Hacker News new | ask | show | jobs
by Jtsummers 3267 days ago
What are the other approaches for preventing bugs you have in mind that are novel or increasing in application?

We've done testing since the start. Still many bugs. We've done ad hoc modeling (behavior driven) for years with some improvement. Formal methods aren't popular but are successful at least at some (small to the low end of medium) scales or within portions of large scale systems. Type systems can and have been used for codifying concepts from all of these into the program semantics. So I'm not sure how it is that they don't help.
2 comments
naasking 3267 days ago
> We've done testing since the start. Still many bugs.

Exactly, as the OP clearly doesn't understand, but every researcher in programming languages does, test can only prove the presence of bugs, it cannot prove their absence. Types can.

link
z3t4 3266 days ago
what ive found most effective at finding bugs is to have two systems, the data represented in two different type systems, then cross check them on crucial intervals. the second most effective at finding bugs is to check function parameters both when entering the function and when returning from the function, for bound, range, etc. and third... study and understand the code, actually writing code that is easy to coprehend. then tests are very good at preventing regressions. despite this, there will still be bugs though smile so debuability, effective debugging is really important!
link