Re: [RFC/PATCH] attr: map builtin userdiff drivers to well-known extensions

On Thu, Aug 25, 2011 at 07:44:25PM -0400, Eric Sunshine wrote:

quoted

How well do our cpp patterns do with header files? I imagine they're
better than the default, but I don't think I've ever really tried
anything tricky.

I scanned through a number of revisions for one of my long-running
C++ projects comparing the diff of header files with and without "*.h
diff=cpp". In some header files in this project, the oft-used C++
keywords public:, protected:, and private: appear at start-of-line.
In such cases, the default diff emits a less-than-useful hunk header:

    @@ -19,8 +19,8 @@ public:

whereas, "diff=cpp" emits:

    @@ -19,8 +19,8 @@ class Foobar

Thanks. My C++ is so rusty that I didn't think immediately of how often
those keywords appear in header files. Also, code in inline
functions in either C or C++ will be found in header files. So I think
defaulting *.h and *.hpp to cpp is sensible.

-Peff