code style only

[dirk-thomas]

Connects to ros2/ros2#21.

[esteve]

+1

[wjwwood]

I would not add a space here. The lack of a space here will not be a problem on any C++ compiler we will support.

[dirk-thomas]

The latest version of uncrustify (0.61) introduces this configuration option for it: sp_permit_cpp11_shift

I updated the config and all PRs where necessary.

[wjwwood]

This one also doesn't look right, I would have expected:

create_service(
    std::string service_name,
    std::function<void (const std::shared_ptr<typename ServiceT::Request> &,
        std::shared_ptr<typename ServiceT::Response>&)> callback,
    rclcpp::callback_group::CallbackGroup::SharedPtr group=nullptr);

Or...

create_service(
    std::string service_name,
    std::function<void (
        const std::shared_ptr<typename ServiceT::Request> &,
        std::shared_ptr<typename ServiceT::Response>&)> callback,
    rclcpp::callback_group::CallbackGroup::SharedPtr group=nullptr);

Regardless of what I expected, the result from the linter is pretty confusing to me, since parameters to nested items end up on the same line.

[dirk-thomas]

uncrustify does not change the way the first argument is positioned. It only decides the indentation of the wrapped arguments here. We have to manually wrap the first argument in this case: 9e735e3

[wjwwood]

What about the whitespace before these? I would personally prefer it to be horizontally aligned with the class keyword. I don't like having leading whitespace which is not a multiple of 2 and does not horizontally align with anything above or below it.

I would propose to have no extra space before the access modifiers.

However, that's just my opinion, what do you guys think?

[dirk-thomas]

Then it looks good as-is.

[tfoote]

As is seems fine, with the standard spacing.

[wjwwood]

With the class definitions styled as we discussed in the meeting, +1.

[dirk-thomas]

@wjwwood Position of colon in class initialization addressed: c6b8fb2

[tfoote]

lgtm