mcba@newt.phys.unsw.OZ.AU (Michael C. B. Ashley) (11/01/90)
I'm driving a LaserJet III (with HP PostScript cartridge + 2Mbytes) via a serial line at 19200 baud from a DECstation 5000 running ULTRIX 4.0, and I'm having a lot of trouble. The problem is that when the LaserJet's input buffer becomes full (which appears to be after 60000 bytes) it sends a control-S to the DECstation, but the DECstation can transmit up to another 40 or so characters before stopping. Unfortunately, the LaserJet only appears to be able to buffer another 30 characters before it gives an IO CONFIG ERR. My local HP response centre have not been particularly helpful and assure me that the problem is with the DECstation. However, I have used the DECstation with Apple LaserWriters and NEC LC890s without difficulty. Perhaps HP's "Robust XON/XOFF" is to blame ... Can someone confirm the 30 character buffering limitation? Is there a way around this (e.g, with some obscure PostScript system call?)? Another unfortunate feature of the LaserJet is that when you remove the print tray, the printer immediately sends a control-S to the computer. It would have been better to delay sending a control-S until the printer was actually about to print a page. That's how Apple do it. Another problem is the inability to swap between PCL and PostScript without having to switch the printer off and remove the cartidge. I thought I was buying a machine that would run with both PCL and PostScript with some sensible method of software switching. Enough problems. The JaserJet is really a very neat printer, and the resolution enhancement technology is a vast improvement over other 300dpi printers. Thanks for any advice on the XON/XOFF problem. Michael Ashley mcba@usage.csd.unsw.oz.au Astrophysics Dept. / Uni. of NSW / Australia.
rjn@hpfcso.HP.COM (Bob Niland) (11/02/90)
re: > The problem is that when the LaserJet's input buffer becomes full > (which appears to be after 60000 bytes) it sends a control-S to the > DECstation, but the DECstation can transmit up to another 40 or so > characters before stopping. Unfortunately, the LaserJet only appears > to be able to buffer another 30 characters before it gives an IO > CONFIG ERR. If the DEC is indeed sending 40 chars, I can believe this is happening. > My local HP response centre have not been particularly helpful and > assure me that the problem is with the DECstation. That would be my conclusion as well. There is no formal spec for Xon/Xoff handshaking. After HP got its fingers burned on the 2225D serial ThinkJet, which has only an 8 byte FIFO, we reportedly determined that industry practice is 32-byte Xoff hysteresis. Either end may send up to 32 more bytes after receipt of an Xoff. I would expect the LaserJet to have a 32-byte FIFO, but probably not more. I'm really surprised that DEC ships 40 bytes. For all practical purposes DEC invented Xon/Xoff, so they should know better. Some HP workstations allow you to configure the FIFO on each serial port. Does DEC have some similar capability in the device file minor number, an ioctl call or as an 'stty' option? > However, I have used the DECstation with Apple LaserWriters and NEC LC890s > without difficulty. Perhaps HP's "Robust XON/XOFF" is to blame ... The robust Xon/Xoff should actually be called "robust Xon". It's function is to prevent loss of communication due to lost Xon's (Xon sent while the host was already in a FIFO overflow situation). The way it works is: * If the printer is not been busy, and * 5 seconds have elapsed since sending the last Xon, * Then send an Xon. This can cause problems for hosts that fail to read and discard unexpected Xons (Xons arriving with no prior Xoff). Their input FIFOs can fill up with Xons, resulting in unpredictable consequences. I assume you have tried running with "robust" disabled? > Another problem is the inability to swap between PCL and PostScript > without having to switch the printer off and remove the cartidge. I > thought I was buying a machine that would run with both PCL and PostScript > with some sensible method of software switching. All the PostScript users inside HP agree with you on this, and I fully expect future LaserJets to be more flexible in this regard. Regards, Hewlett-Packard Bob Niland Internet: rjn@hpfcrjn.FC.HP.COM 3404 East Harmony Road UUCP: [hplabs|hpfcse]!hpfcrjn!rjn Ft Collins CO 80525-9599