Use default CXX, use boost::placeholders::_1 instead of _1

[lucasw]

I'm curious if std::bind with std::placeholders also works (and will likely try it), but here I kept using boost for a smaller change.

[SteveMacenski]

Is there any benefit to this? Mixing and matching doesn't seem like a great policy to maintain

[lucasw]

This PR as-is doesn't involve any mixing and matching or changeover to std::bind and is purely boost::bind, I'm only wondering about std::bind vs. boost::bind in my comment above but not proposing it here.

[lucasw]

The practice of declaring the Bind placeholders (_1, _2, ...) in the global namespace is deprecated.

Then use the qualified names like boost::placeholders::_1 directly, or locally do using namespace boost::placeholders

There's mention of including <boost/bind/bind.hpp> instead of <boost/bind.hpp> but I didn't see any existing bind.hpp includes so didn't make any changes there. I'd have to look further but I think the old include that defines the macro for preserving the old global namespace _1/_2 is removed entirely in later versions of boost, and compiling against those is what is motivating this PR.

https://stackoverflow.com/questions/53203970/why-boostbind-insists-pulling-boostplaceholders-into-global-namespace

[ayrton04]

Thanks for the PR. I think it's the right call to be explicit and use, e.g., boost::placeholders::_1 instead of _1. But with noetic being end-of-the-line for ROS 1, I'm not sure we'll gain much if we later try to move to the STL equivalents (though I obviously prefer them in general).

[ayrton04]

Why did this need to be removed?

[lucasw]

It's a log4cxx 0.12 issue that results in

/usr/include/log4cxx/boost-std-configuration.h:10:18: error: ‘shared_mutex’ in namespace ‘std’ does not name a type
   10 |     typedef std::shared_mutex shared_mutex;
      |                  ^~~~~~~~~~~~```

when building with < C++17.

Ubuntu 20.04 noetic already defaults to C++14 and has an older log4cxx, newer systems with newer log4cxx default to C++17, and no need to support older systems in this noetic branch, so removing it seems cleanest, but I can see setting it to 17 instead. Also putting it into a separate commit in this PR, or a separate PR (or don't accept if at all if you prefer, it's easy to maintain a fork with one line of difference but much nicer for others with the newer log4cxx not to have to).

[ayrton04]

I'm fine with the branch assuming whatever standard is adopted by the ROS release in question. Thanks for the info!