Announcing 1.0.0
  July 17, 2018

Today I would like to announce the 1.0.0 release of, which brings Behaviour Driven Development to the OCaml programming language. Before I start though, I would like to thank Thegaram for his contributions and feedback to this.


For those of you who are not familiar with the OCaml language, it is a
multi-paradigm, strongly, structural and, inferred typed language of the ML family of languages which is generally used for its functional properties and strong module system. Companies which use it include Facebook, who created the ReasonML language, which is a thin wrapper around OCaml, and Jane Street Capital, who use it for market making in the financial markets. I would highly recommend having a look.


While many of OCaml's language features tend to reduce the need for testing in general, I still feel that communicating between stakeholders and programmers is important and Cucumber is well-placed to help in that. Additionally, I felt that this was a good way for me to contribute to the OCaml ecosystem in a positive way.


There are a few complications when working with The first and the most important is that delegates its Gherkin parsing to the gherkin-c implementation using OCaml's foreign function interface. This means that, before you can use, you will need to compile and install the gherkin-c as a shared library on your system. On Linux, this means you will need to install in /usr/lib and copy the gherkin-c header files into /usr/include/gherkin. uses the Dune build system for OCaml. You will need to have this installed before you can build Unfortunately, I have not yet added to the Opam package manager. If anyone would like to help me do that, I would be very grateful.

Once you have checked out the code and have the above prerequisites installed, all you need to do is:

dune build && dune install

which will install the cucumber package into your local Opam repository. From there you can write your step definitions in OCaml and run your feature files against them. For more information on how to use, please see the README file on the repository.

One of the more unusual aspects of is that it assumes that once you start executing your step definitions that it will control the command line arguments. This can be surprising so please be aware of it. Basically, the executable created by the compilation process is the cucumber runner and it is that executable which will run your step definitions.


There are many features that come with

  • Gherkin 5.0 is supported (Feature, Background, Given, When, Then, etc.)
  • Datatable transforms (the user can define transformations for datatables but this is not done automatically and must be applied by the user)
  • Before and After steps
  • State passing is handled by the runner and is passed to Step definitions automatically

The state passing feature is, I believe, a first for a Cucumber implementation.

Roadmap will work today and please let me know if you would like any more features. However, here are a few things that I (and anyone else) will be working on.

  • Update to Gherkin 6.0 using gherkin-go
  • Lwt cooperative threading
  • Bring fully within the Opam package manager

My main focus will be the first thing on the list so I am looking forward to moving to the gherkin-go universal runner.

I hope you enjoy using it as much as I have had making it. If you have any questions, you can ask on the Cucumber Slack or you can drop an issue on the Github issue tracker.