dave@micropen (David F. Carlson) (08/10/88)
In a recent posting someone flamed John Plocher for being a Microport stooge. After the fire-play of/about Dimitri recently I would like to make an appeal to keep the ad-hominem attacks off this forum. It is unproductive and hurtful. For the benefit of those who have just started to deal with Microport: John Plocher *used* to be very active in the user community organizing bug lists, source archives and moderated the Microport maillist that was the predecessor of this forum. Last month he moved from Wisconsin to California to work harder at making Microport work. I feel he should have the chance to get his shit together before people flame him for decisions made up to two years before he joined there. I have great hope that by getting concerned and interested people into Microport, things will get better. John has shown himself in this and other forums as working for a better UNIX for Intel. Best of luck, John. There is no doubt that the Intel 80286 processor in the IBM PC/AT environment is a son-of-a-gun to support. No doubt. Serial boards using the 8250 chip have notorious interrupt noise immunity coupled with an outrageously small 64K kernel stack and a very expensive context switching mechanism makes for real headaches in the serial device drivers. I have looked at the serial device driver code and have wrung my hands at the double panics but it is a rock and a hard place situation--no one wins. At 9600 baud you will have an interrupt every millisecond. The service routine for pushing current context, completing the interrupt polling of the uart, and getting out cleanly consumes very close to 1 millisecond. So, you lose keyboard interrupts, or the floppy drive has multiple timeouts, or the interrupt stack grows while in another driver's critical region until the 64K stack overflows (aka, protection fault or double fault). What a mess. The 386 product is much better from a computer engineering viewpoint. I would recommend SV/AT only for hackers, masocists and cheapskates. We used it commercially for over a year with multiple users and full usenet support. I ported news 2.10.4 here in September of '86 months before those at site uport even knew it existed. I made 16 bit compress work without source manipulations--I was impressed that a toy computer could actually download 2Meg of news at night--and still stay up for a month at a time! My company under my recommendation junked SV/AT in favor of the *much* better 386 UNIX which has performed admirably in the 9 months we have supported it. If anyone needs serious UNIX, buy SV/386. So please people, ad-hominem attacks get us no where. Good accurate, verified bug reports with hardware, ROM rev., system config, reproducable circumstances gets a software product better. Best of luck in your new job John. Don't lose patience. Yours for a *better* Intel family UNIX, -- David F. Carlson, Micropen, Inc. micropen!dave@ee.rochester.edu "The faster I go, the behinder I get." --Lewis Carroll
ken@gatech.edu (Ken Seefried iii) (08/11/88)
Also note that it is pretty unusual for a company to have highly placed, very
active individuals on usenet acting in an official capacity.
Perhaps if people were a bit less vehement in attacking these individuals, more
companies would be willing to provide official info through usenet.
ken seefried iii ...!{akgua, allegra, amd, harpo, hplabs,
ken@gatech.edu inhp4, masscomp, rlgvax, sb1, uf-cgrl,
ccastks@gitvm1.bitnet unmvax, ut-ngp, ut-sally}!gatech!kenmp1@sdcc12.ucsd.EDU (Leroy Dorman) (08/12/88)
In article <531@micropen>, dave@micropen (David F. Carlson) writes: > > There is no doubt that the Intel 80286 processor in the IBM PC/AT environment > is a son-of-a-gun to support. The operative word is 'IBM PC/AT environment'. I've run 80286's on Multibus I (THE porting base for V/286) with few problems. > small 64K kernel stack and a very expensive context switching mechanism > makes for real headaches in the serial device drivers. I have looked at > the serial device driver code and have wrung my hands at the double panics > but it is a rock and a hard place situation--no one wins. At 9600 baud > you will have an interrupt every millisecond. The service routine for pushing I've eaten 19200Kb sustained rate (no handshaking!!) into an Intel smart serial board to Unix V/286 (using a general purpose terminal driver) and didn't lose a character. I only had 1Mb or ram and 1K kernel stack. Smart hardware wins again. > David F. Carlson, Micropen, Inc. > micropen!dave@ee.rochester.edu Regardless, John, I suspect, will do a kick-a** job just as he always has. No one has a right to fault him for things not under his control. As far as V/AT goes, running it on an AT is probably more trouble than it's worth. I've been running V/286 from the release tape on Multibus I with MUCH more stability that an AT. Don't know why . . . (actually, I do. Stupid H/W vs. Smart H/W) Eric Dorman Inch by painful inch University of California, San Diego toward a better tomorrow Scripps Institution of Oceanography -Me siolmd!eric@sdsioa.ucsd.edu mp1@sdcc12.ucsd.edu Attn: eric dorman@mplvax.nosc.mil Attn: eric Disclaimer: Just because I think this way doesn't mean anyone else does.