Path: utzoo!attcan!uunet!husc6!bbn!rochester!rutgers!bellcore!faline!thumper!ulysses!andante!alice!debra
From: debra@alice.UUCP (Paul De Bra)
Newsgroups: comp.unix.xenix
Subject: Re: SCO XENIX/386 and a slow lp
Keywords: XENIX 386, lp is real slow
Message-ID: <8207@alice.UUCP>
Date: 16 Sep 88 03:34:36 GMT
References: <5470@killer.DALLAS.TX.US> <908@riddle.UUCP>
Reply-To: debra@alice.UUCP ()
Organization: University of Antwerp
Lines: 34

Andrew Beatty says in his reply to Tom Armistead:
>In article <5470@killer.DALLAS.TX.US> toma@killer.DALLAS.TX.US (Tom Armistead) writes:
>>I have just installed 386 Xenix (SCO) and am having some problems
>>with the lp spooler. It runs **REAL** slow, printing about 1 line
>>every 1 to 2 seconds, even when there is absolutely nothing else
>>happening on the system...
>
>I'll bet my bottom dollar that you are running 2 parallel printers.  This
>problem is known and the fix is described in the release notes of your
>(HW) manual.

I'll bet Andrew loses the bet. There is a REAL problem with the printer
driver:
1) an interrupt driven driven driver (the default) only works fine on some
   combinations of printers and systems, depending on very critical timing.
   I used an AT-clone with an Epson MX100 and discovered that at 8Mhz and
   10Mhz the printer would print slow. Only if I changed my AT to run at
   9Mhz the printer would print fast. This may seem very unlikely but it
   really did happen. And there is nothing SCO can do about it, because
   there are too many different computers and printers out there to get
   the timing right for all of them.
2) the polling driver, which the manual suggests to try next, is again
   depending on rather critical timing: The idea is to stay in a tight
   loop after sending a character to the printer, hoping that the printer
   will be ready to receive the next character very soon. This is repeated
   until there either are no more characters to print or until the printer
   does not respond quickly enough. The tight loop is short to reduce the
   influence on the total system performance. For some combinations of
   (fast) systems and (slow) printers the loop is still too short. SCO had
   to "guess" some reasonable number, and this may sometimes not work.
   If you are the unlucky guy who has this problem you should write your
   own (polling) printer driver. (mine is only 65 lines of C-code)

Paul De Bra.