GitLab

* add selector_id! and selector_bytes! proc macros

* implement blake2x256! macro

* re-export blake2x256! macro from ink_lang crate

* apply rustfmt

* add BLAKE2b to hunspell dictionary

* add UI tests for blake2x256! macro

* improve span for non-literal inputs to blake2x256! macro

* add non-literal input failure UI test to blake2x256! macro

* improve error span for non-literal selector_{id,bytes}! macro inputs

* rename UI test blake2x256 -> blake2x256_macro

* rename UI test

* add UI tests for selector_id! proc. macro

* fix UI test

* fix UI test expectation

* add UI tests for seletor_bytes! macro

* make flaky and broken codecov CI happy again ...

Updates the requirements on [pretty_assertions](https://github.com/colin-kiegel/rust-pretty-assertions) to permit the latest version.
- [Release notes](https://github.com/colin-kiegel/rust-pretty-assertions/releases)
- [Changelog](https://github.com/colin-kiegel/rust-pretty-assertions/blob/main/CHANGELOG.md)
- [Commits](https://github.com/colin-kiegel/rust-pretty-assertions/compare/v0.7.1...v1.0.0

)

---
updated-dependencies:
- dependency-name: pretty_assertions
  dependency-type: direct:production
...
Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* [storage] Allow one variant enum to derive SpreadLayout

I have an enum that has only one variant, and I want to store the enum value in the storage. I might add new variants later in development, so the enum is not useless.

The current implementation doesn't allow to derive the `SpreadLayout` trait for one variant enums, and I fixed it on this PR.

# Sample code

```rust
#![cfg_attr(not(feature = "std"), no_std)]

use ink_lang as ink;

#[derive(
    Copy,
    Clone,
    scale::Encode,
    scale::Decode,
    ink_storage::traits::SpreadLayout,
    ink_storage::traits::PackedLayout,
)]
#[cfg_attr(
    feature = "std",
    derive(scale_info::TypeInfo, ink_storage::traits::StorageLayout)
)]
pub enum MyEnum {
    A,
}

#[ink::contract]
mod enum_test {
    use super::MyEnum;

    #[ink(storage)]
    pub struct EnumTest {
        value: MyEnum,
    }

    impl EnumTest {
        #[ink(constructor)]
        pub fn new() -> Self {
            Self { value: MyEnum::A }
        }

        #[ink(message)]
        pub fn get(&self) -> MyEnum {
            self.value
        }
    }
}
```

Without this change, I get this error.

```
$ cargo +nightly contract build
 [1/5] Building cargo project
    Updating crates.io index
   Compiling enum_test v0.1.0 (/private/var/folders/zn/l2f569z56vnghtt524x1mv6w0000gn/T/cargo-contract_FM50JF)
error: proc-macro derive panicked
  --> /snip/enum_test/lib.rs:10:5
   |
10 |     ink_storage::traits::SpreadLayout,
   |     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
   |
   = help: message: can only operate on enums
```

* Update crates/storage/derive/src/spread_layout.rs
Co-authored-by: Robin Freyler <robbepop@web.de>

* cargo fmt
Co-authored-by: Robin Freyler <robbepop@web.de>

* use ir::Receiver ToTokens impl in codegen

* slightly alter codegen of fn definitions in item impls

* impl ToTokens for Visibility and use it in codegen where possible

* Fix `env::block_timestamp()` example

* Add clarifying note for `env::block_timestamp()`

* Add example for `env::block_number()`

* Add `-Clink-dead-code`

* Test trait cross-call only when not measuring code coverage

Otherwise a linker error will happen due
to `__ink_enforce_error` being included.

* Fix `clippy::derivable_impls`

* Fix `dead_code`

The `struct` is only a public-facing API.

* Fix yaml

* Remove `allow(dead_code)`

* Actually use `$POLL_SLEEP`

* Produce `unreachable!` instead of linker error for `codecov` CI stage

* Move `if` condition outside of `quote`

* Improve macro hygiene

* Reduce code duplication

* remove some duplicated AsRef impl code in generators

* improve #[ink::test] macro hygiene and error message upon failure

* import standard ink! traits anonymously in codegen

* modernize and improve macro hygiene of metadata codegen

* modernize and simplify and improve macro hygiene of storage codegen

* extend selector_works unit test

* fix minor error in item impl error message

* move ink_lang_macro UI tests into macro-specific folders

* fix GitLab CI formatting pass

* migrate back to a single trybuild test case

This is because our CI currently fails if we have multiple test cases.

* implement Display for ink_lang::Error

* re-export HexLiteral trait from ink_lang_ir crate

* use big-letters for hex-encoded message and constructor identifiers in codegen

* improve panic messages in dispatch codegen for call and deploy fns

* improve macro hygiene in dispatch codegen

* add ir::Selector::into_be_u32 method

* improve and modernize dispatch selector namespace codegen

* check that argument in #[ink(namespace = "argument")] is a Rust identifier

* add UI tests for #[ink(namespace = "..")] ink! property

# Conflicts:
#	crates/lang/macro/tests/compile_tests.rs
#	crates/lang/macro/tests/ui/fail/N-01-namespace-invalid-identifier.rs
#	crates/lang/macro/tests/ui/fail/N-01-namespace-invalid-identifier.stderr
#	crates/lang/macro/tests/ui/fail/N-02-namespace-invalid-type.rs
#	crates/lang/macro/tests/ui/fail/N-02-namespace-invalid-type.stderr
#	crates/lang/macro/tests/ui/fail/N-03-namespace-missing-argument.rs
#	crates/lang/macro/tests/ui/fail/N-03-namespace-missing-argument.stderr

* fix error message for non-identifier namespace argument

* add HexLiteral utility trait to expand hex-encoded integer literals

* add ir::Selector::hex_lits method

* expand into hex-encoded integer literals for metadata selectors

* simplify Selector::hex_lits method using array::map API

* add unit test for Selector::hex_lits method

* miscellaneous improvements to macro hygiene in derive macros

* fix storage_derive UI tests

* make #[ink(selector = ..)] take an integer instead of a string

Also warn about deprecation if a user still uses the old string parameter.

* update tests for new selector int parameter

* update contract examples for new selector int parameter

* remove unused regex dependency from ink_lang_ir

Also remove unnecessary regex dependency from ink_lang_codegen

* update README for new selector int parameter

Also update namespace parameter description.

* update preliminary release notes

* update spellcheck dict

* Derive `Default`

* Fix `field is never read`

* Fix `clippy::derivable_impls`

* Fix `dead_code`

The `struct` is only a public-facing API.

* Get rid of `allow(dead_code)`

* Revert "Get rid of `allow(dead_code)`"

This reverts commit 6b9a95f9.

* Revert "Fix `dead_code`"

This reverts commit 42ab18be.

* New approach to fix `dead_code`

* Fix naming

* Reflect that `return_type` never needs to be read

* Fix typo

* Bump version to rc5

* Add release notes

The function was recently marked as safe:
https://doc.rust-lang.org/nightly/core/arch/wasm32/fn.unreachable.html.

>This function is safe to call and immediately aborts the execution.

Follow-up to https://github.com/paritytech/substrate/pull/9550.

* Remove `DispatchRetCode`

* Replace unwrap with expect

* Import quickcheck macro only for bump allocator

The fuzz tests exist only in the bump allocator.

* Apply `cargo fmt`

* Add clippy flag `--all-targets`

* Fix `redundant_closure`

* Fix `redundant_clone`

* Fix `clone_on_copy`

* Fix `needless_borrow`

* Fix `bool_assert_comparison`

* Fix `len_zero`

* Fix `stable_sort_primitive`

* Forward feature

* Allow `type_complexity`

* Fix `unnecessary_mut_passed`

* Fix `manual_map`

* Fix `match_like_matches_macro`

* Remove duplicate test

* Fix `new_without_default`

* Fix `match_ref_pats`

* Allow clippy rules for  tests

* Fix `unit_arg`

* Apply `cargo fmt`

* Add explanatory comments

* Remove `--all-targets` for Wasm tests

* Fix `or_fun_call`

* Fix `bool_comparison`

* Fix `needless_collect`

* Revert "Forward feature"

This reverts commit 472c50ed.

* Fix `bool_assert_comparison`

* Deactivate test because of feature issue with dev-dependencies

* Add flag

* Fix `unique_topics` tests

* Allow `clippy:bool_assert_comparison`

* Revert "Fix `bool_assert_comparison`"

This reverts commit bceb2dc2.

* Revert "Fix `bool_comparison`"

This reverts commit 02960f98.

* Fix flags positioning

* Add comment for clarification

* Fix comparison

* Do not check all targets for target = wasm32

* Adapt `check-workspace.sh` with clippy flag

* Add clarification comment

* Ignore test

* Revert `CLIPPY_FLAGS` hack

* Link ink! issue

* Fix `assert`'s

* Fix `assert`'s

* Improve language

* Add test with mocked chain extension

* Export off-chain trait `ChainExtension`

* Add missing space

* Remove unequal assert

* Force the `DEBUG_ENABLED = enabled` code to be executed when RPC is used

* Fix clippy

* Flip the conditional

* Add `Fixed` entry for #842

* Replace `3.0.0-rc3` with `3.0.0-rc4`

* Add bump allocator to release notes

* Add fuzz tests for `StorageVec::binary_search*`

* Decrease number of fuzz tests which are run

We found that bugs are usually already found
with low numbers.

* Apply suggestions from code review
Co-authored-by: Hernando Castano <HCastano@users.noreply.github.com>

* Improve expect

* Fix variable naming
Co-authored-by: Hernando Castano <HCastano@users.noreply.github.com>

* Add bump allocator skeleton

* Implement `alloc` for our bump allocator

* Make the allocator usable globally

* Remove unused `init()` function

* Nightly RustFmt

* Use global mutable static instead of Mutex

This will reduce our use of dependencies which will hopefully
reduce our final Wasm binary size.

Also, apparently spinlocks aren't actually all that efficient.
See: https://matklad.github.io/2020/01/02/spinlocks-considered-harmful.html

* Stop assuming that memory is allocated at address `0`

* Remove semicolon

* Use correct address when checking if we're OOM

* Remove unnecessary unsafe block

* Return null pointers instead of panicking

Panicking in the global allocator is considered undefined behaviour.

* Use `checked_add` when getting upper limit memory address

* Use `MAX` associated const instead of `max_value`

* Inline `GlobalAlloc` methods

* Turns out I can't early return from `unwrap_or_else` 🤦

* Rollback my build script hacks

* Add initialization function to allocator

* Add some docs

* Make the bump allocator the default allocator

* Allow bump allocator to be tested on Unix platforms

* Remove unecessary checked_add

* Add error messages to unrecoverable errors

* Remove `init` function from allocator

Instead we now request a new page whenver we need it, regardless
of whether or not it's the first time we're allocating memory.

* Try switching from `mmap` to `malloc` when in `std` env

* Fix `is_null()` check when requesting memory

* Stop requesting real memory for `std` testing

Instead this tracks pages internally in the same way that the Wasm
environment would. This means we can test our allocator implementation
instead of fighting with `libc`.

* Gate the global bump allocator when not in `std`

* Allow for multi-page allocations

* Update the module documentation

* Override `alloc_zeroed` implementation

* Forgot to update Wasm target function name

* Appease the spellchecker

* Use proper English I guess

* Get rid of `page_requests` field

* Explicitly allow test builds to use test implementation

* All link to zero'd Wasm memory reference

* Check that our initial pointer is 0 in a test

* Add `cfg_if` branch for non-test, `std` enabled builds

* Simplify `cfg_if` statement

error: constructor `lazy` has the same name as the type
	   --> crates/storage/src/lazy/mod.rs:121:5

	       121 | /     pub(crate) fn lazy(key: Key) -> Self {
	       122 | |         Self {
	       123 | |             cell: LazyCell::lazy(key),
	       124 | |         }
	       125 | |     }
		   | |_____^

	= note: `-D clippy::self-named-constructor` implied by `-D warnings`
	= help: for further information visit https://rust-lang.github.io/rust-clippy/master/index.html#self_named_constructor

* Remove not used Salt

* Updated unit test to cover case with code generation for trait.

* Renamed seconds "new" constructor. Seems the same naming for constructors with different signature causes compilation error on linux platform.

* Let's try to move trait definition into contract. Maybe it will resolve compilation error in CI.

* Moved trait implementation to separate test.

* Changed version of docker image to staging.
Added `cc --version` in info section

* Apply suggestions from code review
Co-authored-by: Michael Müller <mich@elmueller.net>
Co-authored-by: Michael Müller <mich@elmueller.net>

* Prepare rc4 release notes

* Improve language

* Fix link

* Fix spelling mistake

* Remove duplicates of "the the"

* [storage] Implement binary_search

* [storage] Add binary_search testcase for missing smaller element

* [storage] use indexing instead of Vec::get
Co-authored-by: Robin Freyler <robbepop@web.de>

* [storage] Port binary_search tests from core

* [storage] Format tests

* [storage] Format tests

* Remove old comment
Co-authored-by: Robin Freyler <robbepop@web.de>

* [storage] Incorporate review feedback

* [CI] add comparator to spellcheck

Porting binary_search to the storage Vec, including the documentation found in core, led to a failure in spellcheck. Since core is considered to be correct, it is updated on spellcheck.

* incorporate review feedback

* add permalink to stdlib source

* Split long line

* fix examples

* RustFmt the doc examples
Co-authored-by: Robin Freyler <robbepop@web.de>
Co-authored-by: Hernando Castano <hernando@hcastano.com>

* [lang] catch illegal struct destructuring pattern in ink! message arguments.

* fmt