> There seems to be an inconsistency here.  Can (c_start >= c_len) be
> relaxed to c_start > c_len?

Done. `substring' was a useful reference.
 
> It would be nice to document if it's an open, closed or half-
> open/closed range.  E.g. see the documentation of 'substring':

Done.

> It seems a bit weird to support [start] and [end] for utf8->string but
> not for utf16->string and utf32->string.

I will gladly implement those too. I wanted feedback on the code before
I implemented all of them, which you provided, thank you. I still want
thoughts on the name of the new C function. If that is okay, then I can
implement the rest following the same name pattern.

>> +      c_start = scm_to_size_t (start);
>
> This seems suboptimal because if start > SIZE_MAX,
> then this will throw an 'out-of-range' exception without attributing
> it to 'utf8->string' (untested).

Switched to scm_to_unsigned_integer, that does bounds checks. This is
what `substring' does.

The updated patch is attached.

~ Vijay