multicoretests

multicoretests copied to clipboard

Add Util functions to make tests with logging of generated test values

1

This has been useful quite a few times for us, for instance to investigate deadlocks or livelocks, and build standalone reproducers. This might be more generally useful. This is still...

shym

Support multiple ts in Lin and Lin_api signatures

23

This PR takes a stab at supporting multiple `t`s in the signature descriptions of `Lin` and `Lin_api`, thus fixing the `Lin` part of #62. Overall, it supports multiple `t`s by...

jmid

Currently we are printing `string`s, `list`s, etc. unmodified which can sometimes get a bit long to the point that it may detract from seeing the structure of the program causing...

jmid

Add (and use) a runner wrapping every test, to improve error reporting

1

This supersedes #205, providing a custom runner written in OCaml Add a custom runner that allows to display the result of a test in the same way on Unix and...

shym

I've played with computing coverage to have an idea of how much ground we cover. For now, the approach takes a copy of each tested `Stdlib` module and instruments that...

jmid

Faster Lin check for pure operations

2

This is just a suggestion, feel free to discard! (... especially since it changes the public API) I think the sequential consistency check can be a tiny bit faster if...

art-w

#266 reminded me of a couple of model-based memorizations that `STM` could potentially benefit from. [...] It reminds me of an optimization from the [original QuickCheck race-condition paper](https://smallbone.se/papers/finding-race-conditions.pdf): > This...

jmid

Add information to logs with a custom runner

8

Add a bash script to use as runner for our test suite This script adds information to the logs: - explicit names of the test run when they start -...

shym

We should revise the preliminary `Effect`-mode for `Lin` to better match asynchronous code in `Effect` and `Eio` fibers. It would be cool if we could hook in to detect race...

jmid

As listed in `Lin`'s current limitations in the manual page https://ocaml-multicore.github.io/multicoretests/0.1/qcheck-lin/#current-limitations it currently doesn't have combinators for tuples akin to the usual `*` type syntax. This hasn't been much of...

jmid