Use c_schar instead of c_char

[pyropeter]

The signedness of the C type 'char' is implementation defined. The
rust type c_schar exists for this reason. Use it.

[nox]

The type c_schar is always signed, whereas c_char's signedness is implementation-defined. Where did you read that c_schar was to be used for the equivalent of C's char type?

[pyropeter]

I didn't. But bindgen uses c_char for clangs CXType_SChar, which is obviously a signed char.

[nox]

That sounds wrong indeed. It seems like IntKind should have an additional variant SChar.

[emilio]

That sounds wrong indeed. It seems like IntKind should have an additional variant SChar.

That's right. Please update the PR to do that instead, thanks!

[pyropeter]

Why do you want to have an additional variant SChar in IntKind? As far as I understand, clang takes care of the ugly char type and only uses signed char and unsigned char afterwards, so there really is no need for a variant representing char, is there?

One could of course rename the Char variant to SChar, but that would break the naming scheme of your variants. In that case I would prefer to add an explicit S prefix to all signed variants. (Like it is done in clang)

[emilio]

Why do you want to have an additional variant SChar in IntKind? As far as I understand, clang takes care of the ugly char type and only uses signed char and unsigned char afterwards, so there really is no need for a variant representing char, is there?

Whoops, you're totally right here. My memory failed me and I thought that there was a CXType_Char, but apparently there isn't, so this looks fine to me.

@bors-servo r+

[bors-servo]

📌 Commit 4f1e3da has been approved by emilio

[bors-servo]

⌛ Testing commit 4f1e3da with merge 1a4b8e6...

[bors-servo]

☀️ Test successful - status-travis
Approved by: emilio
Pushing 1a4b8e6 to master...