ietf-822
[Top] [All Lists]

Re: Language tags and 10646

1993-03-10 03:50:14
John has been suggesting that perhaps
language tagging could be considered to be "markup", and therefore not
part of "text/plain".  Is he right?

Just to clarify, I think there are two logically-possible ways to
identify language information:
  -- in the header somehow, whether as funny words made part of
     'charset' or as separate parameters or fields, but, in any event,
     applicable to a whole body part.
  -- in the body part, such that one can specify the point at which we
     switch from one language to another.

I'm convinced the latter is "markup", simply because I want to be able
to build automata to take advantage of it, and I don't want automata
looking heuristically at putative plain text to try to figure out if it
really isn't "plain".

I think that we will need language-markup (the second case) sooner or
later.  I don't think it is a problem we need to solve in-line with
RFC1341bis.

By contrast, if we would benefit significantly from per-body-part
language markup, then it seems to me that we should straighten that out
and define fairly quickly, as it has significant bearing on how we
think, operationally, about "character set".   Whether the would be real
benefit there depends critically on the question of how much text we
expect to have floating around in languages that can be adequately
disambiguated by per-body-part language identification (or with no
language identification at all).

If we really expect most messages to be subtly multilingual, then there
is no point bothering and, in practice, text/plain and very extensive
character sets are incompatible.

   --john

<Prev in Thread] Current Thread [Next in Thread>