brokenprogrammer/brk

brk's Issues

Finish language definition

It is much easier to write a parser if the language have a clear definition.

The definition should be in some "ENBF-ish" grammar and clear enough that the implementation of the parser can begin.

Find a decent test framework and write tests for the lexer & parser.

Ideas for testing frameworks:

• Catch
• Google Test
• Boost.Test

Testing is important for compilers, they will in a sense define the language, the tests will check for correct code and detect incorrect code.
This might be a good opportunity to try out TDD as well..

Update the makefile to properly include the llvm libraries.

Current solution is pure copypaste of what Brk needs from llvm which simply wont do.. There is llvm commands that works in the terminal but not using the makefile such as llvm-config --cxxflags

Fix this so its easier for me to work on diff machines.

Parentheses return nullptr when parsed.

'(' expression ')' is being parsed using e = parseExpression(); and when returned from the parseUnaryExpression function the unique_ptr is a nullptr and the compiler crashes.

Code for the parens expressions is found at line 187 in parse.cpp

Expression codegen

Lexer error handling

The lexer currently only have //Error comments at the places where it should log some form of error or handle some error. This needs to be added to improve debugging when using Brk and developing Brk.

Create basic types

Parse statements

Parser error handling

The parser currently only have //Error comments at the places where it should log some form of error or handle some error. This needs to be added to improve debugging when using Brk and developing Brk.

Parse expressions

Since the expressions have an initially defined grammar it can now be parsed so lets implement the parsing logic for expressions and then write tests that also evaluate the parsed expressions such as:

• 1
• (1)
• (1 + 2)
• (1 - 2- 3)
• (2 * 3 - 5 * 4)
• 2 * (3-5) * 4
• -1
• 2+-3

These tests will allow us to see that parentheses expressions work as they should and that the precedence work as intended as well.
This will also allow me to get an overview of how the parser should be structured and I can also tune the lexer and token class while this is being done.

Improve Lexer structure.

Some missed tokens are currently not being lexed which needs to be fixed. At the same time as I'm fixing this I might as well improove the lexing by re-implementing the Token struct as a linked or doubly linked list.

Using Integer expression after a floating point expression crashes Brk

Example Brk code that would skip the 1 completely:

55.54 + 123.2;
1 + 2;

Example Brk code that would cause Brk to crash:

55.54 + 123.2;
12;