# Changelog

All notable changes to this project will be documented in this file starting from `2.0.0`.

The format was adapted from [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).

The following shorthand will be used, in this order:  
`+` - New feature (`_.X._`)
`/` - Breaking Change (`X._._`)
`-` - Removal (`X._._`)
`~` - Misc. Change (Any, really)
`!` - Bug-Fix (`_._.X`)

This project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).

## [2.0.0] - Unreleased

This update intends to fix mistakes made during the 1.0.0 version with regards to code style
and naming. It also intends to overhaul the `Span` and `Notification` systems to better integrate
with grammars built off of Alkale. Other miscellaneous changes have similar goals— making Alkale
work better as an integrated stage wtihin a larger compiler/grammar.

`+` Added `Spanned`, a replacement for previous instances of `(T, Span)`.  
`+` Added `Spannable`, indicating a type can be converted into a `Spanned`. (All types implement this)  
`+` Added `NotificationBuilder`, a way to construct generic notifications.  
`+` Added `NotificationAcceptor`, indicating types that can aggregate built-in `Notification`s.  
`+` Added `NotificationSeverity`, used for `Notification`s.  
`+` Added `NotificationList`.
`+` Added `SourceCodeScanner`, a replacement for `TokenizerContext`.
`+` Added `FinalizedLexerResult`, a replacement for `LexerResult`.
`+` Added `common` feature to enable/disable commonlib.

`/` `Notification`s are no longer a trait, instead they serve as a default error type.  
`/` Notifications in `TokenizerContext` are no longer stored as dynamic references.  
`/` `Span` type no longer stores line and column position.  
`/` All instances of "tokenizer" replaced with "lexer" for correctness.  
`/` `TokenizerContext` now uses `NotificationList` instead of `Vec<Notification>`. 
`/` Lexers now take strings instead of char arrays.
`/` `LexerResult` now handles accumulating notifications and tokens.

`-` Removed direct iterator results from `TokenizerContext` due to complicating parsing for little benefit.  
`-` Removed `TokenizerContext`, use `SourceCodeScanner` instead.

`~` `Span` type now stores index positioning with a length instead of ending index.   
`~` No longer dependent on `utf8-chars`.