Dhall is a programmable configuration language optimized for maintainability.

You can think of Dhall as: JSON + functions + types + imports

Note that while Dhall is programmable, Dhall is not Turing-complete. Many of Dhall's features take advantage of this restriction to provide stronger safety guarantees and more powerful tooling.

You can find more details about the language by visiting the official website:

• https://dhall-lang.org

I am no longer maintaining this project. I got it to support about 90% of the language but then lost faith in the usability of dhall for my purposes. I am willing to hand this over to someone who's excited about dhall and rust.

This is the Rust implementation of the Dhall configuration language. It is meant to be used to integrate Dhall in your application.

If you only want to convert Dhall to/from JSON or YAML, you should use the official tooling instead; instructions can be found here.

For now, the only supported way of integrating Dhall in your application is via the serde_dhall crate, and only parsing is supported.

Add this to your Cargo.toml:

[dependencies]
serde_dhall = "0.12.1"

Reading Dhall files is easy and leverages the wonderful serde library.

use std::collections::BTreeMap;

// Some Dhall data
let data = "{ x = 1, y = 1 + 1 } : { x: Natural, y: Natural }";

// Deserialize it to a Rust type.
let deserialized_map: BTreeMap<String, u64> = serde_dhall::from_str(data).parse().unwrap();

let mut expected_map = BTreeMap::new();
expected_map.insert("x".to_string(), 1);
expected_map.insert("y".to_string(), 2);

assert_eq!(deserialized_map, expected_map);

dhall requires Rust >= 1.60.0

This implementation currently supports most of the Dhall standard version 20.0.0.

The main missing feature is import headers. See here for a list of the other missing features.

This section will cover how we can get started on contributing this project.

To get a copy of this repository we can run:

$ git clone https://github.com/Nadrieril/dhall-rust.git

But we also might note that it's better practice to fork the repository to your own workspace. There you can make changes and submit pull requests against this repository.

After the repositry has been cloned we need to update the git submodule in the project, i.e. dhall-lang. We can do this by running:

$ git submodule update --init --recursive

A preferred method among the Rust community for developing is to use rustup.

It can be installed by running:

$ curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

or if nix is your tool of choice:

$ nix-shell -p rustup

Once rustup is installed we can get it to manage our toolchain by running:

$ rustup toolchain install stable

Then we can manage our building and testing with the cargo dependency manager:

$ cargo build

$ cargo test -- -q

You can also run tests individually by their name:

$ cargo test tests::spec::name_of_test

Now we can have fun and happy contributing!

The test suite uses tests from the dhall-lang submodule as well as from the local dhall/tests directory. The various tests are run according to the instructions present in dhall-lang/tests/README.md.

If an output test file (a fooB.dhall file) is missing, we will generate it automatically. This is useful when writing new tests. Don't forget to commit it to git !

If one of the specification tests fails but you prefer the new output, you can run the test(s) with --bless to overwrite the result file with the new output. This happens often with ui tests (see below), since we may want to change the phrasing of errors for example. Note that the --bless argument is only accepted by the spec tests and will not be recognized if you also run other test.

$ cargo test --test spec -- -q --bless

In addition to the usual dhall tests, we additionally run "ui tests", that ensure that the output of the various errors stays good. The output of the ui tests is stored in the local dhall/tests directory, even for the tests coming from dhall-lang. They are stored in a .txt file with the same name as the corresponding test.

I try to keep commit messages somewhat in the style of Conventional Commits. That means the commit message should start with feat:, test:, spec:, doc:, fix:, style:, refactor:, chore:, perf: or similar prefixes.

A breaking change should be indicated with ! before the :.

Licensed under the terms of the 2-Clause BSD License (LICENSE or https://opensource.org/licenses/BSD-2-Clause)