Changelog for treescript

12/19/18

Started brainstoriming

12/21/18

Initial specification

Setup Haskell project

• General architecture/idea from previous project
  • Will work very hard to make less bloated though, no unnecessary features
• Although there are a few style differences:
  • More generics / autoderiving
  • Language extensions - in alphabetical order
  • Imports - TreeScript in alphabetical order, then others in alphabetical order
  • Qualified imports always have shortest non-conflicting qualifiers, no special differences for "main" imported class (e.g Text in Data.Text)
  • Try to put small functions above big functions in modules - still order is datatypes, classes, instances, and functions
  • Modules can be bigger, but they'll still be split when too big (overall module structure still shouldn't matter much) - generally 100 lines is OK, 200 or 300 in some cases OK too

12/23/18

Lexing and Lex AST

• Passes a few tests but likely still some bugs

General test suite (taken from previous project)

12/25/18

Parsing and Sugar AST

• Can now analyse this AST and generate code with it even though it might not be well-formed
• Passes a couple tests but likely some bugs

Some docs

• Code design
• Changelog

12/29/18

Significantly changed specification, slightly changed lexing and parsing

• Mainly made specification much more precise, and figured out how to implement
• Updated README, lexing, parsing, and tests
• Sugar and Core ASTs don't have guard support yet, but Lex AST has guard lexemes

1/01/19

Desugaring and Core AST

• Can parse spliced code blocks with plugins
• Still can't parse guards

Basic plugin support

• Plugin settings
• Plugin languages
  • Currently the only languages are scheme and text
  • scheme has a specification and parsing (although dots followed by lists don't work)
  • text has a specification
• Plugin printing skeleton

1/02/19

C template, translation and Translate phase

• Template for C code
• Updated README, added how Descript code will be translated into C, spliced into the template, and this will be compiled to get the Descript executable
• Tested via printing/observing, didn't actually compile code yet

1/03/19

Compiling and CLI

• Compiling both C (last step) and Descript (all steps)
• Tests for compiling - currently only work for AST data
• Command-line interface only supports compiling
• Started to change name to "TreeScript"

1/04/19

Renamed to treescript, added vscode extension to project

Run on source code - treescript run

• Parses code into AST data, runs the executable, and prints it back into code
• Now can take a Descript source file and input, and generate output - treescript compile the source file, treescript run the executable with the input.
• I/O exceptions should now be wrapped in Results, better error messages and handling
• Bugfixes for compiling (fixed memory issue)

1/07/19

Started to add functions and "groups"

• Syntax for functions and groups (Lex to Core phases) - untested, but regular syntax still works
• Some documentation
• TODO Add more documentation, add ability to translate to C (Translate phase)

1/12/19

Libraries

• Renamed "servers" to libraries
• Changed library specification and implementation:
  • Directly inserts library into compiled C program and calls function, instead of communicating with an external command-line program
  • Functions are qualified by library
  • (From previous changes) functions have fixed numbers of arguments, libraries have names in spec
• Added Base library with some simple functions - arithmatic and string appending
  • May change these functions later, e.g. make them fail more gracefully
• Added test for library which also uses Y-combinator
• Bugfixes, refactors, documentation improvements, etc.

Working on VSCode extension / language server on a separate branch

2/15/19

Wrote interpreter in Rust, instead of directly compiling into C

• Currently slower than C version
• But should be faster to debug and improve

Unreleased changes