No, the trailer dictionary is not necessary for the interpretation of the
PDF for this PDFax format. The trailer is only for Acrobat et. al. to read
the file. The PDF reference's notion of reading from the end doesn't apply
to this new format. And, yes, as I've laid it out it will work.
The process would be something like ...
1) Cache all objects found until the first 'page' object it located.
2) Print the page.
3) Dump the cache.
4) Repeat for the next page.
I'm still considering whether it makes sense to add a marking tag "/PDFax"
to all objects that are necessary to render the page as a FAX, that way
extraneous objects can be bypassed when scanning for the necessary objects
to cache... but I want to consider other options first.
Hope this clears things up.
-Rick
-----Original Message-----
From: John Pulera [mailto:jpulera@minolta-mil.com]
Sent: Wednesday, July 31, 2002 7:34 PM
To: Rick Seeler
Cc: IPP-Fax Group
Subject: PDF question
Importance: High
Rick,
One of the requirements for IPPFAX is that the data format be
"pseudo-streamable" so that a receiver would not have to parse the entire
file before being able to render the first page. The group agreed that
allowing random access for no more than a page worth of data at a time would
be acceptable. However, in the sample JBIG-2 encoded pdf file you posted, it
would seem that a parser would have to jump all the way to the end of the
file to read the trailer dictionary first so that it can accurately
interpret all the indirect objects at the front of the file. In section
3.4.4 of the PDF Reference, it even states, "Applications should read a PDF
file from its end."
Is it possible for a parser to still accurately interpret the page &
resource objects without having first traversed the file trailer and catalog
dictionary? Or would linearized PDF be the only other feasible alternative?
Thanks,
John Pulera
This archive was generated by hypermail 2b29 : Thu Aug 01 2002 - 12:13:04 EDT