Implementers Guide comments

["McIntyre, Lloyd" <Lloyd.McIntyre@xxxxxxxxxxxxxx>]

Tamura-san,
I would like to comment on two items, 1) ambiguity with regard to the
definition of "word" size and 2) comments on Section 5 of the Implementers
Guide. Please see below.

Thank you,
Lloyd


1) Word Size

The requirement to begin an IFD on a "word" boundary is specified in TIFF
6.0. TIFF-FX is simply reiterating this requirement. Since TIFF 6.0 is the
based reference for TIFF-FX, any clarification of ambiguity associated with
the definition of "word" size should be done in TIFF 6.0, not TIFF-FX. It
may, however, be appropriate to add an item to Section 5.1 of the
Implementers Guide giving guidance on interpretation of this ambiguity. I
propose revising Section 5.1 of the Implementers Guide V4.0 to read as
follows:

5.1 IFD Placement in TIFF file & Profile S Constraints

a) An IFD is required, by TIFF 6.0, to begin on a word boundary, 
   however, there is ambiguity with regard to the defined size of
   a word. A word should be interpreted as a 2-byte quantity. This
   recommendation is based on examination of Figure 1 and the
   definition of IFD Entry, Bytes 8-11, found in Section 2 of 
   TIFF 6.0.

b) Low memory devices, which support resolutions greater than the
   required Profile-S, may be memory-constrained such that those
   devices cannot properly handle arbitrary placement of TIFF IFDs
   within a TIFF file.

   To interoperate with a receiver that is constrained, it is strongly
   recommended that senders always place the IFD at the beginning of
   the TIFF-FX file when using any of the Profiles defined in RFC 2301.


2) Section 5 of the Implementers Guide

Most of errors document in Section 5.2 were NOT experienced during
interoperability testing, as stated.

Recommendation
a) Delete current items 5.2 h, 5.2 i and 5.2.1 b as they are unfounded and
the recommendations will create incompatibilities.

b) Divide Section 5.2 into two distinct sub-sections, clarifying those that
are observed interoperability errors.

Example:
New 5.2.1 Errors encountered during interoperability testing
This could include items such as current 5.2 a and b.

New 5.2.2 Items that require particular attention
This could include items that were not observed during interoperability
testing.

At the start of this sub-section it may be appropriate to highlight the
general TIFF guidelines of ignoring field values that are not recognized or
are not supported. 

Items that may be included:
current 5.2 c, d, e, f, g
current 5.2.1 a, c, d, e, f, g
current 5.2.2 a, b, c, d
       Revise 5.2.2 C to read as follows: 
       "The strip offset + StripByteCounts does not point 
        outside the file" 
current 5.2.3 a, b, c
current 5.2.4 a, b
       Revise 5.2.4 b to read as follows: 
     "Implementers should make sure when generating a TIFF Profile M
      that the compression used for the layers is correct. Possible
      errors may include the Mask layer not being compressed with a
      binary encoder and the Background and Foreground layers not
      being compressed with a multi-level (e.g. color) encoder."