-----------------------------
Application message id: 04044171211991/8251 X400
Grade of Delivery: Normal
-----------------------------
VMSmail To information: MUVAXA::MRGATE::"mci
_mta::*emsinternet::*mbx1ietf-822(a)dimacs.rutgers.edu::su=822
_extensions"
Sender's personal name: John C Klensin
Nathaniel writes:
I think the question is whether shifting to a new part each
time you shift languages is too heavyweight. I think the answer
is that something less heavyweight might be nice, but that any
software that's smart enough to let you compoes multi-lingual
text gracefully will easily be made smart enough to create a
data stream of multiple parts when necessary.
I can create multilingual character streams with things as
simple as several versions of kermit. It is lots easier than,
e.g., audio or pictures. Text in one language with a few
explanatory words of another is *very* common in some fields, as
is use of characters in, e.g., mathematical discussion that
aren't in, again, e.g., ASCII or ISO 8859-1. Awful lot of
overhead if I have to have an extra body part for a few
characters. And, yes, humans do read the raw datastream or very
close approximations to it.
Again, one fairly simple class of solutions is to follow TEI a
bit further and push on SGML.
Another, much as I dislike the idea personally, is to decide
that *this* kind of document structuring is an ODA problem,
figure out what happened to that WG (I just realized that they
didn't meet in Santa Fe and weren't in the report just posted to
the IETF list), and make sure we can accomodate ODA body parts
in a reasonable fashion (if we are going to rely on ODA for
things like this, and if ODA-over-email rates a WG, then it
probably should not be buried in some application/maybe-you-
can-copy-and-if-you-can't-we-don't-care type).
--john