Safe code as an I/O requirement

[piotrklos via Digitalmars-d]

On Sunday, 28 May 2017 at 16:58:53 UTC, aberba wrote:
> https://lwn.net/Articles/708196/
>
> (...)

Intuitively it would be much better because overwhelming majority 
of the code can be written with @safe, but bounds checking would 
have to be switched off for some plugin code for performance 
reasons, so it would not be 100% secure.