-----BEGIN PGP SIGNED MESSAGE-----
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.
Agreed. The definition of each content-type should describe its
canonical format. Because the format of text varies from one system
to another, the definition of any text/* content-type should state how
newlines are represented in that content-type. But the basic need to
define the canonical format exists for any content-type that must be
represented differently on different systems.
For the case of application/postscript, the canonical format is
adequately described by the published PostScript language definition
referenced by RFC 1521. It happens to include several ways of
indicating end of line, but all PostScript interpreters should accept
all of these.
So for a PostScript file that uses CR, LF, or some combination of
these to delimit lines, no "canonicalization" is necessary. But for a
PostScript file stored on a system that used some other means of
delimiting lines (e.g. record length counts), it *is* necessary to
convert that format into something that uses CR and/or LF.
-----BEGIN PGP SIGNATURE-----
-----END PGP SIGNATURE-----