By definition, acceptance tests should validate the functionality we expose to developers using the server and client libraries. Like the code in examples/, they also act as examples/documentation of correct usage and expected behavior of our offered APIs. Also, like the code in examples/, it doesn't make sense to showcase the API and then not make it available to developers (see previous discussion at https://code.launchpad.net/~vanvugt/mir/privatise-underused-headers/+merge/232865).
This bug is a matter of opinion only to the degree that the definition of the acceptance tests is a matter of opinion (I think it's not). It's a another matter altogether whether this bug should be fixed at the moment or not, for which there are indeed different opinions.
By definition, acceptance tests should validate the functionality we expose to developers using the server and client libraries. Like the code in examples/, they also act as examples/ documentation of correct usage and expected behavior of our offered APIs. Also, like the code in examples/, it doesn't make sense to showcase the API and then not make it available to developers (see previous discussion at https:/ /code.launchpad .net/~vanvugt/ mir/privatise- underused- headers/ +merge/ 232865).
This bug is a matter of opinion only to the degree that the definition of the acceptance tests is a matter of opinion (I think it's not). It's a another matter altogether whether this bug should be fixed at the moment or not, for which there are indeed different opinions.