Andreas Fackler
Andreas Fackler
The `DynamicHoneyBadger` tests currently don't include adversaries that send messages. They should be extended in the same way as the `HoneyBadger` tests: #401 Or should they? What's the best way...
Tests for `ThresholdDecrypt` are still missing. These should probably be similar to the `ThresholdSign` ones: #399
The current `ThresholdSign` tests have been adapted from the previous `Coin` tests, i.e. they are mainly concerned with the pseudorandom distribution of the signatures. We should add proper tests for...
The `SyncKeyGen` tests currently don't include invalid messages. We should add tests with an adversary that sends incorrect messages, and cover different cases, e.g. with invalid `Part`s, partly invalid `Part`s,...
The Subset test adversaries currently only reorder messages but don't send any messages themselves. They should send random messages, ideally with different, somewhat intelligent random distributions that make it most...
Add an adversary that sends random (but "valid looking") messages. The distributions should be general, but somewhat intelligent, so that they are more likely to uncover bugs. Also, the original...
We have two integration test frameworks at the moment, in `tests/net` and `tests/network`. We need to port all the tests to the former, and remove the latter: * [x] `binary_agreement_mitm.rs`...
While it's impossible to cover every conceivable sophisticated attack strategy, we should send at least one fake message of each variant in some test, i.e. for each algorithm have a...
When I run a test with `HBBFT_TEST_TRACE=true`, it produces ten `net-trace_*.txt` files with cryptic numbers. We should document in `tests/README.md` which of them is the last one (corresponding to the...
* [x] The existing tests should verify that none of the correct nodes are ever reported as faulty. * [x] `tests/network' (deprecated) * [x] `tests/net` * [ ] Verify that...