Popper is an OCaml testing library that can be used for writing simple unit-tests as well as property-based ones. Its underlying design is inspired by the Python library Hypothesis.
See the documentation page for information on how to get started.
High-level features of Popper include:
- A uniform API for defining regular unit- and property-based tests.
- Embedded shrinking — invariants used when constructing samples for property-based tests are always respected.
- Compositional design — tests may be bundled and nested arbitrarily.
- Ships with a
ppx
for automatically deriving comparator and sample functions for custom data types. - Deterministic (and reproducible) results.
- Colorful output (cred goes to Alcotest, couldn't resist some inspiration).
- Support for line-number reporting, timing and logging.
See CONTRIBUTING.md for how to build and contribute to Popper.
- Check out the getting started section for a step by step introduction.
- Take a look at some examples in the examples folder.
- Browse the API docs.
Here's what test output might look like:
It was generated from the following code:
open Popper
open Sample.Syntax
type exp =
| Lit of bool
| And of exp * exp
| Or of exp * exp
| Not of exp
[@@deriving show, ord, popper]
(* A buggy evaluator function *)
let rec eval = function
| Lit b -> b
| And (e1, e2) -> eval e1 || eval e2
| Or (e1, e2) -> eval e1 || eval e2
| Not b -> not @@ eval b
(* A simple unit test *)
let test_hello_world =
test @@ fun () ->
equal Comparator.string "hello world" (String.lowercase_ascii "Hello World")
(* Another unit test *)
let test_lit_true = test @@ fun () -> is_true (eval (Lit true) = true)
(* A property-based test *)
let test_false_ident_or =
test @@ fun () ->
let* e = exp_sample in
is_true (eval e = eval (Or (Lit false, e)))
(* Another property-based test *)
let test_true_ident_and =
test @@ fun () ->
let* e = Sample.with_log "e" pp_exp exp_sample in
is_true ~loc:__LOC__ (eval e = eval (And (Lit true, e)))
(* Bundle some tests together *)
let exp_suite =
suite
[ ("Lit true", test_lit_true)
; ("False ident or", test_false_ident_or)
; ("True ident and", test_true_ident_and)
]
(* Top-level test-suite *)
let suite =
suite [ ("Hello World", test_hello_world); ("Expression", exp_suite) ]
let () = run suite
Popper is designed with the following objectives in mind:
-
Make it as seamless as possible to write property-based tests — for instance by using a ppx to derive sample functions for custom data-types.
-
Use embedded shrinking (ala Hypothesis) and eliminate the need for writing shrinkers manually.
The property-based aspects overlap with the OCaml libraries QCheck and Crowbar.
Popper also supports writing simple unit tests and the ability to compose tests into suites. This API and the output is partly inspired by the testing library Alcotest.
Here's a table comparing features across different OCaml testing libraries:
Library | Test suites | Property-based | Embeded shrinking | PPX generators | Fuzzying |
---|---|---|---|---|---|
Popper | ✅ | ✅ | ✅ | ✅ | ❌ |
Alcotest | ✅ | ❌ | - | ❌ | - |
OUnit | ✅ | ❌ | - | ❌ | - |
QCheck | ✅ | ✅ | ❌ | ❌ | ❌ |
Crowbar | ❌ | ✅ | ❌ | ❌ | ✅ |
Base_quickcheck | ❌ | ✅ | ❌ | ✅ | ❌ |
It might be possible to write some adaptors to be able to integrate with these libraries but nothing such exists at the moment.