Update to the latest bindgen #3
Conversation
|
It looks like it's failing on Travis CI, but I can't get the log to load to see why… |
|
(Oh, and thanks!) |
|
I wonder if it has anything to do with the S3 outage today? |
|
Oh, I dunno. I'll try again later I guess. |
|
I think the outage is resolved at this point. Maybe just kick it off again? |
|
Looks like it you go to the requests section of Travis it shows that it's
building correctly.
…On Feb 28, 2017 3:55 PM, "Jesse A. Tov" ***@***.***> wrote:
Oh, I dunno. I'll try again later I guess.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#3 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AC6qEyaZikSMkWe8QiW8VI9Ktwe_4GKFks5rhLQLgaJpZM4MOyAp>
.
|
|
I still can’t see it on Travis, but it works for me on two different machines. I'll take it. Thanks! |
|
Sweet! |
|
So looks like on my localhost the only thing failing is max_align_t: ---- generated::bindgen_test_layout_max_align_t stdout ---- |
|
It’s a generated test, and the generated code is failing it. The generated code defines a struct #[repr(C)]
pub struct max_align_t {
pub __clang_max_align_nonce1: ::std::os::raw::c_longlong,
pub __clang_max_align_nonce2: f64,
}The test is then that the size of this struct is 32 bytes. But it's failing and getting 16? Seems like 16 is right, though. |
|
Yeah I would peg that as 16. 8 for c_longlong and 8 for f64
…On Wed, Mar 1, 2017 at 7:25 AM, Jesse A. Tov ***@***.***> wrote:
It’s a generated test, and the generated code is failing it. The generated
code defines a struct
#[repr(C)]pub struct max_align_t {
pub __clang_max_align_nonce1: ::std::os::raw::c_longlong,
pub __clang_max_align_nonce2: f64,
}
The test is then that the size of this struct is 32 bytes. But it's
failing and getting 16? Seems like 16 is right, though.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#3 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AC6qE16rXemmXkqL_2y_PTU5U9AEP3CJks5rhY3kgaJpZM4MOyAp>
.
|
|
The problem seems to be that the C source has typedef struct {
long double ld;
} max_align;I’ll put together a bug report. |
|
Nice find!
…On Wed, Mar 1, 2017 at 9:46 AM, Jesse A. Tov ***@***.***> wrote:
The problem seems to be that the C source has long double as the type of
the second field, and Rust thinks that has to be aligned at 16. But it's
getting translated to f64. I’ve found a minimal example for bindgen input:
typedef struct {
long double ld;
} max_align;
I’ll put together a bug report.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#3 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AC6qE5ywY0R9-oUPX3jLla1LyRRcmNsSks5rha79gaJpZM4MOyAp>
.
|
|
Looks like they may know about it, and this may be the fix?: rust-lang/rust-bindgen#468 |
|
Not the fix; possibly the cause? |
This updates libffi to the latest bindgen release. This fixed my build problem with size_t not found. 👍 I have rustfmt automatically run when I save a file which changed a few things in build.rs. If you don't like that I can revert those.