thp/CHANGELOG.md

145 lines
4.2 KiB
Markdown

# Changelog
## TODO
- Implement functions as first class citizens
- Implement AST transformation before codegen:
Create a new AST to represent PHP source code
and a THP ast -> PHP ast process, so that the
codegen section can focus only in codegen, not in
translation of thp->php.
- Parse __more__ binary operators
- Store tokens for the semantic analysis phase, to have actual error reporting
- Parse more complex bindings
- Watch mode
- Improve error messages
- Parse other language constructions
- Type checking
- Check for conflicting identifiers
- Namespace identifiers in the symbol table
- Stdlib
- Document code
- Watch mode
- Formatter
- Simple language server
- Decide how to handle comments in the syntax (?)(should comments mean something like in rust?)
- Not ignore comments & whitespace, for code formatting
- Abstract the parsing of datatypes, such that in the future generics can be implemented in a single place
## v0.0.13
- [ ] Begin work on a formal grammar
- [ ] Simplify AST
- [ ] Define the top level constructs
- [ ] Include the original tokens in the AST
- [ ] Implement a hello world until semantic analysis
- [ ] Refactor code
- [ ] Remove `PARSER couldn't parse any construction` error & replace with an actual error message
## v0.0.12
- [x] Infer datatype of an identifier
- [x] Infer datatype of a binary operatior
- [x] Infer datatype of unary operator
- [x] Infer datatype of binary operators
- [x] Infer Int & Float as different types
- [x] Execute semantic analysis on the function's block
- [x] Write tests
## v0.0.11
- [x] Parse binding of form `val Type variable = value`
- [x] Parse binding of form `Type variable = value`
- [x] Infer datatype of `value` in the above for a simple expression (minimal)
- [x] Ensure that the anotated datatype matches the datatype of `value` in the above (binding)
- [x] Infer datatype of a `val variable = value` in the AST: Use the infered datatype (binding)
- [x] Formally define the top level constructs
- [x] Parse bindings and function declarations as top level constructs
- [x] Parse function declaration arguments (`Type id`)
- [x] Parse function return datatype (`fun f() -> Type`)
- [x] Return parsing to variables to var/val
- [x] Write tests
## v0.0.10
- [x] Parse function call parameters
- [x] Codegen function call parameters
- [x] Begin work on semantic analysis
- [x] Minimal symbol table
- [x] Check duplicate function declarations
- [x] Improve REPL/File compilation code
- [x] Check binding duplication in it's scope
- [x] Check function duplication in it's scope
- [x] Transform simple THP expression into PHP statements
## v0.0.9
- [x] Hand-make CLI, remove clap
- [x] Compile a single file
- [x] Display error messages during compilation instead of panicking
- [x] Improve error messages
- [x] Implement code generation for ast nodes implemented as of now
## v0.0.8
- Parse block of code
- Parse multiple statements inside a block
- Parse unary operator (`!` & `-`)
- Parse binary operators
## v0.0.7
- Parse minimal function declarations following a grammar
- Parse function call, binding as statement
- Parse a statement as body of a function declaration
## v0.0.6
- Parse function declarations
- Parse multiple function declarations
- Parse multiple bindings
- Compile from file
- Emit INDENT & DEDENT tokens
## v0.0.5
- Scan single line comments
- Refactor String token to include double quotes (") in its content
- Refactor datachecking of semantic analysis
## v0.0.4
- Explicit datatype of variables
- Improve error messages when a syntax error is found (show offending line and offending token)
- Show different error messages for val/var binding
## v0.0.3
- Get datatype of an identifier from the symbol table
- Improve documentation of the code
- Simple ASI: insert semicolon after a single or series of new lines
- The token stream now always ends with a Semicolon and EOF token, regardless of input
## v0.0.2
- Compilation of `val` and `var` bindings with a number, string or boolean as value.
- Register symbols and datatypes in the Symbol table.
- Add better error messages for lexical errors. Show:
- Offending line
- Pointer to offending character
- Error message
## v0.0.1
- Compilation of a `val` binding with a number.
- Scan all tokens except new lines, indentation.