Re: Printers Revisited

From: Alan Dechert <alan_at_openvotingconsortium_dot_org>
Date: Mon May 24 2004 - 12:26:52 CDT

David,

> Alan has mentioned this 8 seconds several times; it always struck me as
> somewhat unrealistically quick. I looked through his recent notes, and
> he had indicated in one of them that his 1300 was a Postscript-enabled
> printer (I don't know if that's an add-on or a standard feature).
>
Postscript Level 2 emulation.

> As I mentioned several other places, rasterization of postscript (by
> Ghostscript) will probably take about as long, or slightly longer than,
> the actual printing. I also, elsewhere, mentioned some potentially
> advantages of requiring a postscript printer; but doing so is certainly
> not currently a consensus position.
>
> Anyway, I decided to do my own timing--on my not-terribly-recent
> machine and printer. I.e. it's the HP LJ-6L that I mentioned, and the
> computer is a P3 (733Mhz, I think). Printing a ballot takes a total of
> 28 seconds for me. During the first 15 seconds or so, the computer is
> rasterizing the postscript. That is, the "active" light on the printer
> doesn't start flashing until about that time (it flashes while it's
> receiving data). To a voter, this gives no indication at all that
> anything is happening (a message on screen could do so though). Then
> another 13 seconds or so to actually print.
>
Okay, I have to grant you some points on that one. The 8 second figure was
based on printing out the Word mock up I did of the 44 race ballot Steve
Chessin pointed out.

Here's output from a DOS box where I marked the time then copied to lpt port
then marked the time again. Give a half second or so to arrow up and hit
enter to get the time the second time, it comes to about 17 seconds. This is
on an Athlon 800 PC with the HP 1300.

********
Current time is 9:48:38.24a
Enter new time:
D:\>copy ballot.ps lpt1:
        1 file(s) copied
D:\>time
Current time is 9:48:55.87a
Enter new time:
********
So the watermark appears to double the time required.

> The difference between the 6L and the 1300 can easily be 13 vs. 8
> seconds. But the initial rasterization is pretty much going to take a
> few seconds, unless we require a postscript printer.
>
I just tried the same thing with my 10-yr-old HP LJ 4, which has a
postscript module and 26 meg of ram. 52 seconds... That's why we used the
1300 at the demo.
*********
D:\>time
Current time is 10:12:16.80a
Enter new time:

D:\>copy ballot.ps lpt1:
        1 file(s) copied
D:\>time
Current time is 10:13:08.98a
*********
It takes 3 times as long on that old printer (very hard to see any
difference in print quality, btw, even with a magnifying glass--Wright Bros
photo a little darker on the LJ 4).

I'd like someone to show some time trials printing ballot.ps on an inkjet.
I don't buy Karl's strawman about how it gets started quicker--the total
time to print is what matters. All of this may mean we need to simplify the
graphics. I really like the watermark we have, though. I still think that
in production the HP 1300 could produce a ballot in 8 seconds with simpler
graphics--maybe using PCL instead of PS.

Like a lot of things we're looking at now, there are tradeoffs. Trials are
needed to examine tradeoffs.

Alan D.
==================================================================
= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
==================================================================
Received on Mon May 31 23:18:01 2004

This archive was generated by hypermail 2.1.8 : Mon May 31 2004 - 23:18:16 CDT