> We could regard memory safety as a great success in this area that programmers can easily appreciate. No one wants the overhead on current hardware but C can

> We could regard memory safety as a great success in this area that programmers can easily appreciate. No one wants the overhead on current hardware but C can be made memory safe with bounds checked pointers (although not particularly type safe without introducing a lot of code incompatibilities). It's important to distinguish actual language advantages from indulging the common fetish of wouldn't-it-all-be-better-if-we-rewrote-it. There has been fairly little rigorous study of programming factors leading to more reliable software (esp in the kinds of environment most software we use is created and used in, as opposed to say aerospace). Wouldn't it be sad if enormous amounts of effort were spent rewriting everything in a new language only to later discover that other properties of that language, such as higher levels of abstraction or 'package ecosystems' that cause importing impossible to audit millions of lines of code to get trivial features, lead to lower reliability than something silly like using C on hardware with fast bounds/type checked pointers?