feat(types) Avoid allocating a Vec when calling FunctionType::new
#2141
Conversation
|
bors try |
|
bors r+ |
2141: feat(types) Avoid allocating a `Vec` when calling `FunctionType::new` r=Hywan a=Hywan # Description Sequel of #2036. We can go further by defining `Params` and `Results` of `FunctionType::new` as `Into<Box<[Type]>>` directly rather than `Into<Vec<Type>>`. It simplifies the code: `params.into()` rather than `params.into().into_boxed_slice()`. I assume it doesn't allocate an intermediate vector. Since `From<Box<[T]>>` is implemented for `Vec<T>`, I reckon it's not a BC break. # Review - [ ] ~Add a short description of the the change to the CHANGELOG.md file~ not necessary Co-authored-by: Ivan Enderlin <[email protected]>
|
Build failed: |
|
The CI failure has nothing related to this PR. Weird. |
There was a problem hiding this comment.
Since From<Box<[T]>> is implemented for Vec, I reckon it's not a BC break.
From<Box<[T]>> for Vec<T> means that we can go from Box<T> -> Vec<T>, did you mean Into<Box<[T]>> for Vec<T>? If so that's better but I think it's still a breaking change because for a given type GT, if GT implements Into<Vec<T>> but not Into<Box<[T]>> the caller code now has to do gt.into().into() (or because we call one of the intos they'd have to change from FunctionType::new(g, ...) to FunctionType::new(g.into(), ...)
I think this change is probably a good one but I do think it's probably a breaking change. We have lots of (generally very minor like this) breaking changes coming anyways though, we just need to decide when we want to ship a release to get them all in
| enum SecretOption { | ||
| None, | ||
| Some, | ||
| } | ||
|
|
||
| impl fmt::Debug for SecretOption { | ||
| fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { | ||
| match self { | ||
| Self::None => write!(f, "None"), | ||
| Self::Some => write!(f, "Some(...)"), | ||
| } | ||
| } | ||
| } | ||
|
|
There was a problem hiding this comment.
This change seems entirely unrelated. Not sure if you intended to commit it
There was a problem hiding this comment.
It's deadcode, raising a warning from rustc.
|
bors r+ |
Description
Sequel of #2036. We can go further by defining
ParamsandResultsofFunctionType::newasInto<Box<[Type]>>directly rather thanInto<Vec<Type>>. It simplifies the code:params.into()rather thanparams.into().into_boxed_slice(). I assume it doesn't allocate an intermediate vector.Since
From<Box<[T]>>is implemented forVec<T>, I reckon it's not a BC break.Review
Add a short description of the the change to the CHANGELOG.md filenot necessary