Cargo flaky extends cargo to help you find flaky tests in you test suite, and help you solve them.
Flaky tests are notoriously annoying and difficult to debug, because of their very nature: they seem to happen randomly. The idea behind cargo edit is to run you test a great number of times to try to trigger failure of those flaky tests. But this is not sufficient, because knowing that the bug exists is rarely sufficient to debug it, cargo flaky also allows you to record the failing tests using rr, so you can play back the execution of your tests when they failed. Noice.
from source:
git clone https://github.com/meilisearch/cargo-flaky.github
cd cargo flaky
cargo install --path .Right now, cargo flaky doesn't support passing custom arguments to the test suite.
You can run all your tests an arbitrary number of times:
cargo flaky -i 30 # run your test suite 30 times and collect all failuresAnd you can record your failing test (require rr to be installed)
cargo flaky -i 30 -r # run your tests suite 30 time and record failuresoutputs:
--- Found 1 failing test ---
test: test::test stdout, 10/10 (100%)
Test binary: /home/mpostma/Documents/code/rust/cargo-flaky/target/debug/deps/cargo_flaky-2ac6d6022fd
27a8d
message:
thread 'main' panicked at 'explicit panic', src/main.rs:231:9
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
recording available in : recording_20210603213043/record_iter_1
--------------------------------
You can now run rr replay recording_20210603213043/record_iter_1 to open a gdb session and debug a recording of the failing test instance.
You can get more usage information by typing cago flaky --help.
This is still a work in progress, and lacks many features. Do not hesitate to open a PR/issue if you're missing something.
MIT