Hello,
I’m trying to wrap my head around how to use the fuzzer.
My sadness is that I can’t get the fuzz loop to run over my build - either on linux or macos, with 0.14.
The tests just run once and exits with a success - no fuzz loop.
Full repro at src/fuzz_stateful.zig.
Here’s a simplified version:
fn runStatefulTest(_: void, input: []const u8) !void {
// allocate a copy of the data that we can use
// Initialize PRNG with the fuzzer data
var prng = FinitePrng.init(fuzz_data);
var random = prng.random();
const commands = [_]state.Command(Model, System){ }; // list of commands here
// Initialize model and system
var model = Model{};
var sut = System{};
var cmd_seq = state.CommandSequence(Model, System).init(...); // Create command sequence
// Run the stateful test
const result = state.assertStateful(Model, System, &cmd_seq, &model, &sut, .{ .verbose = false });
try std.testing.expect(result == null);
}
// Add a fuzz test that will be picked up by the build system
test "fuzz stateful testing" {
try std.testing.fuzz({}, runStatefulTest, .{});
}
Based on the online demos I’ve seen, my expectation was: zig build test --fuzz triggers the fuzzer to keep calling the test function, until it returns an error. Yet, I don’t see any loop running - the test just succeeds and exits immediately.
Could it be that the issue is in my build.zig, or am I grossly mishandling the API here?
…