TIFF-FX: comments on specification

["Larry Masinter" <LMM@xxxxxxx>]

In general, there are options that are optional for
'readers' but which will give poor interoperability
if readers ignore them. Two in specific 
(a careful review of the revised interop report against
the specification might yield more):

(1) Orientation != 1
4 of he 8 Orientation values, if not interpreted correctly,
would give a mirror image of what was intended, which
is bad even for printers. Screen viewers MUST implement
all that senders might send.

(2) PhotometricInterpretation=1 for profile F
If not interpreted correctly, the result would be
white-on-black reverse image.

Editorial issues:

(3) Section 5.2.2 of RFC 3249 notes difficulties with
the default gamut of profile C for some kinds of
images. This should be in the tiff-fx document.

(4) The document should have an 'applicability statement',
because a format with options that require the sender
to be aware of receiver capabilities has narrow applicability
(1-1 communication between parties that can negotiate
capabilities.)

(5) The document uses the term "TIFF" confusingly
to apply to profiles other than S and F. Reduce the number
of different terms and phrases used to name the two
file formats by their MIME types or distinct terms
("TIFF" and "TIFF-FX" but not "TIFF for Facsimile").

Other:

(6) The patent notice at www.ietf.org/ipr.html doesn't
mention JBIG patents for profile L.