codegen: Support repr(align) #1271
Conversation
|
|
| pub struct a { | ||
| pub b: ::std::os::raw::c_int, | ||
| pub c: ::std::os::raw::c_int, | ||
| pub __bindgen_align: [u64; 0usize], |
There was a problem hiding this comment.
Do we still need to generate this, then?
There was a problem hiding this comment.
Not really, but it doesn't hurt and avoids having the extra codepath and conditionals.
tests/headers/repr-align.hpp
Outdated
| @@ -0,0 +1,6 @@ | |||
| // bindgen-flags: --rust-target 1.24 -- -std=c++11 | |||
|
|
|||
| struct alignas(8) a { | |||
There was a problem hiding this comment.
Could you add another testcase with alignas(double)? And probably do so with 32bit Linux as target, given that's where the original issue is.
Also maybe add testcase for things when alignas is on field rather than type?
There was a problem hiding this comment.
Sure, will add. We don't run tests on linux32 though, so I won't do that, since it wouldn't test anything extra.
There was a problem hiding this comment.
I think adding -m32 should be enough?
|
@bors-servo r+ Although it would still be good if you have a test for |
|
📌 Commit 1923985 has been approved by |
codegen: Support repr(align) Fixes #917.
|
💔 Test failed - status-travis |
|
So turns out this feature is really |
|
☔ The latest upstream changes (presumably #1273) made this pull request unmergeable. Please resolve the merge conflicts. |
Otherwise we can't use repr(align) on stylo.
|
@bors-servo r=xidorn |
|
📌 Commit 853ae4d has been approved by |
codegen: Support repr(align) Fixes #917.
|
☀️ Test successful - status-travis |
Fixes #917.