Since it isn't possible to disable linkage of just GCC startup objects, we now need logic for finding libc installation directory and copying the required startup files (e.g. crt2.o) to rustlib directory.
Bonus change: use the `-nodefaultlibs` flag on Windows, thus paving the way to direct linker invocation.
Rather than injecting a local `_Unwind_Resume` into the current translation unit,
just replace `resume` instruction with a direct call the the `eh_unwind_resume` lang item.
This is likely to be more robust in the face of future LLVM changes, and also allows us to delegate
work back to libgcc's `_Unwind_Resume`.
When going through the docs, it is not clear that binary files cannot be tested. Additionally, it is hard to find the proper structure of a Rust crate and it took me several hours of looking through the docs to find the crates and modules section. I think we can link to it from here and it will be beneficial to those who are coming to the language.
While working on #28711 I found out that
* src/libcoretest/clone.rs
* src/libcoretest/fmt/float.rs
* src/libcoretest/intrinsics.rs
are not used. I am not sure if removing them is the right way to go. If it makes sense to keep (and fix and enable) them, I would be happy to update the PR.
This PR solves the following issues (or at least help users to understand the problem):
```Rust
#![crate_name = "b"]
#![crate_type = "rlib"]
pub fn his_function_has_a_very_long_name_and_should_make_cargo_doc_explodes_because_it_will_want_to_make_a_filename_with_it_in_excess_of_the_max_filename_length_for_most_filesystem_this_is_not_yet_long_enough_i_need_moreis_function_has_a_very_long_name_and_should_make_cargo_doc_explodes_because_it_will_want_to_make_a_filename_with_it_in_excess_of_the_max_filename_length_for_most_filesystem_this_is_not_yet_long_enough_i_need_more_() {}
```
```Rust
#![crate_name = "b"]
#![crate_type = "rlib"]
pub struct his_function_has_a_very_long_name_and_should_make_cargo_doc_explodes_because_it_will_want_to_make_a_filename_with_it_in_excess_of_the_max_filename_length_for_most_filesystem_this_is_not_yet_long_enough_i_need_moreis_function_has_a_very_long_name_and_should_make_cargo_doc_explodes_because_it_will_want_to_make_a_filename_with_it_in_excess_of_the_max_filename_length_for_most_filesystem_this_is_not_yet_long_enough_i_need_more_;
```
For the maximum filename length chosen, @gkoz gave me [this link](http://unix.stackexchange.com/a/32834).
This change has two consequences:
1. It makes `Arc<T>` and `Rc<T>` covariant in `T`.
2. It causes the compiler to reject code that was unsound with respect
to dropck. See compile-fail/issue-29106.rs for an example of code that
no longer compiles. Because of this, this is a [breaking-change].
Fixes#29037.
Fixes#29106.
Stricter checking of stability attributes + enforcement of their invariants at compile time
(+ removed dead file librustc_front/attr.rs)
I intended to enforce use of `reason` for unstable items as well (it normally presents for new items), but it turned out too intrusive, many older unstable items don't have `reason`s.
r? @aturon
I'm studying how stability works and do some refactoring along the way, so it's probably not the last PR.
The text says it's a vector of floats, but the code actually uses a vector of integers. The type of the Vec doesn't really matter, so I just cut it from the text.
Fixes issue #29077.
There **are** API changing guidelines in the RFCs, so it might be prudent to put those there. But I'm leaving that up to another person. This PR just removes bad documentation.
When going through the docs, it is not clear that binary files cannot be tested. Additionally, it is hard to find the proper structure of a Rust crate and it took me several hours of looking through the docs to find the crates and modules section. I think we can link to it from here and it will be beneficial to those who are coming to the language.
The implementation for `into_inner` was a bit more complex than I had hoped for---is there any simpler, less unsafe way of getting around the fact that one can't move out of a `Drop` struct?
See #28968 and rust-lang/rfcs#1269 .
The text says it's a vector of floats, but the code actually uses a vector of integers. The type of the Vec doesn't really matter, so I just cut it from the text.
This commit adds `#[derive(Clone)]` to `std::fs::Metadata`, making that struct cloneable. Although the exact contents of that struct differ between OSes, they all have it contain only value types, meaning that the data can be re-used without repercussions.
It also adds `#[derive(Clone)]` to every type used by that struct across all OSes, including the various Unix `stat` structs and Windows's `WIN32_FILE_ATTRIBUTE_DATA`.
This stems from my comment here: https://github.com/rust-lang/rfcs/issues/939#issuecomment-140524439
