When the fuzzer is restarted, it currently weights all seeds equally despite use the aggregate call count as it does at runtime. The thinking is that an input that took the longest to discover. Using the timestamp of the corpus sort of approximates using the call count at runtime (time and call count both increase as the fuzzer runs). This is how Echidna does it fwiw
When the fuzzer is restarted, it currently weights all seeds equally despite use the aggregate call count as it does at runtime. The thinking is that an input that took the longest to discover. Using the timestamp of the corpus sort of approximates using the call count at runtime (time and call count both increase as the fuzzer runs). This is how Echidna does it fwiw