warning: assigning the result of `Clone::clone()` may be inefficient
--> impl/src/ast.rs:85:21
|
85 | *display = attrs.display.clone();
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ help: use `clone_from()`: `display.clone_from(&attrs.display)`
|
= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#assigning_clones
= note: `#[warn(clippy::assigning_clones)]` on by default
warning: assigning the result of `Clone::clone()` may be inefficient
--> impl/src/expand.rs:158:9
|
158 | display_implied_bounds = display.implied_bounds.clone();
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ help: use `clone_from()`: `display_implied_bounds.clone_from(&display.implied_bounds)`
|
= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#assigning_clones
warning: assigning the result of `Clone::clone()` may be inefficient
--> impl/src/expand.rs:402:21
|
402 | display_implied_bounds = display.implied_bounds.clone();
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ help: use `clone_from()`: `display_implied_bounds.clone_from(&display.implied_bounds)`
|
= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#assigning_clones
Prefer using core crate over std in macro expansions wherever a symbol has
a stable equivalent in core. This makes it easier to eventually support
no_std in the future once error_in_core stabilises.
Prior to PR 251, 2 traits were imported (DisplayAsDisplay, PathAsDisplay),
and in some cases only 1 would be used. Now it's 1 trait and is always
used.
Rather than having separate traits implementing as_display method,
replace DisplayAsDisplay and PathAsDisplay traits with a AsDisplay
trait. The difference between those two traits is in the result
returned by the as_display method. With AsDisplay trait this is
captured by an associated type.
The main motivation for the change is making it simpler to support
no_std builds in the future. Previously, PathAsDisplay would have to
be handled specially in such builds on the side of macro expansion.
Now, thiserror-impl doesn’t need to be aware of any complications
around AsDisplay since they are all contained in thiserror crate.
When a macro generates part of the derive input, the call-site hygiene
may be different than the hygiene of a field. Therefore, we need to
be sure to use the same hygiene information for any identifiers we
generate, instead of relying on the hygiene from a particular span
via `quote_spanned!`
