Rework decoding of Boxes, Rcs, Arcs, arrays and enums (stack ov…

…erflow fix) (#426)

* Rework decoding of `Box`es, `Rc`s, `Arc`s and arrays

* Update changelog

* Bump version to 3.5.0

* Fix typo

* Reduce the size of an array in tests to make miri not run forever

* Make the `TypeInfo` match exhaustive

* Update dependencies

* Update benchmarks to criterion 0.4

* Update `BitVec` benchmarks too

* Fix the benchmarks for real this time

* Use `Layout::new` instead of `Layout::from_size_align`

Co-authored-by: Squirrel <giles@parity.io>

* Move blanket decoding of wrapper types into a default impl

* Fix stack overflow when decoding big nested enums

* Bump version of `parity-scale-codec-derive` to 3.5.0

* Move the crate-level docs into README and include README as docs

* Add a note regarding deserializing big arrays to the README

* Update the CHANGELOG again

* Remove one of the intermediate casts

This cast was a no-op anyway, but it's slightly confusing
as the length of the array in the pointer's type was incorrect,
so let's just remove it.

* README cleanups

* Add `rust` markers and make indentation consistent in README snippets

* Make `max-encoded-len` feature not force-enable `parity-scale-codec-derive`

* Reexport derive macros not when `parity-scale-codec-derive` is enabled but when `derive` is

* Update CHANGELOG again

* Fix `parity-scale-codec-derive` tests compilation

* Update `serde`

* Update comment in `src/codec.rs`

Co-authored-by: Bastian Köcher <git@kchr.de>

* Update comment in `src/codec.rs`

Co-authored-by: Roman <r.proskuryakoff@gmail.com>

* Move the version bounds for `parity-scale-codec-derive`

* Remove `DecodeContext`

* Move the array byte size calculations to compile-time

* Add a TODO about `Box::assume_init`

* Move `DecodeFinished` into its own module

* Add a TODO about `MaybeUninit::slice_assume_init_mut`

* Update the changelog

* `std` -> `core` so that no_std compiles

---------

Co-authored-by: Squirrel <giles@parity.io>
Co-authored-by: Bastian Köcher <git@kchr.de>
Co-authored-by: Roman <r.proskuryakoff@gmail.com>

CHANGELOG.md

@@ -5,6 +5,26 @@ All notable changes to this crate are documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this crate adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [3.6.0] - 2023-06-15
+
+### Added
+
+- Added `Decode::decode_into` to allow deserializing into unitialized memory.
+- Added a `DecodeFinished` type to be used with `Decode::decode_into`.
+
+### Fixed
+
+- Trying to deserialize a big boxed array (e.g. `Box<[u8; 1024 * 1024 * 1024]>`) won't overflow the stack anymore.
+- Trying to deserialize big nested enums with many variants won't overflow the stack anymore.
+- Elements of partially read arrays will now be properly dropped if the whole array wasn't decoded.
+
+### Changed
+
+- The derive macros will now be reexported only when the `derive` feature is enabled,
+  as opposed to how it was previously where enabling `parity-scale-codec-derive` would suffice.
+- The `max-encoded-len` feature won't automatically enable the derive macros nor pull in the
+  `parity-scale-codec-derive` dependency.
+
 ## [3.5.0]
 
 ### Added

Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "parity-scale-codec"
 description = "SCALE - Simple Concatenating Aggregated Little Endians"
-version = "3.5.0"
+version = "3.6.0"
 authors = ["Parity Technologies <admin@parity.io>"]
 license = "Apache-2.0"
 repository = "https://github.com/paritytech/parity-scale-codec"
@@ -12,7 +12,7 @@ rust-version = "1.60.0"
 [dependencies]
 arrayvec = { version = "0.7", default-features = false }
 serde = { version = "1.0.164", optional = true }
-parity-scale-codec-derive = { path = "derive", version = "3.1.4", default-features = false, optional = true }
+parity-scale-codec-derive = { path = "derive", version = ">= 3.6.0", default-features = false, optional = true }
 bitvec = { version = "1", default-features = false, features = [ "alloc" ], optional = true }
 bytes = { version = "1", default-features = false, optional = true }
 byte-slice-cast = { version = "1.2.2", default-features = false }
@@ -21,7 +21,7 @@ arbitrary = { version = "1.0.1", features = ["derive"], optional = true }
 impl-trait-for-tuples = "0.2.2"
 
 [dev-dependencies]
-criterion = "0.3.0"
+criterion = "0.4.0"
 serde_derive = { version = "1.0" }
 parity-scale-codec-derive = { path = "derive", default-features = false }
 quickcheck = "1.0"
@@ -46,7 +46,7 @@ fuzz = ["std", "arbitrary"]
 # Enables the new `MaxEncodedLen` trait.
 # NOTE: This is still considered experimental and is exempt from the usual
 # SemVer guarantees. We do not guarantee no code breakage when using this.
-max-encoded-len = ["parity-scale-codec-derive/max-encoded-len"]
+max-encoded-len = ["parity-scale-codec-derive?/max-encoded-len"]
 
 # Make error fully descriptive with chaining error message.
 # Should not be used in a constrained environment.