Update anonymous box list

[bholley]

These have drifted since we introduced them. Adding a few other misc commits while I'm at it.

---

This change is

[highfive]

Warning

• These commits modify style code, but no tests are modified. Please consider adding a test!

[bholley]

r? @emilio

[emilio]

@bors-servo: r+

[bors-servo]

📌 Commit 2784f44 has been approved by emilio

[bors-servo]

⌛ Testing commit 2784f44 with merge c1409af...

[bors-servo]

☀️ Test successful - android, arm32, arm64, linux-dev, linux-rel, mac-dev-unit, mac-rel-css, mac-rel-wpt, windows-dev

[Manishearth]

The llvm_stable thing breaks the binding generation for me. If I pass it llvm38, I get this error:

  "_clang_Type_getNamedType", referenced from:
      bindgen::clang::Type::named::h017a16ee858af8af in libbindgen.rlib(bindgen.0.o)
  "_clang_Cursor_isFunctionInlined", referenced from:
      bindgen::clang::Cursor::is_inlined_function::hbf3ecb6d1f288ea8 in libbindgen.rlib(bindgen.0.o)

If I pass it my local llvm trunk build, it works but makes changes like:

 -- a/ports/geckolib/gecko_bindings/bindings.rs
 ++ b/ports/geckolib/gecko_bindings/bindings.rs
@@ -147,7 +147,7 @@ use structs::nsIAtom;

 pub type RawGeckoNode = nsINode;
 pub enum Element { }
- pub type RawGeckoElement = Element;
+ pub type RawGeckoElement = ::std::os::raw::c_void;
 pub type RawGeckoDocument = nsIDocument;
 pub enum ServoNodeData { }
 pub enum ServoComputedValues { }

(all mozilla:: things get replaced with void)

[bholley]

Hm, _clang_Cursor_isFunctionInlined and friends are LLVM39 symbols. The whole point of llvm_stable is that it avoids requiring llvm39 symbols, because we need to be on LLVM38 to have any chance at all of making LLVM a build-time requirement for Firefox.

Can you debug why llvm_stable seems to be having the opposite effect as intended?

[Manishearth]

I'll try, sure 😄

a clean build of bindgen didn't seem to change anything but I should try this again.

[Manishearth]

Works now. I cleaned it again, reset env vars, and it's fine.