Use original name when checking allowlist for anonymous enum variants #2006
Conversation
| )] | ||
|
|
||
| pub const RENAMED_MyVal: ::std::os::raw::c_uint = 0; | ||
| pub type _bindgen_ty_1 = ::std::os::raw::c_uint; |
There was a problem hiding this comment.
I'm a bit confused as to why this type is generated but not used in the definition of RENAMED_MyVal.
jethrogb
force-pushed
the
jb/anonymous-enum-variant-rename
branch
from
c80eb92
to
7f45e2d
Compare
I see, I didn't realize that existed. Looking at it briefly, the bindgen-integration is quite monolithic and it doesn't check the exact output. I think the expectation testing framework is much better, so I'd prefer to do it this way. But if you feel strongly about it I can move the tests to bindgen-integration. |
Ok, I'm fine with that, it's not too complicated anyways. Thanks! |
jethrogb
force-pushed
the
jb/anonymous-enum-variant-rename
branch
from
7f45e2d
to
2a546c8
Compare
152: Update bindgen r=raoulstrackx a=jethrogb I took some time this week to make the necessary bindgen changes (rust-lang/rust-bindgen#2004 rust-lang/rust-bindgen#2006 rust-lang/rust-bindgen#2007). This PR updates the bindgen build to use that. Breaking changes: * unnamed types are renamed * C-unions are now actual unions * [x] the field accessor functions that used to exist are easy enough to add back * bitfields are done differently now * some fn-ptrs are now unsafe * some fns and fn-ptrs now take const pointers as arguments * havege.c and timing.c are now not compiled on non-unix platforms, i.e. SGX (probably wasn't working properly anyway) Fixes #5 #14 #61 #72 #88 #121 Co-authored-by: Jethro Beekman <jethro@fortanix.com>
Prior to this PR, this header
in combination with
allowlist_var("^MyVal$")and an implementation ofParseCallbacks::enum_variant_namewould generate no code. This is because the renamed variant name is checked against the variable allowlist. All other allowlist checks are against the unrenamed name. This PR makes that consistent. This is a breaking change but I think the old behavior is clearly wrong.Additionally, I've added a way to test ParseCallbacks in CI.