info-vax@ucbvax.ARPA (07/25/85)
From: D. M. Rosenblum <DR01@CMU-CC-TE> We have a problem here at Carnegie-Mellon University's School of Urban and Public Affairs, that perhaps somebody can help us out with. We have a VAX/780 running VMS V3.7, and a number of DEC PRO/350's hooked up to it by hard-wired 4800-baud lines (which are about to have their speeds increased to 9600 baud). We frequently use PRO/Communications V2.0 (under P/OS V2.0) to use the PRO/350's as terminals to talk to the VAX, emulating VT102 terminals. The editor that is most used on our VAX is the VMS version of Gosling emacs. (This is hard to change because most of our users are non-hacking-oriented end-users who also use C-MU's TOPS-20's, which have MIT (TECO-based) EMACS as their main editor, and such users would not be too willing to learn another completely different editor such as SOS or EDT.) Here's our problem: at 4800 baud when PRO/Communications is emulating a VT102 (and rarely when it's emulating a VT52), and at 9600 baud when it is emulating either a VT52 or a VT102, emacs often puts stuff out to the PRO/350 faster than the PRO can handle it. By the way, the same problem sometimes occurs on actual VT100 terminals and Gigis (VK100s) at either speed. Does anyone know of a way in VMS Gosling emacs to allow the system (rather than emacs) to handle XON and XOFF characters for their usual purpose (much as, in MIT TECO, one can use the FS TTY PAGE MODE flag to do the same thing)? Alternatively, does anyone know of a way to alter emacs's internal definition of the VT102 to allow enough timing delays, especially after insert and delete line, to make the problems we've observed substantially unlikely? (The definition of VT1xx terminals is apparently hard-wired in Gosling emacs (it doesn't go through the termcap file).) Are there any other ideas as to how to deal with this problem? Any assistance would be appreciated. Thanks in anticipation. By the way, most of our PRO/350's are also hooked up to our VAX by ETHERNET lines, and we have the standard PRO/DECnet software to support these hook-ups. Although this allows us to bypass PRO/Communications entirely, it seems to fail miserably when one tries to run emacs because PRO/DECnet's terminal emulation lacks much of what emacs expects to be there. If anyone has any suggestions for making emacs work through this connection instead of through the communications lines, that would also be appreciated. Daniel M. Rosenblum, Ph.D. candidate, School of Urban and Public Affairs Carnegie-Mellon University BITNET address: DR01@CMCCTE (? maybe DR01@CMU-CC-TE) MAILNET address: DR01%CMU-CC-TE@CARNEGIE ARPANET address: not sure -- DR01@CMU-CC-TE might work; if not try DR01%CMCCTE.BITNET@WISCVM or DR01%CMCCTE@CU20B or DR01%CMCCTE@COLUMBIA-20 -------
info-vax@ucbvax.ARPA (07/26/85)
From: Keith F. Lynch <KFL@MIT-MC.ARPA> We have this kind of problem a lot at SAIC. DEC is very XON/XOFF oriented. In fact the VT200 series terminals (which is what I think the PRO/350 is really emulating) will not really send a ^S at all. When you type ^S, the screen will freeze and the terminal will not transmit a ^S unless the VAX keeps sending data. With the version of Gosmacs that we have, there is a "set xon mode on" command. In this mode ^S will pause Emacs, and ^Q will continue it. Of course this means that ^S and ^Q cannot be used as commands or as parts of commands. My only other suggestions are to reduce the speed to 2400 baud, try SET TERMIANL /CRFILL=9 /LFFILL=9 (or some lesser values) to pad things with nulls (this won't work if you have an older version of Gosmacs that only works in /PASSALL mode), and turn off smooth scroll if it's on. ...Keith
info-vax@ucbvax.ARPA (07/26/85)
From: H Eidnes <H_Eidnes%vax.runit.unit.uninett@NTA-VAX> I know of a Emacs subset work-alike from Sweden - AMIS, and how they've solved the problem with Xon-Xoff. The editor has now been given a command "M-X XonXoff Mode", which maps the character C-S to C-H ("hit" - you use C-B for "backwards" anyway...) and C-Q to C-\ (in the style of Unix quote...). The Xon and Xoff characters are then freed to do flow control with. There is one problem, however - how do you insert a C-Q in your text (or C-S for that matter)? I have read somewhere that the author of Goslings Emacs is *very* much against using ^S and ^Q for flow control - what he said (wrote) was basically that he did *not* intend to support inferior terminals uncapable of receiving data at sufficient speed without flow control...