std: Enforce Unicode in fmt::Writer
This commit is an implementation of [RFC 526][rfc] which is a change to alter the definition of the old `fmt::FormatWriter`. The new trait, renamed to `Writer`, now only exposes one method `write_str` in order to guarantee that all implementations of the formatting traits can only produce valid Unicode. [rfc]: https://github.com/rust-lang/rfcs/blob/master/text/0526-fmt-text-writer.md One of the primary improvements of this patch is the performance of the `.to_string()` method by avoiding an almost-always redundant UTF-8 check. This is a breaking change due to the renaming of the trait as well as the loss of the `write` method, but migration paths should be relatively easy: * All usage of `write` should move to `write_str`. If truly binary data was being written in an implementation of `Show`, then it will need to use a different trait or an altogether different code path. * All usage of `write!` should continue to work as-is with no modifications. * All usage of `Show` where implementations just delegate to another should continue to work as-is. [breaking-change] Closes #20352
This commit is contained in:
parent
cd614164e6
commit
e423fcf0e0
25 changed files with 320 additions and 359 deletions
|
|
@ -1662,6 +1662,7 @@ mod test {
|
|||
use serialize;
|
||||
use codemap::*;
|
||||
use super::*;
|
||||
use std::fmt;
|
||||
|
||||
// are ASTs encodable?
|
||||
#[test]
|
||||
|
|
@ -1687,6 +1688,6 @@ mod test {
|
|||
exported_macros: Vec::new(),
|
||||
};
|
||||
// doesn't matter which encoder we use....
|
||||
let _f = &e as &serialize::Encodable<json::Encoder, io::IoError>;
|
||||
let _f = &e as &serialize::Encodable<json::Encoder, fmt::Error>;
|
||||
}
|
||||
}
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue