We have two important issues here. One is when newline canonicalization is
required, and the other is what to do about PostScript.
Since Microsoft RTF basically is a "decorated text" format just like
text/enriched, wouldn't it make more sense to have it (re)registered as
text/rtf -- just like it originally was intended?
The fact that data does not have a top-level type of 'text' does not
indicate that the type does not use newline canonicalization. I think it
could be argued that the fact that data HAS a top-level type of 'text' does
not necessarily indicate that data does require newline canonicalization,
but I would rather not think about that.
I think that there should be a requirement that any type requiring newline
canonicalization explicitly say so in its official description.
Furthermore, any type that is borderline but does not permit newline
canonicalization should specify that, too. It might be simplest just to
require that every single description give a yes, no, or don't care to
newline canonicalization.
Now, as for PostScript:
It is quite possible to construct a file compliant with PostScript level 1
or 2 that will be rendered INOPERABLE by newline canonicalization done
without knowledge of the PostScript language. I have in fact seen such
files myself more than once in the course of dealing with real PostScript
generated by real applications.
Furthermore, such canonicalization is NOT (in general) reversible. You
cannot (with reasonable convenience) repair a file that has been improperly
canonicalized.
Therefore, if we require canonicalization, we will render
application/postscript unusable for some set of real PostScript files.
Real PS interpreters are not supposed to care about such conventions.
Users with interpreters or utilities that do care should do the
canonicalization themselves (possibly with the aid of their MUA). That
way, you leave the transformation at the final step, where it can be
reversed (by rereading the message or simply using the raw file instead of
the processed one) if canonicalization damages the file.
I therefore think that the proper thing to do is not to canonicalize
application/postscript for transmission.
--
Steve Dorner, Qualcomm Incorporated
"There's nothing wrong with you that can't be cured
with a little Prozac and a polo mallet." - Woody Allen