强行迫使程序因输入不同而崩溃的策略是什么？

Question

I recall hearing about an effective testing strategy that helps find bugs in software. It involved using another program to generate various inputs. By trying thousands of different inputs, we hopefully cause the program to go down an execution path that it does not normally go down. The goal being to get programs to crash, or throw exceptions, giving us insight to improve the quality of software. I believe it was even had some kind of heuristic, where it was able to know whether it was getting somewhere different in the code based on the inputs it was sending.

I can't learn more about it, because I can't recall what this method of testing was called.

If it helps, I think it was mentioned in regards to C++. Maybe it was a compiler extension. I remember the speaker showing how he would write some kind of test rule, then the implementation, and then brute force it to show edge cases that he missed.

Answer 1

这称为模糊测试，您可以在此处找到更多信息： https : //en.wikipedia.org/wiki/Fuzz_testing