XHTML M12N: Clarify "ContentType"

Dear HTML Working Group,

  Regarding all your specifications that refer to something that is
referred to using the %ContentType; parameter entity in HTML 4.0
and subsequent technical reports, what is the lexical space of these
attributes and how are implementations required to process legal and
illegal values, are there differences between the attributes that
have this content model, are there differences between the various
specifications that refer to this type? Specifically, for the
following cases, what kind of error, if any, do they constitute and
how are implementations (which?) required to process them?

  a) the attribute refers to a type that cannot be found on
     ftp://ftp.isi.edu/in-notes/iana/assignments/media-types/
     for example, "application/xhtml+xml"

  b) the attribute refers to a type that cannot be found on
     http://www.iana.org/assignments/media-types/
     for example, "application/ecmascript"

  c) the attribute refers to an experimental type
     for example, "text/x-javascript"

  d) the attribute refers to syntactically illegal type
     for example (in case this is an illegal type),
     "björn 2004"

  e) the attribute refers only to a top level media type as
     per RFC 2045, for example "image"

  f) the attribute refers to media type, sub type and a legal
     parameter, for example

       'application/xhtml+xml;
         profile="http://www.w3.org/TR/xhtml-basic/xhtml-basic10.dtd"'

  g) as in f) but with an illegal parameter value,
     for example (in case this is illegal) 

       'application/xhtml+xml;profile="(ö)"'

  h) as in f) but with an illegal parameter,
     for example (in case this is illegal) 

       'application/xhtml+xml;x=y'

If parameters are allowed, please further clarify their effect on the
mechanisms for alternate resources (the "alternate" keyword for the
rel/rev attributes) like alternate style sheets, and encoding detection,
for example, if a link refers to a HTML document without encoding
information and has a type="text/html;charset=utf-8" is the user agent
required to use this as encoding information when decoding the document?
Is it, if not required, allowed to do that? What is an implementation
required to do if there is encoding information for the referenced
document? If there is a "charset" attribute on the same link, which of
the specification takes precedence? Is it an error if those contradict?

regards.

Received on Sunday, 23 May 2004 14:23:23 UTC