Info-IBMPC@USC-ISIB.ARPA (Info-IBMPC Digest) (03/11/86)
Info-IBMPC Digest Monday, March 10, 1986 Volume 5 : Issue 32 This Week's Editor: Phyllis O'Neil Today's Topics: LPTX Alternate Character Sets Clearing the Screen (2 messages) Re: Microsoft C Compiler Wish List FORTRAN to C query Redistributing INFO-IBMPC RE: AT&T PC6300 Color Display NEC V30 in a Compaq Deskpro SAVVY Turbo Pascal's 64K Limitation Today's Queries: Help with FTP between PC and VM Help needed with device driver INTERRUPT ROUTINES--QUERY Discrete simulation languages query .OBJ file format query Sort and German DOS 3.x query PC's Limited AT Query ---------------------------------------------------------------------- Date: Mon, 10 Mar 86 16:15:09 est From: Mike Ciaraldi <ciaraldi@rochester.arpa> To: info-ibmpc@usc-isib.arpa Subject: LPTX Problems (more details) I downloaded LPTX a few weeks ago from the ARPA archives, and found it works partially. My first problem was not knowing that it had to be processed by EXE2BIN. At first I thought the "No Stack Segment" message from LINK was an actual error. Once I got that figured out, it would actually run. Results: Under DOS 2.1 it seems to work fine for me. This is on a standard IBM PC with Color Graphics Adapter and dual floppies. I tried several different things, such as copying a file to PRN, using printer echo (ctrl-P), etc., and they all seemed OK. I was NOT using any other memory-resident programs. Under DOS 3.1, it worked all right for printer echo and other commands that didn't require disk use. When I tried to do a DIR > PRN or a DIR with printer echo, the first 512 bytes would go into the printer file, then the whole system would lock up. Repeating, other ways of producing printer output (e.g. from a BASIC program) that did not use the disk caused no trouble. So, it would appear that there is a conflict when DOS is trying to handle two disk requests simultaneously, one producing the data and one receiving it. Looking through the code, I see that there is a place where the author saves away the system stack, explaining in a comment that this gets around the restriction against doing#zDOS calls from inside an interrupt handle. The comment makes reference to the DOS 2.1 reference manual. So, my suspicion is that something changed between DOS versions, and LPTX does not take that into account. I hope someone else with more experience than I can figure out how to fix this, as LPTX seems very handy. Mike Ciaraldi ciaraldi@rochester seismo!rochester!ciaraldi ------------------------------ To: jmsk@mc.lcs.mit.edu Cc: info-ibmpc@isib Subject: Alternate Character Sets Date: Sun, 09 Mar 86 23:07:23 -0500 From: jcmorris@mitre.ARPA In your posting of 8 March you asked about the STSC handling of the special character sets used in its APL product. The STSC APL*PLUS/PC package can work with all three standard IBM CRT interfaces (mono, CGA, and EGA). The special PROM is required with the monochrome adapter; it's recommended but not mandatory with the CGA, and cannot be used with the EGA. (There are also several different PROM's for use with other manufacturers' boards.) Assuming that one of the IBM boards is being used, the no-PROM approach uses the techniques shown in the standard IBM documentation. You don't say in your note which CRT adapter you're using. If it's the IBM mono or CGA, both boards use the same PROM. I've not actually done it, but it doesn't look to be too difficult to figure out the addressing scheme and design your own character set if you've got access to a PROM burner. The EGA uses proprietary chips and doesn't have a socketed PROM to be replaced. The IBM manuals show how to specify your own character set for the CGA and EGA. There is no way to avoid a special PROM if you're using the monochrome adapter. Joe Morris (jcmorris@mitre) ------------------------------ Date: Mon 10 Mar 86 12:42:35-CST From: Pete Galvin <CC.GALVIN@R20.UTEXAS.EDU> Subject: Re: Clearing the screen To: JAZBO%BROWNVM.BITNET@WISCVM.WISC.EDU cc: INFO-IBMPC@USC-ISIB.ARPA I believe the best method is one described in the IBM Tech. Ref. It is simply a call to the bios video routines scroll function. Tell it to scroll 25 lines, and the screen is cleared. Sorry I don't have the numbers in front of me... --Pete ------------------------------ Date: 10 Mar 1986 11:01:27 PST Subject: Re: Clearing the screen From: Richard Gillmann <GILLMANN@USC-ISIB.ARPA> To: Pete Galvin <CC.GALVIN@R20.UTEXAS.EDU> cc: JAZBO%BROWNVM.BITNET@WISCVM.WISC.EDU, INFO-IBMPC@USC-ISIB.ARPA I agree that the BIOS call is the best bet. Most MS-DOS machines have a mode that will emulate the IBM BIOS to that extent. The only way to clear the screen that does not depend on ANSI.SYS nor on the IBM BIOS is to write 25 linefeeds to the screen. And even this won't be compatible with displays with more than 25 lines, like the EGA in some modes. ------------------------------ Date: Mon, 10 Mar 86 09:55:22 CST From: nather@ASTRO.UTEXAS.EDU (Ed Nather) To: info-ibmpc@USC-ISIB.ARPA Subject: Re: Microsoft C Compiler Wish List >From: Ya'akov_Miles%UBC.MAILNET@MIT-MULTICS.ARPA >It would be nice if Microsofts "C" compiler ver 3.0 was enhanced by > > (1) Improving the optimizer such that string constants are folded. > Currently the following "C" statements > > PRINTF("\nThis string is stored twice by Microsoft `C`"); > PRINTF("\nThis string is stored twice by Microsoft `C`"); > > causes the string constant to be replicated, wasting space. I don't think they can do that ... the C langauge allows you to write to a string constant and change its value, so each one must be unique. While there is no way to address the one in your example, it is a special case which the compiler would have to identify. And if it were identical to one that you *could* address ... > (2) Increasing the module size that the optimizer will handle before > > warning: function `xxxxxx` too large for post-optimizer [0] That one should be easy. ------------------------------ Date: Mon, 10 Mar 86 13:51:47 CST From: fenchel@rsch.wisc.edu (Bob Fenchel) To: info-ibmpc@isib Subject: FORTRAN to C query Anyone have, or know of a program that can do a reasonable job translating from FORTRAN to C? ------------------------------ To: info-ibmpc@usc-isib.arpa Subject: Redistributing INFO-IBMPC Date: Mon, 03 Mar 86 16:07:23 -0500 From: Craig Partridge <craig@loki.bbn.com> A few weeks ago a note went out to this list which, among other things, expressed some annoyance at those of us who run CSNET for not offering redistribution facilities for mail digests like INFO-IBMPC. We at CSNET view this as a somewhat unfortunate situation -- we like to see ourselves as helping the research community, yet in this particular situation our hands are tied. I've included parts of the original message below with an explanation of why we are unable to help out. I hope this will give people a better sense of why the problem exists. > Date: 19 Jan 1986 13:34:50 PST > Subject: New Year's Resolutions > From: Billy <BRACKENRIDGE@USC-ISIB.ARPA> > To: info-ibmpc@USC-ISIB.ARPA > The editors of INFO-IBMPC met last week. Several changes have taken > place. > > [much text deleted] > We are unresolved as to what to do about sending back digests and > library files to unconnected networks. We never have responded to > requests from usenet hosts unless they were closely connected to th > arpanet, and we are getting increasingly fed up with sending things > via mail to csnet and bitnet. The only reason these nets don't have file > access to our archives is because the political administrators of > these networks are unwilling to implement gateways that can handle > file transfer. I fail to see why we should have to send files as mail > because the users of these connected networks are unwilling or unable > to put pressure on the administrators of their nets to allow them > file access to ARPA hosts. In the mean time some requests will get > answered and others won't depending on the mood of who is editing at > the moment. This sounds harsh but only reflects the reality of the > last year or so as requests have grown beyond our capability to handle > them. We'd love to offer file access to ARPA hosts. In fact we at CSNET had plans to implement (indeed had started work on) a mail interface to the FTP protocol. But when we mentioned this idea to various parties involved in running the DDN we were told that we could not offer mail oriented FTP. Apparently FTP is involved in a large fraction of the security incidents on the DDN, and if we made FTP available to a larger community (i.e. those people who can reach the Internet via mail), the security problems and concerns could become much more serious. > > If someone had a mail based server that could automaticly respond to > these requests, this could solve the problem for the short term. > > [more deletions ] Such programs exist. Both CSNET and BITNET have information servers that are capable of automatically handling requests via mail. Other versions exists on the UUCP network. MOSIS (at ISI) also offers this capability. We (at CSNET) are, however, in no position to start archiving all of the various mailing lists that our members are interested in (we just don't have enough processing and disk resources to handle this much data). As I've said, we'd like to help out, and we are not unaware of this problem. If we find a solution at our end you'll be the first to know. insert Craig Partridge CSNET Technical Staff craig@sh.cs.net (for the daring) or craig@csnet-sh.arpa I stand by my statements that the reason csnet subscribers can't get FTP access to our files and archives is political not technical. If "the various parties involved in running the DDN" really believe FTP access to the ARPANET is a security breach there is something seriously wrong with the DDN. I am sure there is some appropriate mailing list to discuss whatever politics may be involved in this decision, but this isn't the one. I spoke with the MOSIS people and their software is too specific to MOSIS to work for INFO-IBMPC. In the mean time there is a program called netlib that runs on unix that allows people to fetch files from libraries via net mail. I'd like to get that up running. I don't speak unix and don't want to do it personally but would be glad to give assistance to any volunteers. -wab] ------------------------------ Date: 10 MAR 86 14:32-N From: INNO%HWALHW5.BITNET@WISCVM.WISC.EDU To: INFO-IBMPC@USC-ISIB.ARPA Subject: RE: AT&T PC6300 Color Display I have an Olivetti M24 with a color monitor. It has almost the same hardware as the AT&T 6300. I recognized the problem of the color monitor. I had the problem too, when I was running the product SUPERKEY. When I tried to exit the same problem occurred: a snowy display and the left margin in the center of the screen. I didn't have the problem when I booted from an original IBM PC-DOS V2.00 floppy. For the rest I never have problems with my color monitor, even if I run some graphics color demos for the IBM-PC. Inno Frencken Computing Centre Agricultural University Hollandseweg 1 6706 KN Wageningen The Netherlands phone: 08370-83875 EARN-id: INNO EARN-node: HWALHW5 ------------------------------ Date: Mon 10 Mar 1986 12:39:11 EST From: <SAGE@LL.ARPA> Subject: NEC V30 in a Compaq Deskpro To: info-ibmpc@usc-isib I am running my Deskpro Model 4 with an NEC V30 chip in it. The Norton utility performance index doubled, indicating that some instructions now run twice as fast as before. I ran one benchmark using a compute-bound PMATE text-editing macro and found a speed improvement of about 10-15%. My hard disk seems to run noticeably faster. The one problem I have had is with the tape backup. Things seem to go too fast for it! I solved the problem by switching the computer to its low-speed clock when making the tape backups. ------------------------------ Date: Mon 10 Mar 86 10:24:47-PST From: CRESWELL@SRI-AI.ARPA Subject: SAVVY To: INFO-IBMPC@USC-ISIB.ARPA I purchased Savvy about 6 months ago and have used it for database applications for medium sized files. It is easy to use...seems to do all I want it to...though I'm not using it for extremely exotic purposes. Purchased it through the mail for a good deal....installed in my PC hard disk and had it up and running in about an hour. The little instruction books are well written. The manual, like most manuals, is "ok" if you really need to delve into the guts of things altho' I've found that the menus/user books have provided most everything I need. ------------------------------ Date: 10 Mar 86 10:58:53 PST (Monday) From: Eldridge.ES@Xerox.COM Subject: Turbo Pascal's 64K Limitation To: Marvin Zauderer <ZAUDERER@SU-SUSHI.ARPA> cc: info-ibmpc@USC-ISIB.ARPA, Eldridge.ES@Xerox.COM It is true that Turbo Pascal is limited to 64K of code space and 64K of static data space, but this is not really a problem. To get around the code space limitation, Turbo Pascal provides an overlay facility. Chaining can also be used where appropriate. The 64K limitation on the data space applies only to static data structures. If you use dynamic data structures, then you have access to all available memory in the machine. Dynamic data structures are allocated off of the heap. There is a limitation of 65521 bytes for a single dynamic variable. I have heard that Nostradamus software offers a utility that extends the code and data space range of Turbo Pascal, but I have no experience with it. George (Eldridge.es@Xerox.com) ------------------------------ Date: Mon, 10 Mar 86 10:28:07 cet Subject: Help with FTP between PC and VM From: UZR500%DBNRHRZ1.BITNET@WISCVM.WISC.EDU I am developing a FTP-program running on the IBM-PC (every version). Our PC has a coax kable and neither IRMA nor AST-PCOX card for emulation and FTP from/to our IBM/370 host under VM/CMS. At the moment we would need some information exchange with other people who have the same problems or who have already written such a program,no matter in which programming language it has been written. We have several programs which are solving our problem. But every is too slow or it does not run exactly. We have problems in compiling BASICA-pro= grams with the IBM-BASIC-COMPILER and with the QUICK-BASIC-COMPILER. At the end of our list,writing this program in TURBO-PASCAL brings some other problems,too.So,if there is someone who has the same pro= blem or better who has already solved it,I would like to correspond with him in order to solve it. Thank you very much. Sincerely yours Thorsten Glattki Glattki, Thorsten 0228/73-2747 UZR500 at DBNRHRZ1 RHRZ Universitaet Bonn Zimmer 003 Wegelerstrasse 4 D-5300 Bonn 1 Ferderal Republic of Germany (West Germany) ------------------------------ Date: Mon, 10 Mar 86 03:11 EST From: Paul Schauble <Schauble@MIT-MULTICS.ARPA> Subject: Help needed with device driver To: Info-IBMPC@USC-ISIB.ARPA I am attempting to write a device driver for a slightly strange disk and have run into a problem. The disk is different from a standard PC disk in that it uses a 2K byte sector size. I know that DOS has some procedure for dealing with other sector sizes: I have previously seen device drivers that use other than 512 byte sectors. However, when I try to load my driver, it gets the message "Sector size too large in driver.sys". Does anyone know if anything special is needed to set the sector size that DOS uses? Or, just exactly what is the problem here? Thanks, Paul ------------------------------ Date: 10 MAR 86 10:48-N From: DOMMELEN%HWALHW5.BITNET@WISCVM.WISC.EDU To: INFO-IBMPC@USC-ISIB.ARPA Subject: INTERRUPT ROUTINES--QUERY Has anyone in this audience a routine which does the following: 1. Generates a timer interrupt and calls a routine e.g. each second. 2. Can be locked into the system, so keeps interrupting when a FORTRAN program runs. 3. Maybe can do something like the ON KEY .. GOSUB functions of GWBASIC. I'm asking this for a custumor who is working with an Olivetti M21 and wants to write his program in Microsoft FORTRAN. Probably these routines should be made in MASM, which is available also. Any suggestions are welcome. Wim van Dommelen, Computer Centre Agricultural University, Wageningen Holland. DOMMELEN@HWALHW5.BITNET ------------------------------ Date: Monday, 10 March 1986 13:57:55 EST From: Joe.Newcomer@a.sei.cmu.edu To: info-ibmpc@usc-isib.arpa Subject: Discrete simulation languages query (I'm asking this for a friend) Are there any discrete system simulation languages (the last one I used was Simula) available for the PC. He needs to simulate a system with lots of discrete servers (neg. exp. arrival and service times) with setup and "latency" considerations. Graphics output nice but not required. A self-contained system is desired; native language is C (he has no Fortran, so Fortran-based preprocessor simulator things will not work). joe ------------------------------ Date: Monday, 10 March 1986 14:00:48 EST From: Joe.Newcomer@a.sei.cmu.edu To: info-ibmpc@usc-isib.arpa Subject: .OBJ file format query (for another friend) I know someone who wishes to port his favorite compiler to the PC. Ideally, he'd like to write .obj files directly instead of invoking the assembler. The .OBJ file format is rather obscure and the Intel document hasn't helped us much. Does anyone have a public domain or small $ .OBJ-file-writer? Desire is to support the equivalent of Lattice's S,P,D and L models. I don't have time to get involved in this deeply so am looking for ways to help. joe ------------------------------ Date: Mon, 10 Mar 86 15:59:36 PST From: kegel%Juliet.Caltech.Edu@Hamlet.Caltech.Edu Subject: Sort and German DOS 3.x query To: info-ibmpc@usc-isib.arpa, m70b%cbebda3t.bitnet@wiscvm.wisc.edu The stated problem is, why does SORT seem to ignore the umlaut on letters? Sort is presumably using the generalized mechanism provided by DOS for resolving ordering of ascii characters. The DOS function call "Get/set country information" allows you to get or set the address of the subroutine used to order characters. Although the documentation in the DOS 3.0 Technical Reference manual does not specify what exactly the subroutine does, tracing thru it with DEBUG reveals that it is something like the following: translate proc far push bx ; save caller's register mov bx, offset value_table ; get address of our order-value table xlat cs:[bx] ; replace char in AL with its order value pop bx ret translate endp value_table db 0, 1, 2, 3, 4, 5 ; and so on The value_table used by DOS simply assigns the value 'A' to all letter A's regardless of case or accent. To replace the default character-ordering subroutine, you would need to write a small program which would set the address of the your subroutine, then terminate and remain resident. Warning: the above example subroutine is typed from memory, and may be incorrect; you will have to trace thru the default subroutine to make sure. Have fun, and good luck! - Daniel Kegel 1-60 Caltech, Pasadena, CA 91126 ------------------------------ Date: Mon, 10 Mar 86 01:34:41 EST From: "Keith F. Lynch" <KFL@MC.LCS.MIT.EDU> Subject: PC's Limited AT Query To: info-ibmpc@USC-ISIB.ARPA Does anyone have any info on the PC's Limited AT compatible? Is it any good? How can it be so much cheaper than the competition? ------------------------------ End of Info-IBMPC Digest ************************ -------