3 ms·

My standpoint is that: - Less exploits is an improvement in frequency - More security is a question of limiting impact

• dbaupp 10y ago
  Risk analysis is typically done including both the probability of a problem and the impact of the problem. Not including the former in an analysis will result in spending vastly disproportionate amounts of effort on exceedingly rare problems, e.g. the consequences of a 1000ft tsunami hitting New York would be huge, but we don't build massive sea walls because it is also ridiculously unlikely. This applies perfectly well to digital security too: reducing the probability of an exploit is just as important as limiting impact if it does happen, which is why (for instance) Mozilla is doing both.
• __david__ 10y ago
  I would say 'security = frequency * impact'. Reduce either one and you're doing better. Reduce either of them to zero and you're golden.
• staticassertion 10y ago
  I don't think I want to get into what is or is not security, but what I will say is that rust's security does not interfere with safety net features, like sandboxing. I've found that sandboxing rust is quite easy, as easy as C or C++. I find it much easier than Java/ Python, which have runtimes that can often make things difficult (you certainly would not want to use seccomp in either). Take a look at DropBox's Brotli - it's a great example of this. It makes use of seccomp, but seccomp is not available on all platforms, so it also relies on rust's safety.