I am continuing my recent habit of following up to old postings. :-)

Just over a year ago (on 1999/01/04), Federico Lupi 
<Federico_Lupi@www.datasys.it> wrote to describe a problem uncannily
similar to one I have recently experienced:

> I have been using NetBSD 1.3.2 for some months and I am very happy
> with the system.

I have been using NetBSD 1.4.1 for several weeks at home and a bit
longer at work, and as usual, I'm delighted with it.  In both cases
this was an upgrade from 1.2, which had been in use for a few years.

> I had a very old 9 pin Panasonic printer connected to lpt0 and printed
> (very) low quality text without problems. 

I had a very old Citizen dot matrix printer connected to lpt0 which
would print OK as long as I put in a filter to slow down the output.
At one point I had had a Canon BJ200,  and was printing to it with no
problems by running PostScript output through a ghostscript filter and
text output through a "<CR><LF>" filter.

> Recently I have substituted it with an HP
> Deskjet 690C; nothing else has changed on the system. 

Recently I have substituted it with an HP Deskjet 695C.  I upgraded to
the latest ghostscript and changed the "device name" in my call to
ghostscript in my former BJ200 filter.

> My PC has a Tyan motherboard with a Cyrix P166+.

Who knows what my silly hardware is -- I can never remember such
things!  :-)  An oldish Pentium, anyway.

> Now printing is very slow both with and without Ghostscript: a single page
> can take two hours. 

Printing became very slow both with and without ghostscript: a
PostScript page could take six hours to print, a few rows of pixels at
a time, while text pages would print normally once started, but took
many minutes to start printing.

> I think that the PC is sending data to the printer continuosly
> (because the led flashes) but very slowly. The printed page, at the end,
> is correct.

I saw exactly the same behaviour.

> I have tried changing the cable without result. I changed the bios setting
> for the port (EPP, ECC) but NetBSD didn't recognize it any more at boot.

I tried another cable, but did not think to play with the BIOS.

> Finally I changed the driver from lpt0 (interrupt) to lpa0 (polled) in
> /etc/printcap and everything started working perfectly and very quickly:
> the same Ghostscript page now takes a few seconds.

Based on Federico's experience, I tried the same, and lo and behold,
sanity returned to my printing environment!  Choirs of angels burst
into song, and I went into a printing frenzy.

Now that I have returned to (relative) normality, I'm curious...

> Do you have any idea of what is happening? Why did lpt0 work with the old
> printer and not now? Is it normal that EPP/ECC are not recognized?

Did anyone ever figure this out?  I would have expected printing
performance to be worse with polling, and I would have thought that
maybe polling would also have a negative effect on system performance;
I actually noticed no effect on the latter.

Also, has anyone written a piece of software to nab whatever the
printer says back to us?  It would be kind of nice to be able to
report printer error messages and so on.


Anne.
-- 
Ms. Anne Bennett, Senior Analyst, IITS, Concordia University, Montreal H3G 1M8
anne@alcor.concordia.ca                                        +1 514 848-7606