Hi James, Someone would need to propose to auto-format everything with clang-format and convince the community that while clang-format will never be perfect, it generates a systematic, sane and just as readable codebase. If we can reach consensus on that, we could add it as a commit hook and no one will worry about formatting.
I dreamed of the world described above, but at least at the time clang-format was still learning about some of the C++11 constructs (especially around lambdas), so it seemed a bit too far fetched. My goal with clang-format currently is for people to have it integrated in their editors to help them get 80, 90% of the way there with single key-stroke, then follow-up with necessary minor edits. We've been converging from both sides to get that percentage higher (new features in clang-format + modifying our style), and we'll continue to make such efforts where it makes sense. MPark On 18 March 2016 at 12:45, James Peach <jor...@gmail.com> wrote: > > > On Mar 17, 2016, at 10:41 AM, Yong Tang <yong.tang.git...@outlook.com> > wrote: > > > > Hi All > > > > > > This email is to announce that the default configuration and the > recommended version of the clang-format is being upgraded to 3.8 (from 3.5) > in mesos. > > > > > > In clang-format-3.8, the newly introduced option "AlignAfterOpenBracket: > AlwaysBreak" closes the largest gap between ClangFormat and the style guide > in mesos. It avoids "jaggedness" in function calls and is worth migrating > for. > > > > > > Along with the changes in clang-format configuration > (support/clang-format), the documentation (docs/clang-format.md) is also > going to be updated to reflect changes in version and the recommended > installation process. > > > > > > More details about this upgrade could be found in MESOS-4906 ( > https://issues.apache.org/jira/browse/MESOS-4906). By the way, thanks > Michael for the help on this issue. > > This sounds really promising. Is the plan to auto-format everything with > clang-format?