: > I have a SCO Box using a Cisco Router as a terminal server with 8 async
: > lines on it. It also goes out over a 56k line to another Cisco serving
: > terminals at another site. I'm trying to hook up one of the async lines
: > to a printer. The routers support LPD. My two options are:
: > 1. Use LPD on the SCO box.
: > 2. Address the port directly through the SCO Print Spooler system.
: > If anyone had any experience with this, I'D appreciate your thoughts.
: As a general rule I'd use LPD. We make Terminal Servers and Printer
: Servers and I would recommend this although we have customers who
: use the second option well as others who use RCP.
[some good reasons to use lpd snipped]
I'll chime in on the negatives for lpd: The problems I've had spring
from the fact that lpd assumes there's a "real computer" at the other
end. For instance, lpq will go chasing off and ask the print server
about the queue status. Many will not respond sensibly. Some (at least
HP 4si) will simply hang, leaving the hapless user who just wanted to
know his job status to figure out how to break out; and who can help him
if he's using some gui interface?
Dumb print servers are often single-threaded. Client lpd spoolers may
get irritated when a host that claims to support multiple queues
refuses to accept jobs for minutes or hours at a time. Sometimes the
print server will get "locked in" to serving a single queue, serving job
after job, while not accepting jobs on other queues.
The most irritating problem with relatively dumb print servers trying to
serve lpd is that in the BSD lpd world, filters are applied on the last
system to handle a job, the system that actually runs the printer. Most
print servers have at best a limited set of canned filters, not an
expandable, programmable set. It is often the case with lpd that one
has different queues for different page orientations or font sets or
whatever; the printer setup commands are added to the print stream in
the output filter. With most print servers, no output filter, so you
end up having to torque the already difficult BSD print system some
more, treating certain printers specially.
One "solution" for some of these problems is to have pseudo-queues whose
output filters end with "| lpr -Pprintserver_queue". This lets you
apply filters, at the cost of added complexity (more queues) and loss
of job identity (lpq claims your job is gone -- instead, it's gone
anonymously to another queue).
My pet peeve is the whining of the users of the HP 4si, who complain
multiple times a week about the banner page it insists on adding; there
is no printcap or equivalent, and no way for me to shut it off, as I
have explained many times.
Enough. Choose your own poison. ... rb