John.Cowan wrote:
>>Here-strings and literal strings both have the same possible range
>>of source representations.
>
>
> Actually not.  A here-string can only represent those characters which
> can be represented in the encoding of the Scheme source.

The latter sentence is saying the same thing I was, but I used the term
"source representation" as shorthand to say that.

I'm saying that both the source representation of a here-string, and the
source representation of a quoted string, can only include those
characters which can be represented in the encoding of the Scheme source.

(I'm talking about the representations of these strings in source code
or reader data, not about the Scheme strings which the source
representations denote.)

> For example, if
> the encoding is ASCII, a here-string cannot represent non-ASCII characters.
> A quoted string (as extended by SRFI-75) can represent any Unicode character.

Right, but in the source representation itself, both here-strings and
quoted strings necessarily have exactly the same possibilities and
constraints.  However, they are interpreted differently by the Scheme
implementation.

> Exactly the same limitations apply to XML's CDATA sections vs. ordinary
> character content: no escaping required, but none permitted either.

Yes, although as I pointed out, here-strings can support any sort of
escaping you desire, merely by processing them through a procedure which
performs the desired escape sequence translation.  Ordinary literal
strings can be thought of as here-strings which are always processed
through a specific procedure, the SRFI-75 escape sequence translator.

> It's explicitly out of SRFI-75's scope what encodings of source code
> R6RS implementations must accept.

Yes.

Anton