Re:The perils of Lisp in production
by brlewis on Wednesday April 04, 2012 @09:11AM
(#3797348)
If you presume that it's OK to rely on compiler errors rather than attempt to exercise all code paths, then yes, Common Lisp, Scheme, Clojure, JavaScript, PHP, Python, Ruby, Perl, and bash will be more perilous than C, C++, C#, Java, ML and Haskell. (Which of the former group are Lisps?) I disagree with the presumption.
[ Parent
| on HN ]
> more perilous than C, C++, C#, Java, ML and Haskell
Right. Unless you use features to escape the type system (such as C's casting, Java's reflection, or C#'s dynamic) which every non-trivial program I've ever worked on did.
I think the most dangerous piece in such system is, indeed, the programmer who thinks the compiler will save him from himself.
Much less dangerous, in my experience, than the programmer that thinks he can keep a huge system all in his head. I've moved back to static languages after over a decade with dynamic languages and I couldn't be happier to let the compiler take care of a lot of low-level consistency issues so I can focus on the code.