Reducing the cost of autodecoding

[Patrick Schluter via Digitalmars-d]

On Saturday, 15 October 2016 at 18:40:11 UTC, Uplink_Coder wrote:
> It can also be written like this producing smaller code.
> But it the cost of slower decoding.
>
> dchar myFront(ref char[] str) pure
> {
>     dchar c = cast(dchar) str.ptr[0];
>     if (c & 128)
>     {
>         if (c & 64)
>         {
>             int idx = 0;
>             int l = charWidthTab.ptr[c - 192];
>             if (str.length < l)
>                 goto Linvalid;
>             c = 0;
>           l--;
>             while(l) {
>               l--;
>                c |= str.ptr[idx++];
>                c <<= 6;
>             }
>             c |= str.ptr[idx];
>
>        }
>         else
>     Linvalid : throw new Exception("yadayada");
>
>     }
>     return c;
> }

Just a question. Do encoding errors not have to be detected or is 
validity of the string guaranteed? Wrong continuation bytes or 
overlong encodings are not detected by this routine.