5 ms·
Well sure, but what are the corrective steps? 38% of bugs could have been caught sooner/prevented with types. The question to ask: What percentage of the 38% wo
by Forge36 8y ago
Well sure, but what are the corrective steps? 38% of bugs could have been caught sooner/prevented with types.
The question to ask: What percentage of the 38% would have been avoided?
Similar problems have been solved in the past: what percentage of index out of bounds errors were caught because the concept was adopted by the language?
- z3t4 8y agoCan you give more context. What language are you talking about ? What if I spent five minutes extra looking at the commit, would that have spotted the bug ? In order to give a fair conclusion, you should also look at commits without any bugs, like a blind test to reduce bias, and measure false-positives (errors that are not bugs). Maybe also use modern tools like inference engines that did not exist back when the bug was made. And compare with other techniques like unit testing to see which one is more effective, and to see if there are any synergistic effects.