Yes. But the point is any unsafe code can be reviewed separately and be subject to more stringent checking than the majority of code which is safe. I've been l

Yes. But the point is any unsafe code can be reviewed separately and be subject to more stringent checking than the majority of code which is safe. I've been looking at the HTML parser they are writing (servo/html5ever) which is incredibly impressive and parses the hideous Daily Mail homepage in my tests in < 30ms. It uses the tendril library which contains lots and lots of unsafe code, but this library is set out to do a very specific thing which is make parsing strings much much more efficient. This is a design tradeoff that is desirable; as long as tendril can be shown to be robust it's a good decision over a slower parser.