Better cgo support

[AlekSi]

This can be made smaller:

 buf: ([1024]foo._Ctype_char) {
  (foo._Ctype_char) 115,
  (foo._Ctype_char) 100,
  (foo._Ctype_char) 102,
  (foo._Ctype_char) 115,
  (foo._Ctype_char) 100,
  (foo._Ctype_char) 102,
  (foo._Ctype_char) 115,
  (foo._Ctype_char) 100,
  (foo._Ctype_char) 102,
  (foo._Ctype_char) 115,
  (foo._Ctype_char) 0,
  (foo._Ctype_char) 0,
  (foo._Ctype_char) 0,
  ...

[davecgh]

This is a reasonable request. I haven't spent much time dealing with cgo types under reflection, so I'm not sure how much is possible, but I'll take a look.

[davecgh]

After looking into this a bit tonight, I see that C.char is considered signed as an int8 under reflection. As a result, the hexdump style functionality is bypassed since it only works with uint8.

There are a couple of ways to handle this. One way is to allow spew to hexdump int8 slices as well as uint8 slices, but I don't think that's really desirable for most cases. Another option that is a bit hackish is to look at the textual type (versus the reflected type) specifically to detect something like the pattern ^.*\._Ctype_char$ and hexdump it from there after a bit of casting.

Any thoughts?

[AlekSi]

I think looking for ^.*\._Ctype_char$ is okay if user is opted in for it.

[davecgh]

This feature, including new tests and an updated test coverage report has been implemented as of commit 1fe9f5c.