Re: [xml-dev] Text/xml with omitted charset parameter

[Rick Jelliffe <ricko@allette.com.au>]

 From: "Bjoern Hoehrmann" <derhoermi@gmx.net>
 
> So, who tells me I
> am wrong and text/xml documents without charset parameter may still be
> UTF-8 encoded (and use non-ASCII characters)? Apache uses text/xml as
> default type for .xml documents, are they asking for interoperability
> problems or what?

The only ways out of encoding hell are:
  -  no data interchange between people on different systems
  - everyone adopt a common character reportoire and encoding (e.g. UTF-8)
  - everyone label their data, and make all protocols strictly require 
      accurate labelling

The first of these is silly. The second can only come slowly, if it comes.
XML supports the last two: a common character repertoire and 
required labelling.   

It is no suprise that this falls apart as soon as we get layers which do
not transfer encoding information.   But that is not a sign that
labelling is bad, it merely means that the other layers need to be
attended to. 

The impracticality of the defaulting rules that MIME uses is the
culprit.  It covers up an issue that should be explicitly handled
and should have widespread awareness. (It is nice to see things
like a text open dialog box on Mac OS X  having an encoding
selection option, in this regard. Shame Java doesn't have it
standard.)   

"Text" entities do not exist. There is only text in a particular
encoding.  It is the long-standing and slack policy of 
defaulting to the locale's character encoding has meant that
our API infrastructure do not provide enough mechanisms
for passing information about character encoding. (IMHO, the C
type byte and char is the big villian; the Java approach
of specifically saying "a character is Unicode using UTF-16"
is a big way out of the problem, but even Java's API
did not at first take the external-internal encoding transition
seriously. )

We take it for granted that our programming languages should
have a data type for "float" which is different from "integer",
and we are taught the difference in computer courses.
Yet I believe that most IT students are not taught anything
about character sets or character encodings. 

Cheers
Rick Jelliffe