Encodings and HTTP [Semi-OT] [was RE: Azeri [Re: LANGUAGE TAG REGISTRATION FORM]]

[Jon Hanna]

> >> At 12:54 03/04/15 +0200, Harald Tveit Alvestrand wrote:
> >>
> >> > a standards conformant web page served by HTTP that does not
> >> > give any character set information is in ISO 8859-1.
> >> > ...
> >>
> >> This is not a pessimist's view, it's a realist's view.
> >> And it is what browsers actually do (if they don't do
> >> heuristic charset detection).
> >
> > Seconded, it's just reality and the ISO 8859-1 default has never been
> > anything but wishful thinking (except when it was the only thing).  This
> > default should be removed from the HTTP spec before it moves to Full
> > Standard.
>
> nah, we'll just make sending text/html without charset= illegal.

That would of course be a de facto removal of the ISO 8859-1 default. The
default is not only improperly implemented, but is of decreasing value with
the rise of XML where the documents default encodings clash with HTTP's.

IIRC however there were issues when RFC 2616 was being written with some
clients being confused by the charset parameter. Perhaps this is less of an
issue now, and charset can be made compulsory on all text-based content? If
so then I think all