Path: utzoo!attcan!uunet!l5comp!scotty From: scotty@l5comp.UUCP (Scott Turner) Newsgroups: comp.sys.ibm.pc Subject: Re: AT&T 386 UNIX Vr3.1 Summary: AT&T unix vs Microport from one who owns both Message-ID: <423@l5comp.UUCP> Date: 2 Sep 88 03:38:10 GMT References: <145@carpet.WLK.COM> <441@uport.UUCP> <176@djs.UUCP> <7339@bigtex.uucp> Reply-To: scotty@l5comp.UUCP (Scott Turner) Organization: L5 Computing, Edmonds, WA Lines: 718 This debate of "Pure" AT&T unix vs. Microport SystemV/386 was just too much, I just had to pop up and add my several thousand dollars worth. :) Why several thousand dollars? That's how much I've sunk into this '386 unix swamp... First let's get my "political" leaning stated up front, any unix system based on System V with no Berk 4.3 goodies tossed in is a continual migrane. I'm not as bad as the GNU "Death before System V" camp but I'm not real big on "pure" System V. Thus, everthing except the Roadrunner's OS is non-choice in my political view. But with Sun not licensing their OS for lowly 386AT machines one is forced to work with available materials. My first choice was Microport's System V/386 2.2 (those "extra" numbers are REAL important when discussing these things folks...) with every optional software package they had (upto and including getting the NSA restricted crypt package.) I also decided to take them up on their hotline support contract and their update contract. Total bill? In excess of $1800. Sounds like ALOT, but it's what I expected a REAL unix to cost. Problem is that this REAL unix doesn't come with REAL WORLD features... The text preparation option is a prime example. It comes ready to talk to a phototypesetter or an obsolete Imagen laser printer (and they don't supply the driver for the typesetter!) No support for Epson's, Laserjet's, or any other type of printer that one would expect to find hooked to a 386AT. I hope they've fixed this, but for some reason they haven't sent this package to me for beta testing along with the rest of the unix... (even though I have ALL sorts of printer hardware to test it with) The Network Services Extension is a sad joke. In a REAL WORLD where everyone speaks TELNET/FTP/NFS, NSE gives you CU/UUCP/RFS. Sure, CU/UUCP/RFS do everything TELNET/FTP/NFS do, but not with anything speaking TELNET/FTP/NFS! Which seems to be most of the REAL WORLD these days, sigh. The cruelest joke though is that TELNET/FTP are available as third party add-ons. $1500 and Excellan will fill the gap, but only after you buy the otherwise worthless NSE!!! The irony of this still rankles me. For $1500 you think they'd GIVE you the damn NSE... But alas the cruel REAL WORLD had even more blows... System V/386 2.2 had no support for ESDI, RLL, or SCSI. No tape drives other than what I THINK is a Televideo cartridge tape unit. The battle was joined and to their credit Microport didn't try to duck the battle. An ESDI capable kernel was delivered up and the real fun began... The REAL WORLD struck again. Big drives (in this case 320Mb FORMATTED) come with LONG defect lists. System V/386 2.2 (and 3.0e for that matter, sigh) only supports mapping out 62 defects. Worse still was that the early ESDI support was in the kernel ONLY. The defect scanning software still thought MFM when trying to map out the bad blocks. The hard drive scanning tool did fopen( "tty") rather than using stderr or stdout so you had to write real fast when copying down defect locations (so you could enter them into the defect list manually after running them through an algorithm on your handy HP16C.) The 320Mb drive became a 230Mb drive to lop off enough defects to get below the limit. Then it turned out the cursed hard drive scanner was only reporting the first bad block on a track. On ESDI drives the defects can be so large they span two sectors it seems... As if unix wasn't enough fun, DOSMerge 386 1.0 had all sorts of problems as well. My favorite was the bug report that DOSMerge didn't work with Maxtor hard drives. Didn't say which model, or for that matter how it could tell the drive was a Maxtor, just that it didn't work with them. :( DOSMerge would cause spontaneous system resets, you'd be watching a DIR listing and all of a sudden the screen would clear and you'd see the AMI BIOS take over. (or if on a serial tube the screen would freeze and the other users would scream for your blood) Good ol' fsck would randomly pick files to kill, /bin/sh, or the getty, or the /etc/passwd... But Microport stuck it out. I've got a PILE of disks with various "extra" numbers tacked on after the DOSMerge and System V/386 designations. System V/386 3.0e and DOSMerge 1.1-U still don't like my Northgate CT-101 (they have that fixed, but DOSMerge 1.1-U is still using the old System V/386 2.2H console driver, sigh) and I get the occasional "Drive not ready" on A:. They threw in an Everex tape driver but it and my controller have yet to talk to each other without locking the whole box up. On the bright side though the current beta stuff is what I would have wished to have had back when I first bought into this mess. Things are for the most part functional. (Although the Northgate CT-101 problem is REAL annoying, it "bounces" the key you typed before the last key entered ie ls gets you lss or uucp will get you uucpc grrrrrr.) Now that I've aired my frustration with Microport what next? Well, I said I put THOUSANDS into this unix mess, the story ain't over yet. :) Bell Technologies. I don't know if I want to talk about these folks on the net or not. They seem real edgy about having customers talk about them. Screw 'em. :) Back in the really dark days with Microport I started wondering if I had picked the wrong unix. Bell Tech claimed ESDI, RLL, a COMPLETE networking setup and all for only $495. It sounded too good to be true. I held out on will power and fresh beta's from Microport. Then came the WGE. I asked Bell Tech to send me some info about their hardware fix for bug 21 in the '386. They sent this huge poster of the display from a WGE. It had me, customers, everyone drooling. Let's face it, 1660x1200 monochrome with coprocessor and X-windows for $1495 was (still is) red hot. It was red hot and SOLD. :) But alas Bell Tech claimed the drivers for the WGE would only work with their System V/386 3.0. Will power destroyed. They missed their initial ship date. They missed the next one, "We needed the boards for a show." I had told them to ship the unix without the WGE after they missed the first ship date, "Oh, I didn't think you'd want the unix without the WGE" even though I had told him I did... Oh well, he was under pressure, we're all human... The unix arrived, sans most of the manuals. When would the manuals arrive? Various "ummm, well, ahhhh" answers. Screw the manuals said I, I won't read 'em anyway. :) At this point I learned something VERY interesting. Bell Tech System V/386 3.0 and Microport System V/386 x.x look ALOT alike. Damn near identical in fact. Little thoughts start nibbling "Why won't the WGE drivers install on Microport?" "How many Microport problems are really AT&T problems and thus shared?"... Well the 62 defect problem is shared. (Damn!) Interestingly though, after claiming ESDI support, Bell Tech won't even boot from an Adaptec 2322 ESDI controller. "Does your controller have EPROMs or ROMs on it?" hmm "Yes, but it's 100% register compatible..." which evoked a line I will remember forever "Well, our driver sometimes gets confused by onboard EPROMs." She couldn't explain how EPROMs could confuse a driver that was using direct register I/O, but she remained convinced the fault was those pesky EPROMs on my 2322. She promised to call back. I decided to try some "gene" splicing experiments. I took the Microport ESDI- capable hd.o and stuffed it into the Bell Tech unix. The hybrid booted just fine. I breath a BIG sigh of relief (I wanted the WGE real bad, not to mention everyone else was clamoring to use it.) Next day she calls back, as promised (one of the two times), and informs me that I need to return the unix. The engineers had told her that they had had very little luck with my hardware: Mylex motherboard and Maxtor hard drive. I was real confused at this point, why it was that everyone keep fingering Maxtor hard drives I just couldn't fathom... I told her I needed the unix so I could use the WGE and got the offer to leave a message for my sales person to call me and give me an RMA to return the unix for a full refund. "But what about the WGE?!?" "I'm sure he'll refund that as well, sir." Sputter, argh, screams of utter contempt for lower life forms found eating ones dinner! I decided to bank on the fact I had it working. I also decided to buy some insurance and called the switch board at Bell Tech and got the name of their chief Unix guru. I have left several messages for this person, I have NEVER ever ever heard back from him/her/it. (It is now over 2 months later as I write this) The WGE arrives. The "ummm, well, ahhhh" manuals arrive with it. I grab the WGE readme manual, waste time reading it (one of those really thin installation guides that turned you off on reading the damn things in the first place!) Decide to screw the Bell Tech unix "I'll not be getting much but an RMA number from them for support" and proceed with installation under Microport's unix. The space.c files compiled, the kernel linked, mcs ran. Hooray! Then I booted it. Hooray! I fired up X and turned eager eyes to the dark (the WGE didn't have support for MDA, CGA, EGA emulation back then so you had to boot with another display installed) big beautiful 19" display. It turned grey, a nice CRISP grey, heart REALLY pounds with excitement and then the X server dies. Couldn't talk to the Intel co-processor... And yet it had, the damn screen turned on after all. Re-read the installation guide, the REGISTERS are MEMORY mapped into cached memory addresses... Kill the cache, and it runs! Like a kid at christmas I start running everything in sight and looking at all the pictures. I nearly die from the pile of people leaning over my shoulder... Running through the demos takes .5hr. awm refuses to work. Some of the listed demos aren't available... About then the thrill of victory turns a tad sour. Then gets REAL sour. I forgot, much like "pure" unix vs Berk 4.3, X10R4 doesn't hold a candle to X11R2... Oh well. About then the system crashes and fsck does another number on the hard drive during reboot. Then the system refuses to boot. splint errors appear, 2.2H release notes indicate this means no ram where the system is told to expect it. For example missing EGA ram, bingo! The card appears very very very DEAD. Bell Tech was willing to fix it, but they were much happier to give me a refund RMA rather than a repair RMA. They really didn't want to work with me on the caching issues either. They also had no answer for what they were going to do on other cached CPU's like the newer Compaq's... As for when the REAL X-Windows would show "We're hoping someone will develop X11 for us." Exit Bell Tech. But I didn't send anything back. :) Claims of X11R2 and MDA emulation dancing in the future I have held onto the WGE and the unix. I just ordered up the MDA emulation EPROM this week so maybe the WGE will fly again. (I now know the splint's to be from a bad kernel, got zonked by fsck...) At this point I got to sit back and think. Microport System V/386 2.2H had arrived, it had GreenHill's C-386, a Graphics driver that ate escape sequences to draw lines on the screen, Kernel debugger, adb, SysViz, Everex tape driver, keyboard key re-mapper, hot key remapper, the list went on and on... It made the Bell Tech unix look pretty weak, but Bell claimed that AT&T would be releasing 3.1 RSN and it would fix the 62 defect problem that Microport kept putting off (and is STILL putting off.) But I still couldn't get the system to run reliably for more than 4 days. (The WGE is on the shelf at this point to keep life simple) And I've installed the unix system about 18 times at this point. That gets old REAL quick (there are close to 20 disks to be loaded each and every time...) In this black hour PCDOS 4.0 shows up. Claims of big partitions, bigger than 32Mb(!), EMS support inside the DOS, and a EEMS emulator for '386 machines did me in. Down to Egghead, $130 gets me the last of 12 copies they got in that MORNING. PCDOS had grown. It now came on 5 disks. It had it's own install program. But it was still DOS. :) Formatting the drive didn't go too well. The old 2.x format problem of hitting a bad block past the 50% point seems to be back. At least for a 260Mb root partition. A 32mb root and a 230Mb extended partition worked though. The '386 EEMS simulator only works with PS/2's. (IBM doesn't make a 386AT after all...) And it uses code in the PS/2 BIOS as near as I can tell (I debug'd out the "Real IBM computer" check.) All the other creepy PCDOS 4.0 bugs showed up (format a disk, load it with software, and a 3.21 system thinks the disk is EMPTY!). The much vaunted DOS Shell wouldn't shell. Emitted crypt error messages rather than running. emacs *\Makefile wouldn't. I missed DOSMerge 386. Especially when the system would hang, with DOSMerge you could always login via a serial terminal and kill -9 the dos task if things got totally out of control. Then arrives Microport's System V/386 3.0e and DOSMerge 1.1-U. I was DOStired and DOSweary. Time to get beat up by unix again. :) But 'lo it installed no sweat. It had this cute new scanner that knew about ESDI drives and ran at a decent clip. Those duel bad sectors per track didn't fool it. It didn't cough when it found more than 62 defects, it didn't map the extras out but it didn't just abort like previous scanners. Life looked good. I'm happy to report that life still looks good. I'm still trying to convince fsdb to let me create a bad block file to map out the extra defects, the keyboard is still flakey, the A: drive is flakey. It's System V, but it beats PCDOS. Now, as for the DOSMerge can run "Hello world" and little else... Of this whole sad saga DOSMerge has been a real reason to keep on going. It's beyond terrific. It's the BEST PCDOS development environment I've run across. It beats the rest, like PCMOS or Concurrent DOS, because not only do you get DOS but you get unix commands. Why use ndmake when you can have the real thing? :) Rather than hacking programs to do what unix commands can already do, you get to do work. If the users can't hack ls, let them use DIR. Even from a unix shell! I've got two DOSMerge bugs at present: 1. The ROM BIOS machine ID is set to 0. Which is illegal and confuses some software, such as Codeview. But you can write a program to patch the byte and insert a call to the program into the autoexec.bat and this problem is fixed (and maybe someday Locus will fix it.) 2. The "virtual" 8259 seems to have a problem being used to poll for IRQ's by the too-smart-for-its-own-good Microsoft BUS Mouse driver. I always wondered why I didn't have to tell it which IRQ to use, well now I know. It finds the BUS chip, then goes into a loop of making the chip issue an IRQ and polling the 8259 to see if any came in. DOSMerge currently won't run Desqview 2.01. Gets some sort of math exception and then locks up tight. But Windows 2.1 does run, minus the mouse though. :) I've run NCSA's Telnet/FTP on a WD8003E under DOSMerge (using the /etc/dosdev to teach it about the WD8003E) and finally have REAL WORLD networking. :) I suspect PC-NFS would run as well so I could have NFS. Under DOSMerge I can run REAL WORLD nroff's that know about REAL WORLD printing devices. In short DOSMerge "fixes" alot of the problems with "pure" unix. It does a nice job of filling in for the missing Berk 4.3 enhancements. (To tip or procomm? Not a real difficult choice. :) Anyway, time to wrap up this rather long posting... Despite all the pain caused by Microport I'm sticking with them because they're sticking with me. This, more than anything else, is why I give Microport higher marks than Bell Tech. Bell is quicker to offer an RMA return than support in my experience. Of course what I could have here is a case of: here is beta-support from Micrport and none from Bell Tech and life would be even more confused if I had beta supprt from both (no pun intended ;). But one thing is for damn sure, I can get tech support from Microport via a 1-800 number which is active 24hrs a day 7 days a week. With Bell Tech you get a harried operator and no calls back except from yer sales person. He's always eager to call you back. :) But as for tech support you have to be scheduled for a call back and in my experience you often don't ever get called back. Or maybe it's "Ooops, phone is busy, take him off the list." :-) But support issues aside, you pay more but it looks like you will get more from Microport. Greenhills, Everex, SysViz, and various tools to make life easier (cmos ram config tool for example). They also give you some example files to work from. The Bell Tech $495 unix is pretty well stripped down for a unix. They don't even give you sample .profile's and .login's with the Bell Tech unix. No extra tools beyond what Microport provides. And no REAL WORLD ethernet support either, you get the same NSE CU/UUCP/RFS as with Microport. No extra ethernet drivers no extra support tools, the same exact bits. No Greenhills, no SysViz, no CMOS config tools... Oh yeah! One last response to a net item, shl doesn't come close to what Microport has with their virtual consoles... (or berk has with window) Scott Turner scotty@l5comp -or- uunet!l5comp!scotty Newsgroups: comp.sys.att,comp.unix.microport,comp.sys.ibm.pc Subject: Re: AT&T 386 UNIX Vr3.1 Summary: They all have weak points... References: <145@carpet.WLK.COM> <171@banzai.UUCP> <2753@jpl-devvax.JPL.NASA.GOV> <12031@steinmetz.ge.com> Reply-To: scotty@l5comp.UUCP (Scott Turner) Organization: L5 Computing, Edmonds, WA This debate of "Pure" AT&T unix vs. Microport SystemV/386 was just too much, I just had to pop up and add my several thousand dollars worth. :) Why several thousand dollars? That's how much I've sunk into this '386 unix swamp... First let's get my "political" leaning stated up front, any unix system based on System V with no Berk 4.3 goodies tossed in is a continual migrane. I'm not as bad as the GNU "Death before System V" camp but I'm not real big on "pure" System V. Thus, everthing except the Roadrunner's OS is non-choice in my political view. But with Sun not licensing their OS for lowly 386AT machines one is forced to work with available materials. My first choice was Microport's System V/386 2.2 (those "extra" numbers are REAL important when discussing these things folks...) with every optional software package they had (upto and including getting the NSA restricted crypt package.) I also decided to take them up on their hotline support contract and their update contract. Total bill? In excess of $1800. Sounds like ALOT, but it's what I expected a REAL unix to cost. Problem is that this REAL unix doesn't come with REAL WORLD features... The text preparation option is a prime example. It comes ready to talk to a phototypesetter or an obsolete Imagen laser printer (and they don't supply the driver for the typesetter!) No support for Epson's, Laserjet's, or any other type of printer that one would expect to find hooked to a 386AT. I hope they've fixed this, but for some reason they haven't sent this package to me for beta testing along with the rest of the unix... (even though I have ALL sorts of printer hardware to test it with) The Network Services Extension is a sad joke. In a REAL WORLD where everyone speaks TELNET/FTP/NFS, NSE gives you CU/UUCP/RFS. Sure, CU/UUCP/RFS do everything TELNET/FTP/NFS do, but not with anything speaking TELNET/FTP/NFS! Which seems to be most of the REAL WORLD these days, sigh. The cruelest joke though is that TELNET/FTP are available as third party add-ons. $1500 and Excellan will fill the gap, but only after you buy the otherwise worthless NSE!!! The irony of this still rankles me. For $1500 you think they'd GIVE you the damn NSE... But alas the cruel REAL WORLD had even more blows... System V/386 2.2 had no support for ESDI, RLL, or SCSI. No tape drives other than what I THINK is a Televideo cartridge tape unit. The battle was joined and to their credit Microport didn't try to duck the battle. An ESDI capable kernel was delivered up and the real fun began... The REAL WORLD struck again. Big drives (in this case 320Mb FORMATTED) come with LONG defect lists. System V/386 2.2 (and 3.0e for that matter, sigh) only supports mapping out 62 defects. Worse still was that the early ESDI support was in the kernel ONLY. The defect scanning software still thought MFM when trying to map out the bad blocks. The hard drive scanning tool did fopen( "tty") rather than using stderr or stdout so you had to write real fast when copying down defect locations (so you could enter them into the defect list manually after running them through an algorithm on your handy HP16C.) The 320Mb drive became a 230Mb drive to lop off enough defects to get below the limit. Then it turned out the cursed hard drive scanner was only reporting the first bad block on a track. On ESDI drives the defects can be so large they span two sectors it seems... As if unix wasn't enough fun, DOSMerge 386 1.0 had all sorts of problems as well. My favorite was the bug report that DOSMerge didn't work with Maxtor hard drives. Didn't say which model, or for that matter how it could tell the drive was a Maxtor, just that it didn't work with them. :( DOSMerge would cause spontaneous system resets, you'd be watching a DIR listing and all of a sudden the screen would clear and you'd see the AMI BIOS take over. (or if on a serial tube the screen would freeze and the other users would scream for your blood) Good ol' fsck would randomly pick files to kill, /bin/sh, or the getty, or the /etc/passwd... But Microport stuck it out. I've got a PILE of disks with various "extra" numbers tacked on after the DOSMerge and System V/386 designations. System V/386 3.0e and DOSMerge 1.1-U still don't like my Northgate CT-101 (they have that fixed, but DOSMerge 1.1-U is still using the old System V/386 2.2H console driver, sigh) and I get the occasional "Drive not ready" on A:. They threw in an Everex tape driver but it and my controller have yet to talk to each other without locking the whole box up. On the bright side though the current beta stuff is what I would have wished to have had back when I first bought into this mess. Things are for the most part functional. (Although the Northgate CT-101 problem is REAL annoying, it "bounces" the key you typed before the last key entered ie ls gets you lss or uucp will get you uucpc grrrrrr.) Now that I've aired my frustration with Microport what next? Well, I said I put THOUSANDS into this unix mess, the story ain't over yet. :) Bell Technologies. I don't know if I want to talk about these folks on the net or not. They seem real edgy about having customers talk about them. Screw 'em. :) Back in the really dark days with Microport I started wondering if I had picked the wrong unix. Bell Tech claimed ESDI, RLL, a COMPLETE networking setup and all for only $495. It sounded too good to be true. I held out on will power and fresh beta's from Microport. Then came the WGE. I asked Bell Tech to send me some info about their hardware fix for bug 21 in the '386. They sent this huge poster of the display from a WGE. It had me, customers, everyone drooling. Let's face it, 1660x1200 monochrome with coprocessor and X-windows for $1495 was (still is) red hot. It was red hot and SOLD. :) But alas Bell Tech claimed the drivers for the WGE would only work with their System V/386 3.0. Will power destroyed. They missed their initial ship date. They missed the next one, "We needed the boards for a show." I had told them to ship the unix without the WGE after they missed the first ship date, "Oh, I didn't think you'd want the unix without the WGE" even though I had told him I did... Oh well, he was under pressure, we're all human... The unix arrived, sans most of the manuals. When would the manuals arrive? Various "ummm, well, ahhhh" answers. Screw the manuals said I, I won't read 'em anyway. :) At this point I learned something VERY interesting. Bell Tech System V/386 3.0 and Microport System V/386 x.x look ALOT alike. Damn near identical in fact. Little thoughts start nibbling "Why won't the WGE drivers install on Microport?" "How many Microport problems are really AT&T problems and thus shared?"... Well the 62 defect problem is shared. (Damn!) Interestingly though, after claiming ESDI support, Bell Tech won't even boot from an Adaptec 2322 ESDI controller. "Does your controller have EPROMs or ROMs on it?" hmm "Yes, but it's 100% register compatible..." which evoked a line I will remember forever "Well, our driver sometimes gets confused by onboard EPROMs." She couldn't explain how EPROMs could confuse a driver that was using direct register I/O, but she remained convinced the fault was those pesky EPROMs on my 2322. She promised to call back. I decided to try some "gene" splicing experiments. I took the Microport ESDI- capable hd.o and stuffed it into the Bell Tech unix. The hybrid booted just fine. I breath a BIG sigh of relief (I wanted the WGE real bad, not to mention everyone else was clamoring to use it.) Next day she calls back, as promised (one of the two times), and informs me that I need to return the unix. The engineers had told her that they had had very little luck with my hardware: Mylex motherboard and Maxtor hard drive. I was real confused at this point, why it was that everyone keep fingering Maxtor hard drives I just couldn't fathom... I told her I needed the unix so I could use the WGE and got the offer to leave a message for my sales person to call me and give me an RMA to return the unix for a full refund. "But what about the WGE?!?" "I'm sure he'll refund that as well, sir." Sputter, argh, screams of utter contempt for lower life forms found eating ones dinner! I decided to bank on the fact I had it working. I also decided to buy some insurance and called the switch board at Bell Tech and got the name of their chief Unix guru. I have left several messages for this person, I have NEVER ever ever heard back from him/her/it. (It is now over 2 months later as I write this) The WGE arrives. The "ummm, well, ahhhh" manuals arrive with it. I grab the WGE readme manual, waste time reading it (one of those really thin installation guides that turned you off on reading the damn things in the first place!) Decide to screw the Bell Tech unix "I'll not be getting much but an RMA number from them for support" and proceed with installation under Microport's unix. The space.c files compiled, the kernel linked, mcs ran. Hooray! Then I booted it. Hooray! I fired up X and turned eager eyes to the dark (the WGE didn't have support for MDA, CGA, EGA emulation back then so you had to boot with another display installed) big beautiful 19" display. It turned grey, a nice CRISP grey, heart REALLY pounds with excitement and then the X server dies. Couldn't talk to the Intel co-processor... And yet it had, the damn screen turned on after all. Re-read the installation guide, the REGISTERS are MEMORY mapped into cached memory addresses... Kill the cache, and it runs! Like a kid at christmas I start running everything in sight and looking at all the pictures. I nearly die from the pile of people leaning over my shoulder... Running through the demos takes .5hr. awm refuses to work. Some of the listed demos aren't available... About then the thrill of victory turns a tad sour. Then gets REAL sour. I forgot, much like "pure" unix vs Berk 4.3, X10R4 doesn't hold a candle to X11R2... Oh well. About then the system crashes and fsck does another number on the hard drive during reboot. Then the system refuses to boot. splint errors appear, 2.2H release notes indicate this means no ram where the system is told to expect it. For example missing EGA ram, bingo! The card appears very very very DEAD. Bell Tech was willing to fix it, but they were much happier to give me a refund RMA rather than a repair RMA. They really didn't want to work with me on the caching issues either. They also had no answer for what they were going to do on other cached CPU's like the newer Compaq's... As for when the REAL X-Windows would show "We're hoping someone will develop X11 for us." Exit Bell Tech. But I didn't send anything back. :) Claims of X11R2 and MDA emulation dancing in the future I have held onto the WGE and the unix. I just ordered up the MDA emulation EPROM this week so maybe the WGE will fly again. (I now know the splint's to be from a bad kernel, got zonked by fsck...) At this point I got to sit back and think. Microport System V/386 2.2H had arrived, it had GreenHill's C-386, a Graphics driver that ate escape sequences to draw lines on the screen, Kernel debugger, adb, SysViz, Everex tape driver, keyboard key re-mapper, hot key remapper, the list went on and on... It made the Bell Tech unix look pretty weak, but Bell claimed that AT&T would be releasing 3.1 RSN and it would fix the 62 defect problem that Microport kept putting off (and is STILL putting off.) But I still couldn't get the system to run reliably for more than 4 days. (The WGE is on the shelf at this point to keep life simple) And I've installed the unix system about 18 times at this point. That gets old REAL quick (there are close to 20 disks to be loaded each and every time...) In this black hour PCDOS 4.0 shows up. Claims of big partitions, bigger than 32Mb(!), EMS support inside the DOS, and a EEMS emulator for '386 machines did me in. Down to Egghead, $130 gets me the last of 12 copies they got in that MORNING. PCDOS had grown. It now came on 5 disks. It had it's own install program. But it was still DOS. :) Formatting the drive didn't go too well. The old 2.x format problem of hitting a bad block past the 50% point seems to be back. At least for a 260Mb root partition. A 32mb root and a 230Mb extended partition worked though. The '386 EEMS simulator only works with PS/2's. (IBM doesn't make a 386AT after all...) And it uses code in the PS/2 BIOS as near as I can tell (I debug'd out the "Real IBM computer" check.) All the other creepy PCDOS 4.0 bugs showed up (format a disk, load it with software, and a 3.21 system thinks the disk is EMPTY!). The much vaunted DOS Shell wouldn't shell. Emitted crypt error messages rather than running. emacs *\Makefile wouldn't. I missed DOSMerge 386. Especially when the system would hang, with DOSMerge you could always login via a serial terminal and kill -9 the dos task if things got totally out of control. Then arrives Microport's System V/386 3.0e and DOSMerge 1.1-U. I was DOStired and DOSweary. Time to get beat up by unix again. :) But 'lo it installed no sweat. It had this cute new scanner that knew about ESDI drives and ran at a decent clip. Those duel bad sectors per track didn't fool it. It didn't cough when it found more than 62 defects, it didn't map the extras out but it didn't just abort like previous scanners. Life looked good. I'm happy to report that life still looks good. I'm still trying to convince fsdb to let me create a bad block file to map out the extra defects, the keyboard is still flakey, the A: drive is flakey. It's System V, but it beats PCDOS. Now, as for the DOSMerge can run "Hello world" and little else... Of this whole sad saga DOSMerge has been a real reason to keep on going. It's beyond terrific. It's the BEST PCDOS development environment I've run across. It beats the rest, like PCMOS or Concurrent DOS, because not only do you get DOS but you get unix commands. Why use ndmake when you can have the real thing? :) Rather than hacking programs to do what unix commands can already do, you get to do work. If the users can't hack ls, let them use DIR. Even from a unix shell! I've got two DOSMerge bugs at present: 1. The ROM BIOS machine ID is set to 0. Which is illegal and confuses some software, such as Codeview. But you can write a program to patch the byte and insert a call to the program into the autoexec.bat and this problem is fixed (and maybe someday Locus will fix it.) 2. The "virtual" 8259 seems to have a problem being used to poll for IRQ's by the too-smart-for-its-own-good Microsoft BUS Mouse driver. I always wondered why I didn't have to tell it which IRQ to use, well now I know. It finds the BUS chip, then goes into a loop of making the chip issue an IRQ and polling the 8259 to see if any came in. DOSMerge currently won't run Desqview 2.01. Gets some sort of math exception and then locks up tight. But Windows 2.1 does run, minus the mouse though. :) I've run NCSA's Telnet/FTP on a WD8003E under DOSMerge (using the /etc/dosdev to teach it about the WD8003E) and finally have REAL WORLD networking. :) I suspect PC-NFS would run as well so I could have NFS. Under DOSMerge I can run REAL WORLD nroff's that know about REAL WORLD printing devices. In short DOSMerge "fixes" alot of the problems with "pure" unix. It does a nice job of filling in for the missing Berk 4.3 enhancements. (To tip or procomm? Not a real difficult choice. :) Anyway, time to wrap up this rather long posting... Despite all the pain caused by Microport I'm sticking with them because they're sticking with me. This, more than anything else, is why I give Microport higher marks than Bell Tech. Bell is quicker to offer an RMA return than support in my experience. Of course what I could have here is a case of: here is beta-support from Micrport and none from Bell Tech and life would be even more confused if I had beta supprt from both (no pun intended ;). But one thing is for damn sure, I can get tech support from Microport via a 1-800 number which is active 24hrs a day 7 days a week. With Bell Tech you get a harried operator and no calls back except from yer sales person. He's always eager to call you back. :) But as for tech support you have to be scheduled for a call back and in my experience you often don't ever get called back. Or maybe it's "Ooops, phone is busy, take him off the list." :-) But support issues aside, you pay more but it looks like you will get more from Microport. Greenhills, Everex, SysViz, and various tools to make life easier (cmos ram config tool for example). They also give you some example files to work from. The Bell Tech $495 unix is pretty well stripped down for a unix. They don't even give you sample .profile's and .login's with the Bell Tech unix. No extra tools beyond what Microport provides. And no REAL WORLD ethernet support either, you get the same NSE CU/UUCP/RFS as with Microport. No extra ethernet drivers no extra support tools, the same exact bits. No Greenhills, no SysViz, no CMOS config tools... Oh yeah! One last response to a net item, shl doesn't come close to what Microport has with their virtual consoles... (or berk has with window) Scott Turner scotty@l5comp -or- uunet!l5comp!scotty