On 22 June 2017 at 13:41, Marc-André Lureau <marcandre.lur...@redhat.com> wrote: > Various refactring questions on previously sent series prompted me to > look at coccinelle to automate some changes again. Alas, semantic > patches are not so easy to express for me, cocci doesn't catch all > cases, is quite slow, and it doesn't seem possible to evaluate > expressions to check if E == E-1 or if E is pow2 for example. > > I started looking at clang-tidy > (http://clang.llvm.org/extra/clang-tidy/) as an alternative to do some > refactoring.
Oh, very nice. I've had "investigate clang-tidy" in my todo list for ages so it's nice that somebody has done it for me ;-) I think it would be particularly interesting to look at whether we can use this infrastructure to automatically detect some kinds of bug, rather than merely stylistic issues. (As a random example, could you detect calls to coroutine_fn functions from normal functions?) thanks -- PMM
