add BOOST_MATH_DISABLE_FLOAT128

[mikaelarguedas]

Description

This fixes #504 the compilation issue on Ubuntu Zesty and Debian Stretch.
This change is only for lunar and doesn't need to be cherry-picked into other branches

Note: I settled for this solution because it seemed the least intrusive to me. There are several other fixes that could be considered:

• By changing the -std=c++11 flag to -std=gnu++11 in the add_compile_option() call but I preferred to keep the flags consistent across all packages of the moveit repository.
• By using (but needs to bump cmake minimum version to 3.0.2):

set(CMAKE_CXX_STANDARD 11)`
set(CMAKE_CXX_STANDARD_REQUIRED ON)

Checklist

• Required: Code is auto formatted using clang-format: didn't change code
• Extended the tutorials / documentation, if necessary reference: doesn't apply
• Include a screenshot if changing a GUI: doesn't apply
• Optional: Created tests, which fail without this PR reference: doesn't apply
• Optional: Decide if this should be cherry-picked to other current ROS branches (Indigo, Jade, Kinetic): not needed

Thank you!

[130s]

Thanks for the work @mikaelarguedas. I've asked a few potential reviewers. I guess with minimum one approval we can move forward.

I'm tentatively testing on my fork 130s#1 with industrial_ci that allows prerelease test on Lunar and Zesty.

To Maintainers: #504 is for ROS Lunar, and this is targetted to kinetic-devel and Travis passed. So this change can go into kinetic-devel and no need to create lunar-devel (for now)?

[mikaelarguedas]

Thanks @130s
Just to clarify this does fix the build of moveit_ros_visualization on Zesty and Stretch but doesn't fix the current moveit_setup_assistant failure on Yakkety (the same failures will still exist on Zesty and Stretch once this is merged). Should I open an issue to track the moveit_setup_assistant failure?

[v4hn]

Should I open an issue to track the moveit_setup_assistant failure?

Yes, please. At first glance, this seems to relate to problems with the urdf package in yakkety.

So this change can go into kinetic-devel and no need to create lunar-devel (for now)?

Yes.

@mikaelarguedas: Please add a short comment above the definition to explain that a boost header included through ogre doesn't compile with -std=c++11 without this.

I didn't test this myself, but it makes sense that this should resolve the compile error.

[mikaelarguedas]

Should I open an issue to track the moveit_setup_assistant failure?

Yes, please. At first glance, this seems to relate to problems with the urdf package in yakkety.

Ticketed #506

@mikaelarguedas: Please add a short comment above the definition to explain that a boost header included through ogre doesn't compile with -std=c++11 without this.

Done 162b83a, feel free to push directly to this branch if you want to improve the wording

I didn't test this myself, but it makes sense that this should resolve the compile error.

I ran prerelease tests for all platforms (Xenial, Yakkety, Zesty and Stretch) and it compiled without error.

[v4hn]

The wording is fine.
Thanks a lot for your support @mikaelarguedas !