigned long, unsigned long, std::allocator const&) () from /lib64/libstdc++.so.6
#8 0x00007f893db23e96 in std::string::_M_mutate(unsigned long, unsigned long, unsigned long) ()
from /lib64/libstdc++.so.6
#9 0x00007f893db2445e in std::string::_M_replace_safe(unsigned long, unsigned long, char const*, unsigned long) () from /lib64/libstdc++.so.6
#10 0x00007f893f487b65 in AMQPMessage::addHeader (this=0x7f8690003000, name=0x7f86ad7eefc0,
value=0x7f86ad7eefb0) at src/AMQPMessage.cpp:138
#11 0x00007f893f48ae3e in AMQPQueue::setHeaders (this=0x7f86900e7fb0, p=0x7f87e4098228)
at src/AMQPQueue.cpp:738
#12 0x00007f893f48a868 in AMQPQueue::sendConsumeCommand (this=0x7f86900e7fb0) at src/AMQPQueue.cpp:674
#13 0x00007f893f489f6b in AMQPQueue::Consume (this=0x7f86900e7fb0, param=0, timeout=1)
at src/AMQPQueue.cpp:511

Core Analysis : frame #12
(gdb) p (p->headers.entries[0].value.value.bytes.len)
$13 = 6
(gdb) p (p->headers.entries[1].value.value.bytes.len)
$14 = 3472615286093461794 ---------> Len appears to be some garbage value .
(gdb)

When run on valgrind , we observed invalid read at AMQPMessage::addHeader() due to un-initialized heap.
orign source of un-initialized block was pointing to malloc() at amqp_decode_table() in amqp_table.c

we tried initializing the memory using memset() after malloc(). With this we are not observing crash and able fetch the headers correctly.
entries = malloc(allocated_entries * sizeof(amqp_table_entry_t));
memset(entries, 0, allocated_entries * sizeof(amqp_field_value_t));

Please note that this issue was not observed with rabbitmq-c 0.7.0.
Also we do not observe this if AMQP headers are set and fetched within c++ applications using rabbitmq-c 0.11.0.

Kindly please check on this. If the workaround we applied is a valid fix , please do consider it to fix in your official release.