Fix calling convention propagation for function pointers. #549
Conversation
This sucks, but works. The full solution is a refactoring that needs more thought than the time I'm able to dedicate to bindgen right now, see the comment for details. Signed-off-by: Emilio Cobos Álvarez <[email protected]>
|
|
|
I need to figure out a way for this to be green in CI given we're "cross-compiling" the test. |
There was a problem hiding this comment.
So the CI issue is that we don't have support for something like
// bindgen-windows-only
in our test runner, right?
Since we generate different bindings with MSVC mangling anyways, it would make sense to have different expectations dirs for different targets. Then we could add appveyor CI.
| typedef void (*my_fun_t)(int, int, int, int, | ||
| int, int, int, int, | ||
| int, int, int, int, | ||
| int, int, int, int); |
| // and https://github.com/rust-lang/rust/issues/40158 | ||
| // | ||
| // Note that copy is always derived, so we don't need to implement it. | ||
| impl CanDeriveDebug for FunctionSig { |
There was a problem hiding this comment.
At least extending these checks is obvious and straight forward.
Yeah, but then we need to add an easy way of updating test expectations across different targets. It seems just giving clang the right |
And I don't want nor have the time to debug it right now.
|
@bors-servo r=fitzgen |
|
📌 Commit 4c07a72 has been approved by |
Fix calling convention propagation for function pointers. This sucks, but works. The full solution is a refactoring that needs more thought than the time I'm able to dedicate to bindgen right now, see the comment for details. r? @fitzgen
|
☀️ Test successful - status-travis |
This sucks, but works.
The full solution is a refactoring that needs more thought than the time I'm
able to dedicate to bindgen right now, see the comment for details.
r? @fitzgen