A helpful primitive for moving chunks of data around inside a slice.
In particular, adding elements to the end of a Vec then moving them
somewhere else, as a way to do efficient multiple-insert. (There's
drain for efficient block-remove, but no easy way to block-insert.)
Talk with another example: <https://youtu.be/qH6sSOr-yk8?t=560>
Not being an enum improves ergonomics, especially since NonEmpty could be Empty. It can still be iterable without an extra "done" bit by making the range have !(start <= end), which is even possible without changing the Step trait.
Implements RFC 1980
Correct some stability versions
These were found by running tidy on stable versions of rust and finding
features stabilised with the wrong version numbers.
Add lint for unused macros
Addresses parts of #34938, to add a lint for unused macros.
We now output warnings by default when we encounter a macro that we didn't use for expansion.
Issues to be resolved before this PR is ready for merge:
- [x] fix the NodeId issue described above
- [x] remove all unused macros from rustc and the libraries or set `#[allow(unused_macros)]` next to them if they should be kept for some reason. This is needed for successful boostrap and bors to accept the PR. -> #41934
- [x] ~~implement the full extent of #34938, that means the macro match arm checking as well.~~ *let's not do this for now*
fix confusion about parts required for float formatting
The documentation for flt2dec doesn't match up with the actual
implementation, so fix the documentation to align with reality.
Presumably due to the mismatch, the formatting code for floats in
std::fmt can use correspondingly shorter arrays in some places, so fix
those places up as well.
Fixes#41304.
Added generic example of std::ops::Add in doc comments
We discussed on IRC how the std::ops examples were potentially missing examples using generics. This PR adds an example to std::ops::Add that shows the use of a generic type T. I'm not sure this is ready for merge as I think the two examples now make the documentation a bit verbose, but I think it's a good starting point. I'd love to hear others thoughts on this. This is in relation to the last item in issue #29365.
remove the #[inline] attribute from drop_in_place
Apparently LLVM has exponential code growth while inlining landing pads
if that attribute is present.
Fixes#41696.
beta-nominating because regression.
r? @eddyb
Removes unused macros from:
* libcore
* libcollections
The last use of these two macros was removed in commit
b64c9d5670
when the char_range_at_reverse function was been removed.
* librustc_errors
Their last use was removed by commits
2f2c3e1783
and 11dc974a38.
* libsyntax_ext
* librustc_trans
Also, put the otry macro in back/msvc/mod.rs under the
same cfg argument as the places that use it.
Added blank lines around example
Added comment to Add example referencing the Output type
Removed whitespace from lines 272 and 273
Removed Debug derivation from Add examples
Added Debug derivation
Make [u8]::reverse() 5x faster
Since LLVM doesn't vectorize the loop for us, do unaligned reads of a larger type and use LLVM's bswap intrinsic to do the reversing of the actual bytes. cfg!-restricted to x86 and x86_64, as I assume it wouldn't help on things like ARMv5.
Also makes [u16]::reverse() a more modest 1.5x faster by loading/storing u32 and swapping the u16s with ROT16.
Thank you ptr::*_unaligned for making this easy :)
Benchmark results (from my i5-2500K):
```text
# Before
test slice::reverse_u8 ... bench: 273,836 ns/iter (+/- 15,592) = 3829 MB/s
test slice::reverse_u16 ... bench: 139,793 ns/iter (+/- 17,748) = 7500 MB/s
test slice::reverse_u32 ... bench: 74,997 ns/iter (+/- 5,130) = 13981 MB/s
test slice::reverse_u64 ... bench: 47,452 ns/iter (+/- 2,213) = 22097 MB/s
# After
test slice::reverse_u8 ... bench: 52,170 ns/iter (+/- 3,962) = 20099 MB/s
test slice::reverse_u16 ... bench: 93,330 ns/iter (+/- 4,412) = 11235 MB/s
test slice::reverse_u32 ... bench: 74,731 ns/iter (+/- 1,425) = 14031 MB/s
test slice::reverse_u64 ... bench: 47,556 ns/iter (+/- 3,025) = 22049 MB/s
```
If you're curious about the assembly, instead of doing this
```
movzx eax, byte ptr [rdi]
movzx ecx, byte ptr [rsi]
mov byte ptr [rdi], cl
mov byte ptr [rsi], al
```
it does this
```
mov rax, qword ptr [rdx]
mov rbx, qword ptr [r11 + rcx - 8]
bswap rbx
mov qword ptr [rdx], rbx
bswap rax
mov qword ptr [r11 + rcx - 8], rax
```
The documentation for flt2dec doesn't match up with the actual
implementation, so fix the documentation to align with reality.
Presumably due to the mismatch, the formatting code for floats in
std::fmt can use correspondingly shorter arrays in some places, so fix
those places up as well.
Fixes#41304.
refactor NonZero, Shared, and Unique APIs
Major difference is that I removed Deref impls, as apparently LLVM has
trouble maintaining metadata with a `&ptr -> &ptr` API. This was cited
as a blocker for ever stabilizing this API. It wasn't that ergonomic
anyway.
* Added `get` to NonZero to replace Deref impl
* Added `ptr` getter to Shared/Unique to replace Deref impl
* Added Unique's `get` and `get_mut` conveniences to Shared
* Deprecated `as_mut_ptr` on Shared in favour of `ptr`
Note that Shared used to primarily expose only `*const` but there isn't
a good justification for that, so I made it `*mut`.
Since LLVM doesn't vectorize the loop for us, do unaligned reads
of a larger type and use LLVM's bswap intrinsic to do the
reversing of the actual bytes. cfg!-restricted to x86 and
x86_64, as I assume it wouldn't help on things like ARMv5.
Also makes [u16]::reverse() a more modest 1.5x faster by
loading/storing u32 and swapping the u16s with ROT16.
Thank you ptr::*_unaligned for making this easy :)
Major difference is that I removed Deref impls, as apparently LLVM has
trouble maintaining metadata with a `&ptr -> &ptr` API. This was cited
as a blocker for ever stabilizing this API. It wasn't that ergonomic
anyway.
* Added `get` to NonZero to replace Deref impl
* Added `as_ptr` to Shared/Unique to replace Deref impl
* Added Unique's `as_ref` and `as_mut` conveniences to Shared
* Added `::empty()` convenience constructor for Unique/Shared
* Deprecated `as_mut_ptr` on Shared in favour of `as_ptr`
* Improved documentation of types
Note that Shared now only refers to *mut, and not *const
process:exit -> process::exit in mem::forget docs
The documentation in mem::forget says "...or call `process:exit`..."
instead of `process::exit`.
r? @steveklabnik
reduce stack requirements for floating-point formatting
Doing this speeds up float formatting by ~10% or so, and also makes the formatting code more suitable for embedded environments where stack space is at a premium.
