HiddenBug

We could split the commits which are only needed to satisfy the clang compiler into a separate PR. @agutenkunst What do you think? See PR #283.

For information on how-to execute a sanitizer locally see PR pilzde/.github#7.

As it turns out gcc already supports address and undefined behavior sanitizing (see [gcc online documentation](https://gcc.gnu.org/onlinedocs/gcc/Instrumentation-Options.html)). Furthermore, using clang introduces false-positive error detection's. This seems to happen when a library...

> +1 for adding sanitizing > > Why not use cmake arguments instead of environment variables for ENABLE_ADDRESS_SANITIZING, etc.? Good question :smile: I will give it a try.

> +1 for adding sanitizing > > Why not use cmake arguments instead of environment variables for ENABLE_ADDRESS_SANITIZING, etc.? OK. I only use the travis config file now. This should...

> > +1 for adding sanitizing > > Why not use cmake arguments instead of environment variables for ENABLE_ADDRESS_SANITIZING, etc.? > > OK. I only use the travis config file...

**Interesting fact:** The address sanitizer detects a memory leak in `libclass_loader`, more precisely in `/opt/ros/melodic/include/class_loader/class_loader_core.hpp:205` (see [Travis log](https://api.travis-ci.org/v3/job/614764921/log.txt)). However, for some reason, maybe because the memory leak detection happens in...