Unicode improvements

Neil Schemenauer wrote:

> On Tue, Aug 30, 2005 at 06:07:44PM +0100, Hamish Lawson wrote:
> > the HTTP standard says nothing about how a
> > sender should decide which encoding to use.
>
> I think that it does (from section 3.7.1):

I supply data to Quixote and it in turn sends it over HTTP to a
client. Section 3.7.1 of the HTTP standard applies only to the latter
interaction, as Quixote does not receive the data from me via HTTP and
so HTTP's provisions about a default character encoding have no
bearing on which encoding Quixote decides to use when I have not
specified it. Quixote's only obligation is to include an appropriate
HTTP charset parameter if the encoding being used is not ISO-8959-1.

>     * If no charset parameter is specified and the content type
>       is text/*, then the client must interpret the charset as
>       ISO-8859-1.
>
> Do you agree with that?

Yes. But, again, Quixote is not an HTTP client with respect to the
data I supply it.

>     * If the client is interpreting the content as ISO-8859-1, then
>       Quixote should not encode the dat in some other charset.
>
> Do you agree?

But the client would be interpreting it as ISO-8859-1 only if this was
explicitly specified or if no charset was specified. But if Quixote
used an encoding other than ISO-8859-1 then this would be specifed in
the HTTP charset parameter.


Hamish