Printing PDF files -Reply

[Rich Greenfield]

Michael Huber <mhuber1 at umbc.edu> 03/10/97 12:55pm  wrote:

UMBC began placing some reserves, mostly professor generated
material, on the Web in Adobe's PDF format.  Some of the documents
are 20+ pages and can run up to 1.5 MB's in size.

The problem I am running into is that they take 20 minutes or more to
print.  (i.e. it takes 20 or more minutes for the file to be fully entered into
the print queue, from which out pay 4 print system prints) ....

Does anyone else deal with PDF files of this size?  Can anyone offer
suggestion on how to reduce the printing time? 

********************************

Rich Greenfield replies:

We've been stuggling with similar problems here at the Library of
Congress for almost a year. We call it the "big file" problem and it
involves not only printing PDF documents but downloading and
manipulating large, multi-megabyte files in various formats. The big file
problem is a serious obstacle to the federal government's efforts to go
electronic since Congressional bills, hearings and reports, even in ASCII,
can be extremely large (check out some of the GPO ACCESS files) and
difficult for the average user to capture and print. 

As to the PDF printing problem, here is my most recent experience. Our
network environment is similar to yours, but our new generation of PCs
are all  Pentiums with16MB RAM and running Win95.  The printer used
below was an HP LaserJet4 Plus.

>>> Rich Greenfield 02/20/97 03:32pm >>> Choking On PDF

I just returned from a trouble call  which may be another wrinkle on the "big
file" problem.  Using the Netscape print command, the user was trying to
print PDF documents of various sizes from the GAO website without
success. He was getting a variety of error messages, most of which had
to do with insufficient memory. The user has one of the new Compaq
4000 Pentiums.

We closed WordPerfect and GroupWise and then tried to print (using the
Netscape Print command) the largest of the documents and the whole
setup just froze, without any message. Then we rebooted and tried
printing the smallest document (20 pages) using the Acrobat printer icon:
this worked. Then we tried the next largest document (60 pages) using
the Acrobat printer icon and got an error message:

"There was an error writing to LPT1: for the printer  ... may be a problem
with the network, or the password may be incorrect."

The strange thing is that this message comes up on top of the "Printing
page 19, etc." message, which keeps on counting all the pages to the
bitter end. From this I would conclude that the buffer in the PC is
sufficiently large to handle the PDF pages and send them to the printer,
but that the printer does not have adequate memory and is simply being
overwhelmed.

Does anyone have other explanations? Can anyone say what the
difference is when one prints a PDF document using the Netscape
pulldown print command and using the Acrobat printer icon?

Thanks for any insights.
****************************************
One of our network specialists replied:
	            Once again We run into the BIG FILE problem. Most users
are limited to
5-10MB of network space. When mail messages with attachments are
sent and not deleted, they take up space on the network. The user might
have 1-2MB of free network space (used for spooling network print jobs
and storing important files) allowing the 20 page document to print. But
when you print the 60 page document, it puts you over the amount of free
space available. PDF files generate large print jobs because all
formatting and graphics are included in the job. Ask system administrator
to release the files attached to E-mail messages for that users. This will
free up space to allow the printing of the large files.
*********************************
Following up on this hint, I did a little testing with PDF documents [one
thing I fogot to test was dropping the dpi from 600 to 300]:

Test 1

Taking the second document that the user wanted, a 735k PDF file, I tried
printing it on the AU FrontDesk printer. Watching it load with the Print
Console, it turns out that the 60-page PDF document becomes a 10.88
MB file when it loads in the printer (a ratio of more than 10 to 1). 

Sending the pages to the printer only took a minute, but preparing them
for the print process (a kind of unpacking, in that the file size leaps when
all the instructions to the printer are added) took an additional three
minutes. Printing took still another 18 minutes, for a total of 22 minutes
before I had the document in hand. 

The printer choked on a PDF image on page 6 for more than 12 minutes
of that time and it was also the image which inflated the file size so
tremendously (see test 2 for a more common document inflation ratio of 5
to 1 when there are no images). By printing the page with the image
separately, I determined the image alone was 7.29 MB (and it isn't even
very attractive - I've posted it on the bulletin board).

Test 2

Next I took the first document in the user's list, a 1.01 MB file that
expanded to 4.88 MB sent to the printer and which printed in 12 minutes.
What a difference a single image makes!

Conclusion

Users will not be able to know in advance when a PDF file will
successfully print on their network printer because:

1) they can never tell in how large a file will become when it goes through
it's self-initiating prep-process in the user's assigned network space
before being sent to the  printer. However, the existence of any images in
the document as viewed with Netscape will be an indication that the file
will be larger - probably much larger;

2) most users will not know how large their network space allocation is
and will only learn through experimentation how many pages (without
images) they can print at anyone time. Other factors, such as multiple
fonts, may prevent an accuarate estimate of what can and cannot be
successfully printed.

I leave the ultimate solution to the PC and Network Teams. We can
selectively allocate more network space to users who request it; we can
recommend that users print large documents in small sections; we can
look for a technical work around so that Netscape print commands go to
a very large, common network file space; or we can have workstations in
each division that have large or unlimited network space assigned to
them and are available to all division staff for printing large documents.

Any other suggestions?

************************

This resulted in the following comment back from our network specialist:

It's possible for each TSA/Division Administrator to reset the permissions
for the mail files weekly. But the that would only clear up to the original 10
MB of allotted space, minus any other files residing in the users HOME
directory. The issue is server space. If there are 70 users in a division,
and each are allotted 10MB, that's 700 MB. I do not know when the
servers will be upgraded, but when they are, I could see upping the limit to
25 MB. That would probably be enough to handle the printing problem.
***************************************

We have always hoped that implementation of byte-served PDF files
(what used to be called Adobe's Amber) would enable users to click on  a
hyperlinked table of contents that is delivered to the screen first and then
go immediately to the relevant sections of large PDF documents which
could be selectively printed.  The problem here is that most PDF
documents have not been rebuilt for byte-serving; furthermore, even if they
were, many users want the whole, humongous document, even if it is not a
handy and convenient as an original, printed publication might be. I don't
know of any simple solution to this problem and have the feeling that it is
only going to get worse.