https://github.com/andyfinnell/kalerl

An Erlang implementation of the Kaleidoscope language from the LLVM tutorial.
https://github.com/andyfinnell/kalerl

Last synced: 2 months ago
JSON representation

An Erlang implementation of the Kaleidoscope language from the LLVM tutorial.

• Host: GitHub
• URL: https://github.com/andyfinnell/kalerl
• Owner: andyfinnell
• License: mit
• Created: 2015-12-23T17:48:54.000Z (almost 9 years ago)
• Default Branch: master
• Last Pushed: 2015-12-24T14:40:52.000Z (almost 9 years ago)
• Last Synced: 2023-08-03T18:05:47.431Z (over 1 year ago)
• Language: Erlang
• Size: 83 KB
• Stars: 12
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# kalerl

##Description

An Erlang implementation of the [Kaleidoscope language from the LLVM tutorial](http://llvm.org/docs/tutorial/index.html). It's not a straight port as it targets BEAM, the Erlang VM, instead of native code. The project is a work in progress for the purpose of learning how to generate code targeting BEAM.

The name is pronounced "Kal-Earl," like [Kal-el](https://en.wikipedia.org/wiki/Superman)'s redneck cousin who's really into compilers.

##Dependencies

* rebar

* make

* erlando (automatically downloaded and built by rebar)

##Build

To build the kalerl compiler just run make.

    $ make

This will place the `kalerl` executable in the `bin` subdirectory.

##Use

The `kalerl` compiler can only take one parameter, the file to compile. For example:

    $ ./bin/kalerl test/add.kalerl

This will generate a BEAM file in the same directory as the source file. In this case it will be `test/add.beam`. From here, the BEAM file can be used just like any other; all functions are exported by default.

    $ cd test

    $ erl

    Eshell V7.2  (abort with ^G)

    $ 1> l(add).

    {module,add}

    

The above moves to the `test` directory, starts the Erlang shell, and loads the `add` BEAM file just built. To call a function:

    $ 2> add:add(3.0).

    5.0

    

Also, all kalerl modules export a `main/0` function that executes all the top level expressions.

    $ 3> add:main().

    5.0

##Language Differences

There are a few difference between kalerl and Kaleidoscope as defined by the LLVM tutorial. The most obvious is there is no REPL in kalerl; it assumes reading and writing from files.

One major deviation from the original Kaleidoscope language is the lack of mutability. BEAM/Erlang doesn't support it (mostly, kinda), so it doesn't make sense for kalerl to either. This means kalerl remains a functional language. Because of the lack of mutability, the `var/in` construct in Kaleidoscope becomes a `let/in` and kalerl does not implement an assignment operator. An example of a kalerl `let/in`:

    let x = 3 + 4 * 5 in

    let y = x + 7 in

    ...

Another difference is the handling of externs. kalerl doesn't handle calling out to C functions; instead it supports calling into Erlang modules. To accomplish that the language syntax changed to support designating the external module name. For example:

    extern foo(a) in bar

    

declares that there is a function `foo/1` in module `bar`. After this declaration, kalerl code can call `foo/1` like any other kalerl function and it will do the right thing.