tweag
Tracking issue: Stdlib testing
We had a bug introduced right before releasing 0.3, which led to releasing 0.3.1 right after. This bug could have been avoided by having a more systematic testing of the stdlib. Right now, the stdlib is mostly undertested, if tested at all.
Possible approaches include property-based testing. We could expose a primitive in Nickel to generate random values given a few parameters, or have the Rust side of the testing infrastructure generating inputs to the tests in dedicated files input.ncl that are then consumed by the tests.
The recent work of @jneem now checks in the CI that the examples given in the stdlib's documentation do run and produces the expected result. We can probably do more than that though, so keeping this open.