rich@devvax.JPL.NASA.GOV (Richard Pettit) (09/18/87)
[grrrr] Ok, I'm sick of all this massive confusion about who, what, where, and why UNIX 5.3 for the 80386 is, isn't and should be. Since I have the scoop, I think I'll pass it all on to the world. In the beginning, there was AT&T. (You knew it would start that way didn't you ?) AT&T developed UNIX 5.3 in their labs getting it to run on their line of 3B computers. At this point, they figured that it would be a great idea to put the system on a 80386 chip. Why not ? They were going to end up selling 80386 machines some day soon anyway. So, where is the best place to go to get your OS ported to a chip ? How about the people who made the chip. Well, that's what they did. Intel was contracted to port the 5.3 code to the 80386. Intel, not wanting this massive responsibility for themselves decided to subcontract the work. They chose the best company that they could to do the port because they (Intel) were responsible for delivering back to AT&T a running version of 5.3. This company is Interactive Systems, Inc. in Santa Monica, CA. Interactive (call them IS for short) went quickly to work porting the code, and in no time at all they had a working version. Not ready to release yet, but working. The way it works is like this: IS would complete a release of the code, be it beta or whatever, and they would release the code back to Intel. Intel would in turn release the code back to AT&T. AT&T would then release the code to those companies which had contracts with them to supply the source to 5.3 as it became available. So, IS is not just a code porting shop. They also sell UNIX 5.3, but they only sell it to companies. They know very well what kind of crap Microport goes through trying to support their systems. (Lets face it, a great deal of people out there are buying UNIX for their AT and they don't know crap about crap when it comes to crap. I wouldn't be a customer support person for anything. People calling asking questions like "I put the floppy in the drive without taking the cover off the disk and it won't read it. What should I do ?".) Anyway, since IS sells this system too, they take the code, make some changes to it and market it as two different products. The first is called 386/ix. This is just the stock 5.3 kernel and associated utilities, such as compilers, uucp, sys adm, and so forth. The second is called VP/ix. This is 386/ix with extensions to support the "virtual PC" mechanism. This is the MS-DOG under UNIX option. It utilizes the virtual-86 mode of the 80386. Now, obviously, since IS is the company doing the port, they are going to be the people who offer the product first right ? Wrong. IS will only sell the system if it is absolutely AT&T SVID certified. And they were, in fact, the first company to have such a product. And to give you a little clue, they only started shipping their 386/ix product about the end of August. What does that mean to those of you who got a similar product from Bell or Microport before then ? That's right. Beta release. Surprised ? "So, whose product is it that I got when I bought my Bell Tech. machine ?" you're asking yourself. Microport. They are one of the companies that buys the source from AT&T when it becomes available. They want to be the first company in the marketplace to offer a 5.3 for the 80386. So they release code that isn't SVID certified. And if you don't believe it, call them and ask. Call AT&T and ask them what companies currently have a SVID certified 5.3 available. Don't worry, it's a short list. Bell gives you a UNIX port when you buy their hardware because it's a great marketing ploy. What a lot of people don't know is that all you get is the base system without documentation. Thanks guys. None of this is to say that any of THESE companies are shifty. As far as I'm concerned, every successful company on the face of the earth is shifty. Caveat emptor. Microport doesn't sell any product called "SVID Certified SystemV/386". It's up to the buyer to find that out. I'm positive that eventually the Microport product will be SVID certified and that just like their SystemV/AT product, it will eventually become a mature, well running, well liked, popular product. The problem is that buyers simply don't beware before they buy something. You've got to watch your butt. How do you think the owners of 1984 Corvettes feel ? That was a new product too. As for the MS-DOG under UNIX option, Locus Computing of Santa Monica, CA does the DOSMerge stuff for Microport, and Interactive works in conjunction with Phoenix Technologies to do the VP/ix stuff. I have used both. I like VP/ix infinitely better. That is just my opinion. Take it or leave it. And as an added note, another company that buys the 5.3 source from AT&T is a little garage shop called Microsoft. Didn't forget about them did you ? This is comp.unix.xenix isn't it ? MS is taking the code, and working with Interactive Systems (they are everywhere, aren't they ?) are producing a 5.3 that will run COFF, Xenix, and DOS executables. Pretty slick, huh ? This product will be called Microsoft UNIX 5.3. Pretty original. Give them marketing people a raise. Does that clear up anything ? I hope so. If you have any more questions, drop me e-mail. I'm going to regret making that statement. Bye. Rich (rich@devvax.jpl.nasa.gov)
ron@vsedev.VSE.COM (Ron Flax) (09/20/87)
In article <411@devvax.JPL.NASA.GOV> rich@devvax.JPL.NASA.GOV (Richard Pettit) writes: >"So, whose product is it that I got when I bought my Bell Tech. machine ?" >you're asking yourself. Microport. They are one of the companies that I beg to differ with you here. Actually Bell's version is 386/ix from Interactive with Bell's device drivers for streaming tape, intelligent serial card, non-intelligent serial card, etc. The only thing unbundled in the Bell release is the Documenter's WorkBench 2.0, and of course vpix (the Interactive Systems DOS executive...). As far as documentation goes, Bell gives you the AT&T hardware specific documentation (differences specific to the 80386 port of UNIX 5.3), the System Administrator's Guide (from AT&T), and the System Administrators Reference Guide (equivalent to section 1 of the normal UNIX manual pages pertaining to administration. They also give you the generic AT&T UNIX 5.3 documentation as published by Prentice Hall for AT&T. You can of course buy as many copies of this as you need at around $100 per complete set. -- ron@vsedev.vse.com (Ron Flax) inet: vsedev!ron@cvl.umd.edu uucp: ..!uunet!cvl.umd.edu!vsedev!ron
kens@ism780c.UUCP (Ken Sarno) (09/24/87)
>I beg to differ with you here. Actually Bell's version is 386/ix from >Interactive with Bell's device drivers for streaming tape, intelligent >serial card, non-intelligent serial card, etc. The only thing >unbundled in the Bell release is the Documenter's WorkBench 2.0, and >of course vpix (the Interactive Systems DOS executive...). This is not quite true. 386/ix is INTERACTIVE's Unix V.3 packaged product, which has quite a few enhancements and additional features. The BTI release, which is the generic 386 V.3 system as written by ISC and certified by AT&T, is an *ancestor* of the 386/ix kernel. All 386 Unix V.3 systems that we know of are based on some release of the code developed by INTERACTIVE. While the Bell Technologies product is not 386/ix, as far as we know BTI ships only the certified release of Unix to its customers, not Beta releases. The enhancements over and above the generic 386 Unix which have been added to 386/ix by ISC include: - faster file system code - hot-key virtual consoles - the VP/ix DOS-under-Unix subsystem - integrated DOS file system support using the V.3 File System Switch facility. This facility, which is independant of VP/ix, allows you to mount a DOS diskette or hard disk partition so that both Unix and DOS processes can access DOS files as though they were normal Unix files (within reasonable limits, e.g. you can't create a 14-character name in a DOS partition). - standard support for the MICOM TCP/IP ethernet controller - standard support for ESDI and RLL disks in a new disk driver - a host of other standard device drivers - Documenter's Work Bench 2.0
sl@van-bc.UUCP (Stuart Lynne) (09/26/87)
In article <7375@ism780c.UUCP> kens@ism780c.UUCP (Ken Sarno) writes: >>I beg to differ with you here. Actually Bell's version is 386/ix from >>Interactive with Bell's device drivers for streaming tape, intelligent >>serial card, non-intelligent serial card, etc. The only thing >>unbundled in the Bell release is the Documenter's WorkBench 2.0, and >>of course vpix (the Interactive Systems DOS executive...). > > This is not quite true. 386/ix is INTERACTIVE's Unix V.3 > packaged product, which has quite a few enhancements and > additional features. The BTI release, which is the generic > 386 V.3 system as written by ISC and certified by AT&T, is > an *ancestor* of the 386/ix kernel. All 386 Unix V.3 systems > that we know of are based on some release of the code developed by > INTERACTIVE. > > While the Bell Technologies product is not 386/ix, > as far as we know BTI ships only the certified release of > Unix to its customers, not Beta releases. All *I* want to know is who is responsible for the async and parallel port drivers in Bell Tech's release. I wouldn't want to flame the wrong party. I've had the 386 box for a couple of weeks. Basically after much frustration I have given up on the parallel port driver. I must admit it *does* function. However printing one line every 5 to 10 seconds does not impress me *to* much. And as bad as the lp driver is, it's at least consistent. The serial drivers on the other hand.... Can you say *schizophrenic*... (let alone spell it). The serial driver supports the two standard AT serial ports. In my case one on the Intel motherboard, and one on an IBM Serial/Parallel Adaptor Card (true Blue, not clone). An attempt has been made to get around the normal Unix serial driver problems or how to handle modem control by having two minor devices for each port. One with and one without modem control. The problem is neither works! I have yet to find a method to tie even a simple ASCII terminal (Qume QVT102) to the 386 reliably. The closest I've come is to simply tie DTR from the terminal high (through a null modem to simulate DCD). This works fairly well. At least until I had to turn the terminal on and off to clear a glitch (QVT's sometimes lock up when you send random garbage at them, simplest way to fix is to just turn off and on); that hung the system! At other times I've seen the following types of problems: - characters sent to the second serial port appear at the rate of about one every 1.5 seconds - /etc/getty tty00 allows login, but if you didn't login the login after about a half a minute, the login message would appear about six times in quick succession after which a message would appear on the system console the /etc/getty was respawning to quickly - running cu -ltty00 displays, Connected, Lost Carrier, DisConnected in quick succession - running cu -ltty00, exiting with ~.<cr> hangs - running cu -ltty00, state change on DCD (low to high), Lost Carrier, disconnect The really annoying part is that it is simply not consistent from day to day (minute to minute sometimes). If the bloody thing would at least fail consistently, I could probably figure out a work around. Of course I'm probably the only person in the world having these problems right. I mean this is AT&T certified code, *right*. And I would *LOVE* to be proved wrong. If anyone can describe how to use a simple parallel printer with the lp driver, or hook a modem up to the asynchronous port for reliable dial in operation I'd just love to be made a fool of. Anyhow even though its a bit slower than we had hoped, the 386 box is quite nice. There's two of us on it right now doing light development work. This should extend to three in the next week or so, and we'll be getting down to brass tacks. I'll let everyone know how usable the box is for this type of use. (PS. I'm using the Bell Tech ICC card for extra users, not the standard serial ports.) -- {ihnp4!alberta!ubc-vision,uunet}!van-bc!Stuart.Lynne Vancouver,BC,604-937-7532
bill@ism780c.UUCP (10/02/87)
> All *I* want to know is who is responsible for the async and parallel port > drivers in Bell Tech's release. I wouldn't want to flame the wrong party. > > I've had the 386 box for a couple of weeks. Basically after much frustration > I have given up on the parallel port driver. I must admit it *does* > function. However printing one line every 5 to 10 seconds does not impress > me *to* much. > > And as bad as the lp driver is, it's at least consistent. The serial drivers > on the other hand.... Can you say *schizophrenic*... (let alone spell it). > Your parallel printer port problem is a configuration problem. Check your config file. The reason your printer runs so slowly is that you've configured it to run off the clock interrupt, rather than the printer port interrupt. So you are sending a character to the printer every clock tick. The ICC driver was developed by BTI. You may also have a configuration problem there as well. You might check with them to find out if you have the latest version of the driver. The base drivers (display, keyboard, parallel, async (COM1 & COM2), CMOS, hd, and fd) were certified by AT&T and were also extensively tested by Intel's beta sites (more than 60 beta customers). They work if configured properly. Note that the standard V.3/386 release does not support more than a single device per interrupt vector. If you have 2 devices on a single interrupt vector, one of them will be serviced when interrupts occur and the other will likely be serviced by clock ticks.