Info-IBMPC@USC-ISIB.ARPA (Info-IBMPC Digest) (11/25/85)
Info-IBMPC Digest Sunday, 24 November 1985 Volume 4 : Issue 133 Today's Editor: Eliot Moore <Elmo@USC-ISIB.Arpa> Today's Topics: Underflow Interrupts MS-DOS Interrupt Parameters EGA via C Lattice Wordstar 3.3 vs. Wordstar 2000 ROMing programs EasyFlow vs. NEC V20 DAK Printers (Silver-Reed, Gorilla, Olivetti) Changes in XT Motherboard Wendin Operating System Toolbox PCVMS Remote vs. Carbon Copy Lugaru's Epsilon Today's Queries: Epsilon 3.01 and Tab Settings LAN's C Floating Point Arithmetic Files & Interrupt handlers Rexx High-speed I/O Boards FAX IBM 3278/79 Card Motherboard Replacements ---------------------------------------------------------------------- Date: Tue, 19 Nov 85 08:26:12 cst From: cody@anl-mcs.ARPA (Jim Cody) To: cody@ANL-MCS.ARPA, steven@BRL-TBD.ARPA Subject: Underflow Interrupts Let me suggest an alternative. ALL arithmetic on the 8087 is done with 80 bits of significance, with the final result rounded to working precision. There is no time penalty, therefore, in doing things in double precision. The advantage is that the exponent range is much wider, thus pushing the underflow problem to where it may not matter with you. The disadvantage is increased storage for variables. ------------------------------ Date: 19 Nov 1985 09:15-EST From: Israel.Pinkas@ISL1.RI.CMU.EDU Subject: MS-DOS Interrupt Parameters To: Info-IBMPC.Digest.<Info-IBMPC@USC-ISIB.ARPA>,amc543@uiucuxa According to "IBM-PC Assembly Language", by Leo Scanlon, MS-DOS Interrupt 5 is the Print Screen routine. DOS function 5 is Print Character. Thus the code that Aaron wants looks like: MOV DL, <character to print> MOV AH, 5 INT 21H No status is returned on this call. As far as I can tell, there is no way to direct the output to a printer other than PRN:. If you wish to print at a different printer port, you must use Interrupt 17. This would involve the following: MOV AL, <char to print> MOV DX, <printer to use (0-2)> INT 17H This returns with a code in the AH register which looks like: BUSY | ACK | PAPER-OUT | SEL | I/O-ERROR | UNUSED | UNUSED | TIME-OUT These are asserted high. (i.e paper is out if bit 5 is high.) Please note that my reference describes DOS 1.x. There may be better ways to do this in DOS 2/3.X, but these methods do guarantee that all versions of DOS will work with your program. (Or is that your program will work with all versions of DOS?) -Israel igp@isl1.ri.cmu.edu ------------------------------ Date: Tue, 19 Nov 85 13:13:39 cst From: cody@anl-mcs.ARPA (Jim Cody) To: cody@ANL-MCS.ARPA, steven@BRL-TBD.ARPA Subject: Underflow Interrupts Again One last bit of information. Professional Fortran runs your demo program without incident. It must be the flag settings in MS Fortran. ------------------------------ Date: 19 Nov 85 16:28:00 PDT From: SHEN, HWA N <shen@ge-crd> Subject: EGA via C To: lichtenberg.pa <lichtenberg.pa@xerox.arpa> Reply-To: SHEN, HWA N <shen@ge-crd> Halo Graphics will support Enhanced Graphics Adaptor for Lattice C and other languages (Frotran, Basic, Microsoft C, etc.) Halo is available from Lifeboat Associates or directly from Media Cybernetics. Shen@ge-crd.arpa for Marc Laymon ------ ------------------------------ Date: Tue, 19 Nov 85 12:19:47 PST From: Greg Coleman <bilbo.gregc@LOCUS.UCLA.EDU> To: INFO-IBMPC@USC-ISIB.ARPA CC: dana, jjr Subject: Lattice Lattice used a call to avoid writing self modifying code. The only way to do an int is with imediate mode addressing. The software interrupt number can not be in a register or anything. They could have used a 512 byte table to do it. Gregc ------------------------------ Date: Tue, 19 Nov 85 14:22:40 EST From: Thomas_Ackenhusen%UB-MTS%UMich-MTS.Mailnet@MIT-MULTICS.ARPA To: INFO-IBMPC@USC-ISIB.ARPA Subject: Wordstar 3.3 vs. Wordstar 2000 This is in reply to the request for a comparison between Wordstar 3.31 and Wordstar 2000. Wordstar 2000 is user-friendly, easy to use, and unbearably slow on anything other than an AT with hard disk. Unless you can stand waiting, don't get 2000. Thomas W. Ackenhusen P.S. Any answers on my query about the Underground Wordstar book of user patches for Wordstar? I really would like any info. about this book. Thanks. ------------------------------ Date: 19 Nov 85 22:11:50 EST From: Dee Snider <AD0R@TE.CC.CMU.EDU> To: info-ibmpc%usc-isib@rCMPSYAk Greatest-supernatural-horror-author-of-all-time: H. P. Lovecraft Phone: x4844 Room: Smudge B-104 Subject: ROMing programs a fiend and me have been discussing the possibility of putting kermit onto a prom and putting into the public pc's here (Carneie Mellon University) so that the pc's will boot into it instead of rom basic. I think it would be incredibly complex, because you don't have dos and the exe loader around. He thinks it would be simple. Any ideas? Al Datri Carnegie Mellon University ad0r@cmu-cc-te arpa, for you lucky doodes ad0r%cmu-cc-te mailnet, for the few ad0r%cmu-cc-te@cu20b bitnet Beak is. -------- ------------------------------ Date: Tue, 19 Nov 85 21:29:01 PST From: walton%Deimos@CIT-Hamlet.ARPA Subject: EasyFlow vs. NEC V20 A while back, Bob Stine wrote in that he had a program by the name of EasyFlow Plus which did not work on a V20-equipped PC, but did run on the same machine when the 8088 was plugged in. He is correct, though HavenTree (the company responsible for EasyFlow) gave him the wrong reason. EasyFlow contains a subroutine which in its entirety, is MOV AX,BX MUL DX MOV BX,AX RET When EasyFlow calls this routine, it apparently considers it an error if the result of the multiply is longer than 16 bits. Intel's documentation for the 8088, and NEC's for the V20, both confirm that the carry and overflow flags are set when the high-order result of a MUL is non-zero, and cleared when the result is zero, and that the other flags are undefined after a MUL. It turns out that the 8088 also sets the zero flag if the high-order result is zero, and clears it if it isn't. HavenTree chose to use this undocumented feature of the 8088 to check for overflow by looking at the zero flag, rather than the carry or overflow flags. The V20 doesn't change the zero flag after a MUL, hence EasyFlow doesn't work on a V20. I would emphasize, though, that in my opinion the fault is entirely that of the authors of the software, since they count on an undocumented and therefore unreliable feature of the 8088. Does the 80286/386 do this? If it doesn't, EasyFlow won't work on an AT or the NGAT (Next Generation AT). In pawing through EasyFlow, I found some real howlers, by the way. (1) The first thing EasyFlow does is copy itself downward in memory by 8D7 (hex) bytes for reasons I can't fathom. If it needed the extra memory, why didn't it just do a JMP around the space it needed? (EasyFlow is a .COM file.) (2) EasyFlow contains no DOS function calls accessed through INT 21. All DOS function calls are done by putting the function code in CL (yes, CL) and executing the instruction CALL 0005, which links to the FAR CALL in the program segment prefix (PSP) to the DOS function dispatcher. (3) Similarly, the code contains no INT 20 instruction--rather, it contains a CALL 0000, which goes to the INT 20 in the PSP. (4) The second thing EasyFlow does is calculate its own checksum. Fine, but the loop involved ends with the following four instructions: DEC CX MOV AL,CH OR AL,CL JNZ start_of_loop How's that for a wordy replacement for a LOOP start_of_loop instruction? I strongly urge Info-IBMPC readers to avoid EasyFlow. Between lousy coding and a strongly coupled-in use of an undocumented 8088 feature, it seems likely to break on many machines. Steve Walton Caltech Solar Astronomy walton%deimos@cit-hamlet.arpa swalton@citdeimo.bitnet ------------------------------ Date: Wed, 20 Nov 85 00:36 EST From: Hess@MIT-MULTICS.ARPA Subject: DAK Printers (Silver-Reed, Gorilla, Olivetti) To: INFO-IBMPC@USC-ISIB.ARPA This probably isn't the right forum, but I must give a counterpoint to Joe Newcomer's observations: I have used all three printers mentioned in the previous messages. The Silver-Reed printers, esp. the EXP-550 are all just fine, although we (Mark of the Unicorn) never had one for a long time. The one we had was Diablo compatible. The Gorilla Banana is *OUTRAGEOUSLY* NOISY! Pretty boring printer otherwise. The Olivetti sparkjet does screech, but it is a low-decibel high-pitch noise, sort of like a poorly fitted belt swishing on some piece of equipment nearby. It's quiet enough that a roommate doesn't complain when I print downstairs at 2:00 AM. Basically, I like it. The brown vs. black is inexplicable; I have seen it both ways and suspected some interaction between the solid ink ampule and the paper for brown. Mine at home is black all the time. I ordered a whole bunch of ink ampules when I ordered the printer, and they were delivered on time. DAK still advertises them as a separate item, about a year later, so they are still probably available. I only use the Olivetti for printing drafts and receiving Telexes, so I don't care too much about how the output looks. Double-wide works OK. But there's no backspace on the printer, and no bare carriage return without linefeed. The worst feature of the Olivetti is that it seems to create ozone when it prints, so sitting within a yard of the printer if it prints more than 2 or 3 pages smells funny, and may be hazardous somehow. ------------------------------ From: brown@nicmad.uucp Subject: Changes in XT Motherboard Date: 20 Nov 85 05:32:22 GMT To: info-ibmpc@usc-isib.ARPA We have just recently received new XTs from our local Computerland. To make a long story short, just today we discovered that IBM changed the design of the XT mother board so that it will no longer work with older versions of their hard disk controller board or will it work with the short slot OMTI hard disk controller board. The OMTI controller cards work fine in the older version mother board, but not in the new layout. We get our hard disks and controllers from Kamerman Labs. I called them today and I am the first to report a problem. I gave them the information that I had, ie, the changes to the boards and the new board part number. There are two ways you can tell that the board has been changed. One, there are six colored wires added to the component side of the old version board to make it new and two, the new layout is not green, but copper colored. At least my new version was copper colored. The new version now has a 74LS32 in the spare chip spot and gets labeld U90. So, if you buy third party hard disks, make sure it works on the new layout. I will put up more news as I find out about it. I will have some kind of comment from Kamerman Labs in a few days. --- ihnp4------\ harvard-\ \ Mr. Video seismo!uwvax!nicmad!brown topaz-/ / decvax------/ ------------------------------ Date: 20 Nov 1985 11:22:47 PST Subject: Wendin Operating System Toolbox From: Billy <BRACKENRIDGE@USC-ISIB.ARPA> To: info-ibmpc@USC-ISIB.ARPA I have taken a closer look at the Wendin Operating System Toolbox, and must correct some statements I made earlier. I incorrectly stated that it could not be used with Turbo Pascal. If Turbo Pascal is one of your "favorite languages" they support it, but you will still need the latest Microsoft C compiler if you wish to re compile the kernal. I was also a bit fuzzy about the program license agreement. You must buy one program for each developer working with their source code. You can move the source code to several machines or use it on a network if only one person is using it at a time, but you can sell freely whatever you develop using their code. If you don't like MS-DOS and had thought about writing your own operating system for the PC, but still wanted to run MS-DOS programs this is the product for you. As they say in their literature "we did the hard part". You are really buying an operating system kernal. This includes I/O, process, memory, date & time, and all the stuff real operating systems have and DOS lacks. This should be just the ticket if there is still anyone out there writing integrated environments for the PC. The intent of the package is that you will write your own shell and replace command.com, and presumably under that shell you should be able to write your own applications or run existing DOS programs. I had intended to use it to allow me to handle several speech I/O subsystem on a PC while talking to a network. Unix is out of the question for this and I would have has to write my own extentions to DOS to handle the multiple processes. Unfortunately they haven't replaced the DOS or BIOS I/O subsystems. I really can't get the concurrency as DOS is still busy waiting on the disk drives. They promise fixes to this in future releases as well as network and virtual memory support. ------- ------------------------------ From: jrodrig@EDN-VAX.ARPA (Jose Rodriguez) Date: 20 Nov 1985 1751-EST (Wednesday) To: info-ibmpc@usc-isib.ARPA Cc: well!wendin@lll-crg.ARPA Subject: PCVMS Considering I am an old fan of DEC' VMS, I recently ordered PCVMS. After a reasonable wait of 2-3 weeks (they told me they were swamped with orders) I got an impresive document and three diskettes. After reading through the documentation, I was impressed on how much of VMS is emulated, particularly it looked like most system calls and the objects they deal with have been reproduced. They have QIO, QIOW, SPAWN, TRNLOG, wait on and/or of event flags, mailboxes, logical names, etc. One thing though, they do not reproduce all the extesive software packages VMS comes with - no choice of CLI, the signal and command line parsing packages are not there, etc. (sorry I can't remember VMS names but I haven't used it in almost 3 years). So one Saturday morning I tried to use PCVMS (in a Tandy 1000 with 384k and two floppy drives - I should get a hard disk soon) and was somewhat disappointed. First when typing "show <something>" the system looks all right, even "show system" where you get the description of two processes running (oh by the way, no symbols). So I tried running some programs in the background with the spawn command and everytime it will return with some error message about some problems with the .exe file. Then I tried submitting some batch files (the @ notation works) and they do work but (forgive my Unix) only in the foreground!! In other words I could not create some background process on the system - which the documentation asures it's possible. Some exe files have trouble running under PCVMS - Perfect Writer and Edlin didn't work correctly. In the other hand Kermit worked ok. I also tried conecting a login processes on COM1:, which is part of the startup file (which I don't think gets executed), with no success. To summarize, I have had some trouble with PCVMS, and it seems it has been released ahead of time. One of its most interesting features, multi-tasking, does not work. I have version 1.3 (that is what the startup program says even if the floppys say in their labels version 1.4) so maybe I got a pre-release version. Who knows. I should get back in touch with them to at least get version 1.4. If someone can help me I would much appreciate it. If there is any further progress I will keep you posted. Jose M. Rodriguez SDC McLean RC jrodrig@edn-vax ------------------------------ Date: Wed, 20 Nov 85 19:50:23 EST From: "Richard P. Wilkes" <wilkes@hopkins-eecs-bravo.ARPA> To: Keith Dale <kdale@VAIHINGEN-EMH.ARPA> cc: info-ibmpc@usc-isib.ARPA Subject: Remote vs. Carbon Copy I think that you'll find that Carbon Copy from Meridian Technology works far better than Remote. We tested Remote and found several incompatibilities, including one with ANSI.SYS. Carbon Copy so far has run with everything we use with the exception of being able to transmit the hi-res EGA graphics (coming, I'm told). -r ------------------------------ Date: Sat, 16 Nov 85 21:01:48 PST From: Richard K. Jennings <jennings@AEROSPACE.ARPA> To: info-ibmpc@usc-isib Subject: Lugaru's Epsilon I just picked up a copy of this emacs clone, based in part upon the reviews it has gotten over the past year. After playing with it for about an hour (and reading the manual) two, for me, very unsettling observations: 1) if you are not familliar with emacs and C this editor is probably not for you, the documentation is not easy to use and although a tutorial of some kind is mentioned in the manual it must be some kind of an extra cost option. 2) as nearly as I can tell there is no way to 'extend' Epsilon to talk to COMM ports. Why is this a problem? Because they fool around with the keyboard interrupt, and things like mice load the keyboard buffer -- with characters not read until the DOS command processor gets control after Epsilon is done. It does permit you to run compilers, XLISP, and other very well behaved programs while you edit, but MS Windows will get you that capability (and with more than one subprocess). For someone dropping of a VAX, who programs 1/2 Mbyte of C code using emacs might very well like this editor. People familiar with Framework, the new Final Word, or the other SOA PC text editors may be disappointed. Although I don't know emacs I have to learn anyway, and I am quite familiar with C -- so if I can't break the thing next week I will probably keep it. I just can't believe that a company would sell a non-mouseable text editor for almost $200! Rich. ------------------------------ Date: Wed, 20 Nov 85 16:12:18 pst From: sonia@aids-unix (Sonia Orin Lyris) To: info-ibmpc@usc-isib Subject: Epsilon 3.01 and Tab Settings With the 2.0x test versions of Epsilon, you could put those commands you always wanted executed after startup in EINIT. With version 3.01, you can save the entire state of Epsilon at any time, but there doesn't appear to be a way to set tabs automatically. Has anyone sucessfully tackled this in the new version? Thanks in advance. Sonia Lyris [arpa: sonia@aids-unix] ------------------------------ Date: Sun, 17 Nov 85 18:02:56 EST From: Dave Swindell <dswindel@BBN-LABS-B.ARPA> To: info-ibmpc@usc-isib.ARPA cc: lsher@BBN-LABS-B.ARPA Subject: LAN's A non-net friend asked me to post this for her. She works in a small office associated with a non-profit health agency and which is using a basic 2 floppy disk-based IBM PC for database management (DB2) and word processing (MultiMate). They are thinking of purchasing a second, hard disk-based machine to relieve the demand on their current machine, but would like to be able to share resources between the two machines (printers, disks, files, and programs). Does anyone know of any good articles or references on PC networking equipment? What are the costs associated with networking two machines? How difficult/expensive is it to add in a third (fourth, fifth,...) machine into an existing network? How much system resources do network programs require (the current PC only has 256kb of RAM installed, and I've heard that some of the current network systems (3Com, etc...) require this much memory alone!)? Is DOS 3.1 required to do networking? If not, what does 3.1 buy you over 2.xx relative to networking? Can a user sitting on the floppy-based PC run MultiMate (or some other program) from the hard disk on the other machine? Would the office be just as well off by shuffling floppy disks between their two machines for the time being? As those of you who are familiar with networking can probably tell, I don't know too much about this area, so any knowledgeable responses would be appreciated. Thanks, Dave Swindell BBN Laboratories Incorporated ARPA: dswindell@bbn ------------------------------ Date: Tue, 19 Nov 85 02:20 EST From: Paul Schauble <Schauble@MIT-MULTICS.ARPA> Subject: C Floating Point Arithmetic To: Info-C@BRL-TGR.ARPA, Info-IBMPC@USC-ISIB.ARPA Does anyone know of a C compiler for the IBM PC that does NOT force all floating point arithmetic to be done in double precision? 8087 support is necessary. Thanks, Paul ------------------------------ Date: Tue, 19 Nov 85 05:16:50 PST From: dgb%Deimos@CIT-Hamlet.ARPA Subject: Files & Interrupt handlers To: INFO-IBMPC@USC-ISIB.ARPA Has anyone had any experience maniulating files from an interrupt handler? Since I need to use DOS to do the manipulation (I do not want to use the BIOS for file work), what precautions must be taken to insure not interrupting DOS at some critical time? So far as I can tell, the file functions are the only DOS functions I need. I can leave the rest alone. For instance; are the "traditional file management functions" any more robust than the extended functions group, or visa versa? The obvious solution is to route any interrupt/function that could concievably collide with my routine through a "guard function", but I would prefer to take over as few interrupts as possible. Thanks for any help in advance. --Daniel Briggs Caltech Solar Astronomy (818) 356-3803 dgb%deimos@cit-hamlet.arpa ------------------------------ Date: Tue, 19 Nov 85 10:18:02 CST From: C3669S%UMVMA.BITNET@WISCVM.ARPA (Kamil Marcinka) To: info-ibmpc@usc-isib.ARPA Subject: Rexx Does anybody have any experience with Rexx for PC? (Rexx is an interpreter for VM rel. 3.) /km ------------------------------ Date: Tue, 19 Nov 85 13:50:13 EST From: "Lawrence D. Sher" <sher@BBNJ.ARPA> To: info-ibmpc@usc-isib.ARPA Subject: High-speed I/O Boards ************************** WANTED **************************** Synchronous I/O board for PC/AT that supports bit-oriented protocol, e.g., SDLC or HDLC, at 1 to 2 megabits/sec. ------------------------------ Date: Tue 19 Nov 85 15:53:22-EST From: Marc B. Jones <MBJ@MIT-XX.ARPA> Subject: FAX To: info-ibmpc@USC-ISIB.ARPA cc: wax.es@XEROX.ARPA Perhaps I was not clear enough in my request about FAX machines for PCs. The machine I am looking for is used to covert hardcopy text, i.e. a typewritten page, into a ascii file. The ad I once saw had a picture of a box about 15" x 15" x 6". The paper is placed on a flat, non moving surface. You (or the machine, I am not sure which) ran a small scanner mounted on rails from left to right across each line. In order to improve it's accuracy, I think you had to "teach" it about the particular type face of the original. The name Omnireader and a cost of $500 sticks in my mind, but it could be incorrect. Any help in locating this particular machine or type of machine would be appreciated. Marc Jones ------- ------------------------------ Date: Tue, 19 Nov 1985 13:02 EST From: Jim Ennis <JIM%UCF1VM.BITNET@WISCVM.ARPA> Subject: IBM 3278/79 Card To: info-ibmpc <info-ibmpc@usc-isib.arpa> I am having problems using the IBM 3278/3279 card with an AST 6-pack card. If the clock on the AST card is enabled then the IBM 3278 card does not work correctly. Has anybody else had this kind of problem with the mix of IBM 3278/AST cards? If we disable the clock on the AST card the IBM 3278 card will work okay but this is not an acceptable solution. We have the problem on both IBM PCs and PC/XTs. Thanks, Jim Ennis snail mail: Jim Ennis University of Central Florida Computer Services PO BOX 25000 Orlando, Fl 32816 fast mail: JIM%UCF1VM.BITNET@WISCVM.ARPA fastest: (305) 275-2931 ------------------------------ Date: 19 Nov 85 17:41:00 EDT From: "LANDBERG, TED" <landberg@nbs-vms.ARPA> Subject: Motherboard Replacements To: "info-ibmpc" <info-ibmpc@usc-isib.arpa> Reply-To: "LANDBERG, TED" <landberg@nbs-vms.ARPA> I am looking for a replacement for my IBM PC-1 motherboard. Would like to reuse cards, 135w power supply, keyboard from old unit. Prefer XT version with 8088 or 8086 chip. Need source for ROM chip also. Will buy new cabinet with 8 slot backplane. Any sources for high quality parts. Ted Landberg NBS-VMS.ARPA ------ ------------------------------ End of Info-IBMPC Digest ************************ -------