Maybe makes sense to use clang-5, it has a number of new settings.
Should we aim already for 5 even though xenial is shipping 3.8?
I don't think it matters what is shipped with a certain distro. This is auxiliary tooling only for devs and they can make this extra step of adding a PPA and installing 5.0 alongside with their standard clang.
Plus, in my dream world we have all the code base reformatted and a special Travis task that runs each pull request through formatter and fails if code does not match, outputting a diff that a contributor can apply to his code without any need to install whatever version of clang we happen to be using on Travis.
Maybe makes sense to use clang-5, it has a number of new settings.
I don't think it matters what is shipped with a certain distro. This is auxiliary tooling only for devs and they can make this extra step of adding a PPA and installing 5.0 alongside with their standard clang.
Plus, in my dream world we have all the code base reformatted and a special Travis task that runs each pull request through formatter and fails if code does not match, outputting a diff that a contributor can apply to his code without any need to install whatever version of clang we happen to be using on Travis.