Skip to content

Latest commit

 

History

History
73 lines (51 loc) · 4 KB

README.md

File metadata and controls

73 lines (51 loc) · 4 KB
Raw

TryMLTS

A web application designed to try and play with the MLTS programming language.

Sources

The source is hosted on GitHub{target="_blank"} as is the demo{target="_blank"}.

Our prototype interpreter is written in λProlog. Its source is located in the lib/data/core folder{target="_blank"}. The more interesting parts are:

Basic usage

You can write code in the left panel and have it evaluated by clicking the Run button or the key combination ctrl+b while the focus is on the editor. If something goes wrong you can stop the execution by clicking Restart kernel.

When running your code will first be translated to the λProlog abstract syntax. Some static checking happen during this phase. The translation's result (be it an actual translation or an error) will quickly appear on the right, under the λProlog tab.

Then the evaluation will start. Result will be shown in the result tab only at the very end of the process. You can witness progress and errors in the Log tab. When results are shown you can click them to quickly locate their position in the source code.

Concrete syntax

MLTS' concrete syntax is based on the on of OCaml. A program written in MLTS not using the new constructs nab in, new in, \ and => should compile with the ocamlc compiler.

  • Datatypes can be extended to contain new nominal constants and the new X in body program phrase provides a binding that declares that the nominal X is new within the lexical scope given by body.

  • A new typing constructor => is used to type bindings within term structures. This constructor is an addition to the already familiar constructor -> used to type functional expressions.

  • The backslash \ is an infix symbol that is used to form an abstraction of a nominal over its scope. For example, X\ body is a syntactic expression that hides the nominal X in the scope body. Thus the backslash introduces an abstraction.

  • The @ eliminates an abstraction for example, the expression (X\body) @ Y denotes the result of substituting the abstracted nominal X with the nominal Y in body.

  • Rules within match-expression can also contain the nab X in rule binder: in the scope of this binder, the symbol X can match existing nominals introduced by the new binder and the \ operator. X is then bound over the entire rule (including both the left and right-side of the rule).

Known issues and missing features

This work is highly experimental and you should expect bugs to appear. If so please report them on the issue tracker{target="_blank"}.

  • Error messages are not always meaningful of even shown.
  • Mutual recursion is coming soon.
  • Pairs must always be written using parenthesis: (a, b) is a pair, a, b is not.
  • There is no syntactic sugar for non-empty lists. Use a::b::c::[] instead of [a;b;c].

How it works

Our prototype interpreter for MLTS is written in λProlog. When you press the "Run" button the MLTS concrete syntax is transpilled to it's λProlog counterpart and Elpi, an embeddable λProlog interpreter written in OCaml, runs it.

All this happens locally in your browser thanks to the js_of_ocaml js_of_ocaml.

Build dependencies

Ocaml modules dependency(install them via opam) : menhir elpi js_of_ocaml js_of_ocaml-ppx base64

Credits

The "backend" is powered by OCaml, λProlog, Elpi and js_of_ocaml.

The frontend uses a standard mix of Bootsrap and Jquery but also the Ace code editor, highlight-js for additional syntactic coloring and open-iconic for icons.