steve@wlbr.UUCP (Steve Childress) (09/10/85)
...
....
In response to the request for general info on OS9 ... a MC68xxx O/S ...
- See Usenet group net.micro.6809 ... there's a buncha OS9 fans there,
especially Peter Dibble.
A brief discussion of OS-9:
o Authored by Microware Systems Corp, Des Moines, Iowa
o It's about 5 years old, starting out as a spin-off from a contract
between Microware and Motorola. Mot. was looking for a super-BASIC
and an O/S to exploit the superb architecture of the MC6809.
o OS-9 exists in three flavors:
- Level I for 8-bit sytems using 64K on CPUs having a 6809
- Level II, same but has bank-switching for up to 1MB on 6809s
- Level I for 680x0 machines with large memories (>=256K?)
o Key features:
- Costs about 1/8 of Unix
- Small, assembly-lang Kernel
- Reentrant & position-independent code everwhere which makes
multiuser and multitasking practical on < $5K computers
- Concurrent, priortized processes similar to Unix
- Tree-structured file system directories with long file names
- Pipes & some of Unix's signals like quit, abort, send_signal
- Interrupt driven scheduler and I/O drivers
- Loadable device drivers and dynamic extensions to the O/S
- Numerous Unix-like utilities; not as many as Unix though
- Reasonalby good interrupt latency and process switch times
- Microware's BASIC09 is noteworthy -- it's an incremental compiler
which is interactive, compiles to psuedocode, is FAST, and
sort of a merger of PASCAL's procedures and BASIC's easy debugging.
o Machines:
- Many (older) 6809s, eg., Southwest Technical, Gimix, Smoke
Signal, Hazlewood, Tandy Color Computer, Apple II w/add-in board.
- A (very) few retailed 68000 machines, eg., Smoke Signal,
Gimix, and perhaps two or so more.
- Supposedly more widely used as a real-time exec for controllers
I've used it for several years on 6809 machines and enjoyed it. I did`
a port of the Kernel to the upgraded Apple II -- it was fairly easy
because Microware's designers used some of the best design practices
and beautiful regularity I've ever seen. I have a 12MB hard disk
which has run under OS-9 (8-bit version) for 4 years without losing
a file. I can't say that for many others except DEC's RSX-11.
Because the code is reentrant (including BASIC09 and other applications)
and position-independent, you REALLY CAN run 2-3 users on a VERY inexpensive
machine .. so long as the users don't run many unique programs. EG., everyone
BBS'ing, editing or running BASIC09. I've even run three users on an Apple II
in BASIC09 --- it works so long as they use the floppy lightly.
Tandy sells it for their ($300) Color Computer ... it's a real
paradox in that (diminishing) market. Look for its manuals at your
nearby Tandy computer center.
The trouble (with OS9 for the 68000) is that it's not available on
a popular computer so there's never been any good non-system softare
so there's never been any users --- etc, etc, catch-22.
I considered going to OS9/68K until I spent a weekend with an AT&T
UNIX-PC. For < $5,000, that machine, with its bit-mapped graphics & Sys 5.2
just beats the heck out of any other high-end machine for the serious
hobbiest. Compare that machine to Smoke Signal's SS50 bus ( == no peripherals)
Regulus machine, or to Gimix's high-priced 68020 computer, or even to
(yech!) a PC/AT w/Xenix.
So in SUMMARY I'll editorialize:
OS9-6809 is the best small-computer O/S I've seen but
it has never gained popularity to attract software houses.
OS9-68000 is better than the 8-bit version (I've heard), but
is encumbered even moreso than its 8-bit predecessor.
Microware advertises in "Mini-Micro Systems", "Unix World", and ?
Microware's ads speak of several new manufacturers incorporation
of OS9/68K and announcements Real Soon Now. (Kudos to Jerry).
Personally, I don't suggest holding your breath!
Send out a CQ on net.micro.6809 for more info. I wrote several related
articles published in "Micro" magazine (now defunct) in '83 (or was it
'84?). Also wrote a big article in Kilobaud Microcomputing on a disk
device driver for OS9.
Hope this helps those curious about OS-9.
Regards,
Steve Childress
Eaton IMS R&D Group MS 43
31717 La Tienda Drive
Westlake Village, CA 91360
(818) 889-2211 X2148
{trwrb, scgvaxd, ihnp4, voder, vortex} !wlbr!steve
or ...wlbr!wlbreng1!stevesean@ukma.UUCP (Sean Casey) (09/16/85)
Flame on!
OS9 has a real problem, and it's not in the software, but with the management.
Microware will not release source code to OS9, not even to developers! I find
this apalling, and it is probably one of the main reasons that OS9 is not more
popular than it is. They have really stuck it to themselves there, because
AT&T is starting to push SysV for the 68000, for which you can get source code.
I realize that OS9 and SysV are different products. Their target markets are
considerably different at the ends, but have a large overlap in the middle.
If Microware wants a good share of that middle, they are going to have to give
the developers a means to practically support the product. Otherwise, well,
I told you so!
Flame off.
--
- Sean Casey UUCP: sean@ukma.UUCP or
- Department of Mathematics {cbosgd,anlams,hasmed}!ukma!sean
- University of Kentucky ARPA: ukma!sean@ANL-MCS.ARPAdavid@ukma.UUCP (David Herron, NPR Lover) (09/16/85)
In article <2193@ukma.UUCP> sean@ukma.UUCP (Sean Casey) writes: >Flame on! >... >I realize that OS9 and SysV are different products. Their target markets are >considerably different at the ends, but have a large overlap in the middle. >If Microware wants a good share of that middle, they are going to have to give >the developers a means to practically support the product. Otherwise, well, >I told you so! Sean, Sean, Sean, please be careful, the last time you flamed you totally melted a keyboard. And on your salary .... :-) Seriously... My experience with OS-9 (COCO) was that it was small enough that one could disassemble it and stare at it for a few days and understand it. In fact, the RS OS-9 manual set is almost good enough for that understanding as is. (It documents the internal data structures and such well enough that it's just a little thinking before you understand the system). Maybe once you actually started using the code it acts different. (I've not had much chance to play with it because I've only got one disk drive). I'm curious about one thing ... Does Microsoft distribute MSDOS source code to developers? Or just some fancy documentation? What exactly does Microware distribute for OS-9 developers? -- --- David Herron --- ARPA-> ukma!david@ANL-MCS.ARPA --- UUCP-> {ucbvax,unmvax,boulder,oddjob}!anlams!ukma!david --- {ihnp4,decvax,ucbvax}!cbosgd!ukma!david Hackin's in me blood. My mother was known as Miss Hacker before she married!
dibble@rochester.UUCP (Peter C. Dibble) (09/17/85)
> Flame on! > > OS9 has a real problem, and it's not in the software, but with the management. > Microware will not release source code to OS9, not even to developers! I find > - Sean Casey UUCP: sean@ukma.UUCP or Microware has incomplete source packages available for OS-9. For OS-9/68K you can buy a portpack which includes all the system dependant code and some of the other stuff. Everyone I know with a real need and a willingness to sign a non-disclosure agreement has been able to get any source they need from Microware.
moriarty@fluke.UUCP (Jeff Meyer) (09/18/85)
A quick correction: In article <347@wlbr.UUCP> steve@wlbr.UUCP (Steve Childress) writes: >o OS-9 exists in three flavors: > - Level I for 8-bit sytems using 64K on CPUs having a 6809 > - Level II, same but has bank-switching for up to 1MB on 6809s > - Level I for 680x0 machines with large memories (>=256K?) Actually, Level I for 680x0 machines recommends a minimum of 128K of user RAM) (256K is recommended) for general user purposes; however, since you could put the entire OS-9 operating system in ROM and only have enough RAM for OS-9 uses, this could be smaller for special applications. A small ROM-based control system running OS-9 could (according to the manuals) get by with 32K RAM. Anyone wanting more information on OS-9 should drop a line to: Andy Ball Microware Systems Corporation 1866 NW 114th Street Des Moines, Iowa 50322 >From the looks of things, I don't think it'd be tough to port OS-9 to an Amiga.... "For I perceive that behind this seemingly unrelated sequence of events, there lurks a singular, sinister attitude of mind." "Whose?" "MINE! HA-HA!" Moriarty, aka Jeff Meyer ARPA: fluke!moriarty@uw-beaver.ARPA UUCP: {uw-beaver, sun, allegra, sb6, lbl-csam}!fluke!moriarty <*> DISCLAIMER: Do what you want with me, but leave my employers alone! <*>
berger@datacube.UUCP (09/20/85)
We use OS9/68k as a real time kernel / executive as opposed to a
development system. I would never want to use it as a development
system though since I think unix is such a nice development
environment.
My feeling is this is where Microware is making a BIG mistake. They
should be promoting it against VRTX and MTOS. Not against UNIX and
MS/DOS. (I'd rather develop under OS9 than MS/DOS though! At least
Microware got the slashes to go in the right direction!).
We do all our development on a nice Pyramid 90x that runs both SYSTEM
V and Berkley 4.2 (We hardly ever use SYSTEM V). We cross compile
using the Microware C Cross compiler to prom based OS9 68000 on the
VME and Multibus. We have a binary downloader and make extensive use
of OS9 trap libraries to minimize download time.
OS9 has some great features over the other prom executives:
- Price
- Flexiblity
- Unix likeness (Not as much as it should though...)
- I/O system and file managers, heirarchical file system.
- Data Module / Trap library concept (This is the best thing about it!)
- You could use it as a disk based development system if you had to.
- Plus all the other things mentioned in the previous note.
We have added more unix likeness to it. Especially the ability to
write unix style device drivers in C, and System V style ioctls and
other system calls for easy source code portability.
We have also developed a tool for automatically building trap
libraries from object modules. This technique still needs some work
to make it completly general, in particular handling memory
allocation.
We payed Microware to port the Cross Compiler to the Pyramid. I
would like to see them make it a regular product. Officially they
only support the VAX (running unix or vms), PDP-11 (running unix) and
OS9 disk systems as cross hosts. (I'd like to see them support SUN
as well).
Overall I am pretty happy with OS9 as a target system. We will be
getting an upgrade any day now that will hopefully fix most of our C
compiler complaints. OS9 is significantly less expensive than other
routes. About half of what it would cost to get something like VRTX
and the Greenhills compiler. And VRTX does not have as much
functionality.
Of course VRTX might have better real-time context switching, etc and
Green Hills is probably a better compiler... Its all trade offs.
Also I think Microware WILL sell source for OS9 and the C compiler.
They just charge outragous amounts for source. They charge very
reasonable amounts for port-paks (everything you need to port to new
hardware including source for key things). Their licensing fees are
also significantly less than VRTX.
Bob Berger
Datacube Inc. 4 Dearborn Rd. Peabody, Ma 01960 617-535-6644
ihnp4!datacube!berger
decvax!cca!mirror!datacube!berger
{mit-eddie,cyb0vax}!mirror!datacube!bergerrb@ccivax.UUCP (rex ballard) (09/20/85)
> In response to the request for general info on OS9 ... a MC68xxx O/S ... > A brief discussion of OS-9: > > o Authored by Microware Systems Corp, Des Moines, Iowa > o It's about 5 years old, starting out as a spin-off from a contract > between Microware and Motorola. Mot. was looking for a super-BASIC > and an O/S to exploit the superb architecture of the MC6809. > o OS-9 exists in three flavors: > - Level I for 8-bit sytems using 64K on CPUs having a 6809 > - Level II, same but has bank-switching for up to 1MB on 6809s > - Level I for 680x0 machines with large memories (>=256K?) > Tandy sells it for their ($300) Color Computer ... it's a real ^^^^ the computer, not OS-9 > paradox in that (diminishing) market. Look for its manuals at your > nearby Tandy computer center. > The trouble (with OS9 for the 68000) is that it's not available on > a popular computer so there's never been any good non-system softare > so there's never been any users --- etc, etc, catch-22. Neither was CP/M 68K (re: TOS) until about two weeks ago. Intuition is a real commonly used OS around here? (I guess MS-DOS was a household word before the PC!!!) > I considered going to OS9/68K until I spent a weekend with an AT&T > UNIX-PC. For < $5,000, that machine, with its bit-mapped graphics & Sys 5.2 ^^^^^ > just beats the heck out of any other high-end machine for the serious > hobbiest. $5000 for a hobby machine?!!! (Wouldn't you rather have a buick?) > So in SUMMARY I'll editorialize: > > OS9-6809 is the best small-computer O/S I've seen but > it has never gained popularity to attract software houses. There is an awfull lot of "basic-09" and OS-9 code on radio-shack bulletin boards, much of it source. Since CoCo owners tend to get no software support from RS, OS-9 is the only way to get much of anything for it. (See earlier article about buying ROM to Throw it away (AMIGA DOS NOT IN ROM??)) I even remember seeing one called GKS.OS9 or something like that. This is one of the few systems where you can sell a LIBRARY! There is also some "C" stuff. Porting would involve expanding screen widths (CoCo uses 256x190 graphics or 32x16 text), or you could put them in windows. OS-9 uses a library of graphics commands similar to extended BASIC (GWBASIC to IBM'ers), which can be used from C or BASIC-09. Since I haven't got a GEM guide handy, maybe someone will tell me if there is something BASIC can do that GEM can't? I would guess that GEM could be called instead of a run-time library. If GEM can do more, great!!
uhclem@trsvax (09/20/85)
<Die!, die!, make your bit bucket flow-over!> /* Written 5:48 pm Sep 15, 1985 by ukma.UUCP!sean in trsvax:net.micro.68k */ Flame on! OS9 has a real problem, and it's not in the software, but with the management. Microware will not release source code to OS9, not even to developers! I find this apalling, and it is probably one of the main reasons that OS9 is not more popular than it is. They have really stuck it to themselves there, because AT&T is starting to push SysV for the 68000, for which you can get source code. ... /* End of text from trsvax:net.micro.68k */ Microware isn't alone. Microsoft does not normally give source out for anything. For example, MS-DOS is always shipped as a "black-box" with instructions on how to interface the drivers you write to the black box. This is their "link-kit" package. Even modules like FORMAT have a black-box that calls your disk driver to format, then you return and it calls you back with other requests to build the directory. Microsoft sends you sample drivers for some machine that is totally different from a IBM PC, (I think it was a Intel 310 or some CDC box) that you can use as a general guide (very general) on how to talk to the boxes they supply. If you buy their Xenix products, you get the same thing, unless you are willing to spend many kilo-bucks more for source, assuming that they are currently offering it for sale. I hate working without source, but that seems to be one of the current market trends. <The opinions expressed here may not be those of the management and personel of trsvax. "Thank you, Uh Clem." Frank Durda IV @ <trsvax!uhclem>
moriarty@fluke.UUCP (Jeff Meyer) (09/23/85)
In article <360@wlbr.UUCP> steve@wlbr.UUCP (Steve Childress) writes: >Is Microware's rep still listening to net.micro.6809? Any comment? Re 68K? Yes. And listening to net.micro.68k.
jpm@bnl44.UUCP (John McNamee) (09/25/85)
>OS9 has a real problem, and it's not in the software, but with the management. >Microware will not release source code to OS9, not even to developers! I find >this apalling, and it is probably one of the main reasons that OS9 is not more >popular than it is. They have really stuck it to themselves there, because >AT&T is starting to push SysV for the 68000, for which you can get source code. According to the licensing information I got from MicroWare about 6 months ago, full source code is provided to OEMs who purchase a high volume distribution license. You must also realize that OS9 source code isn't as necessary as Unix source code. It is possible to bring up OS9 on a totally new system without sources. It is also possible to extend the kernel without source code (i.e. you should be able to do NFS for OS9 without kernal sources). This is not to say that sources aren't a good thing to have (I would sure want them), but that they aren't as essential for OS9 as they are for Unix. But they are available, so this whole discussion is moot. -- John McNamee ..!decvax!philabs!sbcs!bnl44!jpm jpm@BNL44.ARPA "MS-DOS is a communist plot"
kim@mcrware.UUCP (Kim Kempf) (09/25/85)
> /* Written 5:48 pm Sep 15, 1985 by ukma.UUCP!sean in trsvax:net.micro.68k */ > Flame on! > OS9 has a real problem, and it's not in the software, but with the management. > Microware will not release source code to OS9, not even to developers! I find > this apalling, and it is probably one of the main reasons that OS9 is not more > popular than it is. They have really stuck it to themselves there, because > AT&T is starting to push SysV for the 68000, for which you can get source code. Full Source code for OS-9 is available, as UNIX source is from AT&T, but not for free. Check out the source code license prices for OS-9 and UNIX and see which you can afford! OS-9 is most commonly sold without full source code because the full source code is not required porting or maintenance. To change or develop a new driver requires only a small amount of system definition source to implement. In fact, adding another device to a running system is trival with OS-9. Try doing that with UNIX without recompiling/linking a new kernel module. Microware sells a product called a PortPak (TM) that includes all the software and software development tools required to install OS-9/68000 on a new system for evaluation by potential licensees or for special purpose, one-of-a-kind computers. A signed PortPak license must be received by Microware prior to shipment of the software. A standard Microware OEM license must be obtained prior to any kind of distribution. Each portpak includes: - OS-9/68000 Object Modules - Sample source code files for "boot rom" and I/O driver modules - Macro assembler, linker, file transfer and development utilities - A non-transferable single system software license PortPaks are available for the following operating systems: OS-9/6809 Level II OS-9/68000 Level I VAX/BSD 4.2 Unix VAX/UNIX System V VAX/VMS systems Contact Andy Ball at Microware for pricing information. Kim Kempf R & D Manager Microware Systems Corporation 1866 N.W 114th Street Des Moines, Iowa 50322 (515) 224-1929
kim@mcrware.UUCP (Kim Kempf) (09/25/85)
> My feeling is this is where Microware is making a BIG mistake. They > should be promoting it against VRTX and MTOS. Not against UNIX and > MS/DOS. (I'd rather develop under OS9 than MS/DOS though! At least > Microware got the slashes to go in the right direction!). > We are promoting OS-9 against UNIX not on VAXes, Pyramids or other high-end hardware but on the lower end (under $5000). Any system that is capable of running UNIX requires 1) hard disk(s) and 2) memory management. Both of these elements fix the minimum cost for the system at a relatively high base value. Consider one floppy disk and no memory management, the base cost for the machine drastically drops into the affordable range. Unfortunately, UNIX can't be squished enough to operate on such a configuration. This is where OS-9 comes in. > Also I think Microware WILL sell source for OS9 and the C compiler. > They just charge outragous amounts for source... Again, compare the source code prices to a UNIX license.
jpm@bnl44.UUCP (John McNamee) (09/27/85)
> We use OS9/68k as a real time kernel / executive as opposed to a > development system. I would never want to use it as a development > system though since I think unix is such a nice development > environment. > > My feeling is this is where Microware is making a BIG mistake. They > should be promoting it against VRTX and MTOS. ... > > We do all our development on a nice Pyramid 90x that runs both SYSTEM > V and Berkley 4.2 (We hardly ever use SYSTEM V). We cross compile > using the Microware C Cross compiler to prom based OS9 68000 on the > VME and Multibus. I'm happy you can afford a nice big Unix system. I sure can't, and there are more people in my position than there are in yours. Given the choice between developing under OS/9 or Unix, I would rather have Unix, but only if I was given the proper hardware to run it on. Unix runs like sh*t on small 68K systems with slow drives and limited memory. Given a limited hardware budget, OS/9 makes a lot more sense than Unix. -- John McNamee ..!decvax!philabs!sbcs!bnl44!jpm jpm@BNL44.ARPA "MS-DOS is a communist plot"
jimomura@lsuc.UUCP (Jim Omura) (09/29/85)
In article <2198@ukma.UUCP> david@ukma.UUCP (David Herron, NPR Lover) writes: >In article <2193@ukma.UUCP> sean@ukma.UUCP (Sean Casey) writes: >>Flame on! >>... >>I realize that OS9 and SysV are different products. Their target markets are >>considerably different at the ends, but have a large overlap in the middle. >>If Microware wants a good share of that middle, they are going to have to give >>the developers a means to practically support the product. Otherwise, well, >>I told you so! > >Sean, Sean, Sean, please be careful, the last time you flamed you totally >melted a keyboard. And on your salary .... :-) > >Seriously... My experience with OS-9 (COCO) was that it was small enough >that one could disassemble it and stare at it for a few days and understand >it. In fact, the RS OS-9 manual set is almost good enough for that >understanding as is. (It documents the internal data structures and such >well enough that it's just a little thinking before you understand the >system). > >Maybe once you actually started using the code it acts different. (I've >not had much chance to play with it because I've only got one disk drive). > >I'm curious about one thing ... Does Microsoft distribute MSDOS source >code to developers? Or just some fancy documentation? > >What exactly does Microware distribute for OS-9 developers? >-- >--- David Herron >--- ARPA-> ukma!david@ANL-MCS.ARPA >--- UUCP-> {ucbvax,unmvax,boulder,oddjob}!anlams!ukma!david >--- {ihnp4,decvax,ucbvax}!cbosgd!ukma!david > >Hackin's in me blood. My mother was known as Miss Hacker before she married! Ummm. Guys, it'd be nice to have more Amiga stuff than OS-9 stuff 'here'. I *would* like to see OS-9 on Amiga and ST-520, but unless someone is going to say "I'm going to port OS-9 to the Amiga", I'd think most people would rather hear what the Amiga does. OS-9 has it's own net. Anyway, you don't need source to develop for OS-9. The only time you need source is when the system doesn't work as specified (a known flaw of *Unix* not OS-9 -- OS-9 works). The documentation available for OS-9 (full manuals cost $40.00 from Microware, so you can buy them and find out pretty much everything) is quite adequate. To quote the OS-9 Tech manual: "All system calls are executed via an SWI2 instruction. 1. Load the 6809 register with any appropriate parameters. 2. Execute a SWI2 instruction, followed immediately by a constant byte, which is the request code. 3. After OS-9 processes the call, it returns any param- eters in the 6809 register. If an error occurred, the C bit of the Condition Code Register is set, and Acc- umulator B contains the appropriate error code. This permits a BCS or BCC instruction immediately following the system call to branch on error/no error." Each call is documented with all the necessary information. And, no, Microsoft won't give you MS-DOS source-code either and yes, you can disassemble OS-9 modules and system calls *iff* you get a disassembler which is extra cost, but fairly cheaply had, especially from 2nd sources. -- James Omura, Barrister & Solicitor, Toronto ihnp4!utzoo!lsuc!jimomura
jimomura@lsuc.UUCP (Jim Omura) (09/29/85)
In article <272@ccivax.UUCP> rb@ccivax.UUCP (rex ballard) writes: >> In response to the request for general info on OS9 ... a MC68xxx O/S ... >> A brief discussion of OS-9: >> >> Tandy sells it for their ($300) Color Computer ... it's a real > ^^^^ the computer, not OS-9 >> paradox in that (diminishing) market. Look for its manuals at your Nobody has accurate figures on Color Computer sales. I suspect the conclusion that the market is shrinking is right, but Tandy only knows. There are rumours of a new computer coming out in the next year, but then again, there were rumours a year ago too. In fact the parts of one computer that was *almost* marketted are now being sold for the CoCo in some US stores. >> nearby Tandy computer center. > >> The trouble (with OS9 for the 68000) is that it's not available on >> a popular computer so there's never been any good non-system softare >> so there's never been any users --- etc, etc, catch-22. What?!!! There's Stylograph (word processor w/mail merge & spelling checker, DynaCalc (spread sheet) and Sculptor (applications generator which you can consider as taking the place of a dBASEIII). This is all that a *lot* of people *ever* use. These are just the ones that I know about too, there are probably more that I don't know about. There is a rich and growing public domain, as you mention later, specifically for OS-9, and the 'C' compiler will translate many Unix calls, making the general public domain 'C' programs and especially the Unix programs fairly readily available. The Pascal should be able to handle a lot of Public Domain programs as well. T.G. Lewis' "Microbook" database manager should work fine, because it comes with an ASCII source file (I can read single sided IBM PC formatted discs on my Color Computer under OS-9 using a utility called 'PCread' and write single sided PC disks with 'PCwrite'). > >Neither was CP/M 68K (re: TOS) until about two weeks ago. >Intuition is a real commonly used OS around here? >(I guess MS-DOS was a household word before the PC!!!) > >> I considered going to OS9/68K until I spent a weekend with an AT&T >> UNIX-PC. For < $5,000, that machine, with its bit-mapped graphics & Sys 5.2 > ^^^^^ >> just beats the heck out of any other high-end machine for the serious >> hobbiest. > > $5000 for a hobby machine?!!! (Wouldn't you rather have a buick?) > >> So in SUMMARY I'll editorialize: >> >> OS9-6809 is the best small-computer O/S I've seen but >> it has never gained popularity to attract software houses. > >There is an awfull lot of "basic-09" and OS-9 code on radio-shack bulletin >boards, much of it source. Since CoCo owners tend to get no software >support from RS, OS-9 is the only way to get much of anything for it. >(See earlier article about buying ROM to Throw it away (AMIGA DOS NOT IN ROM??)) >I even remember seeing one called GKS.OS9 or something like that. >This is one of the few systems where you can sell a LIBRARY! > >There is also some "C" stuff. Porting would involve expanding screen >widths (CoCo uses 256x190 graphics or 32x16 text), or you could put them >in windows. OS-9 uses a library of graphics commands similar to >extended BASIC (GWBASIC to IBM'ers), which can be used from C or BASIC-09. Most OS-9'ers, including Color Computer OS-9'er (such as myself) have hardware 80 column displays. In fact, my scroll rate is faster than an IBM-PC (due to dedicated RAM and a 6845 with it's own clock), I also have smooth scroll, redefinable charactersets, horizontal redefinition (I haven't tried 132 columns but apparently I can do this too), and possibly high-res graphics (black & white) beyond 256 * 192. We also have hard disk systems available. In fact, as a practical matter, OS-9 is better for systems of say 20 meg. than real Unix due to code efficiency and less online surplussage. Putting it into scale, I run 1 meg. online (3 double sided, double density, 40 track floppies) and don't find it enough. I could go 80 track, but really I should get a hard disk of about 10 - 20 meg. I doubt if anyone here is surprised. > >Since I haven't got a GEM guide handy, maybe someone will tell me if >there is something BASIC can do that GEM can't? I would guess that >GEM could be called instead of a run-time library. If GEM can do more, >great!! Huh? GEM isn't a language. -- James Omura, Barrister & Solicitor, Toronto ihnp4!utzoo!lsuc!jimomura
peter@graffiti.UUCP (Peter da Silva) (10/02/85)
> We are promoting OS-9 against UNIX not on VAXes, Pyramids or other high-end > hardware but on the lower end (under $5000). Any system that is capable > of running UNIX requires 1) hard disk(s) and 2) memory management. Both 1) False. HP put out a very nice little machine called the Integral that not only doesn't need a hard disk, but doesn't really work well with one. 2) True, but... I don't want to do serious work on a machine without an MMU any more. It's so frustrating... even sickening... to watch a lost pointer or bad copy of foocalc blow away everything & write garbage over a:\*.* > of these elements fix the minimum cost for the system at a relatively > high base value. Consider one floppy disk and no memory management, the base Why is an MMU that expensive? > cost for the machine drastically drops into the affordable range. > Unfortunately, UNIX can't be squished enough to operate on such a > configuration. This is where OS-9 comes in. Sure can. Xenix, Venix, and PC/IX all run without an MMU, apparently quite well. HPUX runs without a hard drive, and I'd use it if the Integral didn't have a typical Hewlett-Packard price tag. Neither seem to be necessary for UNIX. Personally, I don't care whether it's called UNIX, Xenix, OS/9, or Intuition. Just so long as I can run rogue on it :->.
henry@utzoo.UUCP (Henry Spencer) (10/03/85)
> ... Consider one floppy disk and no memory management, the base > cost for the machine drastically drops into the affordable range. > Unfortunately, UNIX can't be squished enough to operate on such a > configuration... Tsk, tsk, ignorant newcomers. Unix can be and has been run on such machines. See "Unix on a Microprocessor" in the July-August 1978 Bell System Technical Journal, for example. More modern Unixes are harder to squish that far, and everyone agrees that hard disks do wonders for the performance of most any operating system, but it's not impossible. Check out Unix on the Data General 1, for example; yes, it exists. -- Henry Spencer @ U of Toronto Zoology {allegra,ihnp4,linus,decvax}!utzoo!henry
rb@ccivax.UUCP (rex ballard) (10/04/85)
> >Since I haven't got a GEM guide handy, maybe someone will tell me if > >there is something BASIC can do that GEM can't? I would guess that > >GEM could be called instead of a run-time library. If GEM can do more, > >great!! > > Huh? GEM isn't a language. > > > > -- > James Omura, Barrister & Solicitor, Toronto > ihnp4!utzoo!lsuc!jimomura Let me clarify this (sorta). Basic for the Coco has some nice graphics routines (plot, line, fill...) as part of the ROM, could Basic-09 call Gem routines, or are their equivelant VDI routines available in OS-9. By the way, thanks for the list of application software.
jimomura@lsuc.UUCP (Jim Omura) (10/07/85)
In article <294@ccivax.UUCP> rb@ccivax.UUCP (rex ballard) writes: >> >Since I haven't got a GEM guide handy, maybe someone will tell me if >> >there is something BASIC can do that GEM can't? I would guess that >> >GEM could be called instead of a run-time library. If GEM can do more, >> >great!! >> >> Huh? GEM isn't a language. >> >> >> >> -- >> James Omura, Barrister & Solicitor, Toronto >> ihnp4!utzoo!lsuc!jimomura > >Let me clarify this (sorta). Basic for the Coco has some nice graphics >routines (plot, line, fill...) as part of the ROM, could Basic-09 call >Gem routines, or are their equivelant VDI routines available in OS-9. > >By the way, thanks for the list of application software. BASIC09 comes with a graphics extension library called GFX which is called as a Procedure from BASIC09. There's no reason that GFX couldn't be redone for the Atari or any other machine, but subject to a caveat. The Color Computer GFX is oriented to the Color Computer's video chip's limits (4 color sets, 256 x 192 resolution also interfacing to Joystick is in GFX now that I'm mentioning it). The choice would have to be made whether to use the same parameters for portability or the relevant Atari parameters. Personally, I feel that GFX should be re-written to be NAPLPS based in instructions. The NAPLPS approach (hardware independent) is the best I've seen. -- James Omura, Barrister & Solicitor, Toronto ihnp4!utzoo!lsuc!jimomura Byte Information eXchange: jimomura Compuserve: 72205,541 MTS at WU: GKL6
kim@mcrware.UUCP (Kim Kempf) (10/07/85)
> > We are promoting OS-9 against UNIX not on VAXes, Pyramids or other high-end > > hardware but on the lower end (under $5000). Any system that is capable > > of running UNIX requires 1) hard disk(s) and 2) memory management. Both > > 1) False. HP put out a very nice little machine called the Integral > that not only doesn't need a hard disk, but doesn't really work > well with one. That HP machine is a single-user machine. The OS was an HP in-house clone of UNIX. Not even in the same class as OS-9. > > 2) True, but... I don't want to do serious work on a machine without > an MMU any more. It's so frustrating... even sickening... to watch > a lost pointer or bad copy of foocalc blow away everything & write > garbage over a:\*.* > The MMU does a little more than protect memory. It allows manipulation of a process' logical address spaces. In fact, the UNIX fork system call CANNOT operate without an MMU (or some type of segment/base register allocation scheme [ala 80*86]). > Why is an MMU that expensive? > Not only in dollars. In bus cycles. Re: MC68451. > Sure can. Xenix, Venix, and PC/IX all run without an MMU, apparently quite > well. HPUX runs without a hard drive, and I'd use it if the Integral didn't > have a typical Hewlett-Packard price tag. Neither seem to be necessary for > UNIX. Sorry. Those machines use Intel processors with their segment register on-chip MMUs. In this case the MMU is free. ---------------- Kim Kempf, Microware Systems Corporation {{cornell,decvax,ihnp4,sdcsvax,tektronix}!uw-beaver}\ {allegra,gatech!sb1,hplabs!lbl-csam,decwrl!sun,sunup} >!fluke!mcrware!kim {ssc-vax,hplsla,wavetek,physio,cae780,tikal,telematic}/
jimomura@lsuc.UUCP (Jim Omura) (10/09/85)
In article <275@graffiti.UUCP> peter@graffiti.UUCP (Peter da Silva) writes: >> hardware but on the lower end (under $5000). Any system that is capable Actually, as I think I've said, OS-9 has advantages over real Unix well above the $5,000.00 range. You software developers have really funny notions about end users. We end users need lots of memory. Unix eats memory for breakfast and asks for more. My law practice needs at least 20 meg. of disk and 256K for really unrestricted use of OS-9. If I had Unix with that much I'd be back in the same position I'm in now. I don't believe that an end user in my position could use a Unix system with less than about 1 meg. of RAM and 40 - 60 meg. of disk online. Unix overhead is fairly large even if you offload a lot of the unused stuff. But the real problem is that you may find you *need* some of the stuff you presumed was surplusage for later software. As such, even if you think you have an adequate system now, with a small change in software you could find out that you've doubled your minimum and comfortable system levels soon after. My current experience leads me to believe that a system developer can get away with using a much less capable system than an end user. He may *choose* to have a bigger system, but he could easily get by without. >> of running UNIX requires 1) hard disk(s) and 2) memory management. Both > > 1) False. HP put out a very nice little machine called the Integral > that not only doesn't need a hard disk, but doesn't really work > well with one. Maybe. Hard to define "nice". > > 2) True, but... I don't want to do serious work on a machine without > an MMU any more. It's so frustrating... even sickening... to watch > a lost pointer or bad copy of foocalc blow away everything & write > garbage over a:\*.* OS-9 checks CRC before executing. It hasn't got anything to do with *needing* an MMU (I'm not saying MMU's don't help ... ). Other protection schemes can also be added. Gimix OS-9 has protection. > >> of these elements fix the minimum cost for the system at a relatively >> high base value. Consider one floppy disk and no memory management, the base > >Why is an MMU that expensive? > >> cost for the machine drastically drops into the affordable range. >> Unfortunately, UNIX can't be squished enough to operate on such a >> configuration. This is where OS-9 comes in. > >Sure can. Xenix, Venix, and PC/IX all run without an MMU, apparently quite >well. HPUX runs without a hard drive, and I'd use it if the Integral didn't >have a typical Hewlett-Packard price tag. Neither seem to be necessary for >UNIX. (you forgot QNX too, but that's another matter) Take the Tandy 6000 for an example. With the same hardware, an OS-9 would allow me more disk space and more user memory space. Some applications might even run a bit faster (unlikely that any would run slower). For a small time user like me (1 man law-firm) and up to say a 20 man lawfirm, I can't see what advantage Unix would have. Lawyers are *not* made of money. Other end users would probably prefer OS-9 too if they knew what the real choice was. Don't you think it's time you guys got out of the ivory towers and thought about what end users need? > >Personally, I don't care whether it's called UNIX, Xenix, OS/9, or Intuition. >Just so long as I can run rogue on it :->. Good point. I've never used 'rogue' but I hear it's good. -- James Omura, Barrister & Solicitor, Toronto ihnp4!utzoo!lsuc!jimomura Byte Information eXchange: jimomura Compuserve: 72205,541 MTS at WU: GKL6
david@ukma.UUCP (David Herron, NPR Lover) (10/10/85)
In article <138@mcrware.UUCP> kim@mcrware.UUCP (Kim Kempf) writes: >> 2) True, but... I don't want to do serious work on a machine without >> an MMU any more. It's so frustrating... even sickening... to watch >> a lost pointer or bad copy of foocalc blow away everything & write >> garbage over a:\*.* >> >The MMU does a little more than protect memory. It allows manipulation of a >process' logical address spaces. In fact, the UNIX fork system call CANNOT >operate without an MMU (or some type of segment/base register allocation >scheme [ala 80*86]). Tell that to the people who put together V6 Mini Unix. We ran that sucker on a PDP-11/10. NO mmu in that beast. And it regularly died every time you had a pointer fault in your C program. It was bad enough that the friend of mine that ran the machine patched up the shell so it would sync every time it printed a prompt or ran a program. This way he wouldn't have to fix the disk so often. (No fsck). Sorry, I don't know any details of the system -- The only time I really looked at V6 was as an intelectual exsercise, I needed to understand the basic design structure of Unix. -- David Herron, ukma!david@ANL-MCS.ARPA, cbosgd!ukma!david (Soon -- david@UKMA.BITNET, and (hopefully) david@ukma.csnet) Hackin's in me blood! My mother was known as Miss Hacker before she married!
guy@sun.uucp (Guy Harris) (10/12/85)
> > Why is an MMU that expensive? > Not only in dollars. In bus cycles. Re: MC68451. *Some* MMUs are expensive in bus cycles. The one on this machine does not slow down access to memory. It may not be inexpensive in cost, but it's certainly cheap in bus cycles. Guy Harris
peter@graffiti.UUCP (Peter da Silva) (10/13/85)
> In article <275@graffiti.UUCP> peter@graffiti.UUCP (Peter da Silva) writes: > >> hardware but on the lower end (under $5000). Any system that is capable > > Actually, as I think I've said, OS-9 has advantages over real > Unix well above the $5,000.00 range. You software developers have > really funny notions about end users. We end users need lots of memory. Then why are end-users working quite well with IBM-PCs with 2 320K disk drives and 256K of RAM? Someone in that position would be able to survive quite well on an HP-integral with 2 720K drives (that being the size they come in) and 512K of RAM (since it puts '/' and '/tmp' in RAM and ROM). > Unix eats memory for breakfast and asks for more. My law practice > needs at least 20 meg. of disk and 256K for really unrestricted use > of OS-9. If I had Unix with that much I'd be back in the same position > I'm in now. Which UNIX? You can fit all the *required* UNIX utilities into 1 Meg. Since UNIX was originally written for a machine with a 64K-per-task limitation (I know the 6809 is even more limited... the point is that the 68000 is a much bigger machine than either O/S really needs), 256K should be plenty. So long as you stick to V7 and avoid SIII and SV you shouldn't have any problems. > I don't believe that an end user in my position could use > a Unix system with less than about 1 meg. of RAM and 40 - 60 meg. of > disk online. Unix overhead is fairly large even if you offload a > lot of the unused stuff. But the real problem is that you may find > you *need* some of the stuff you presumed was surplusage for later > software. As such, even if you think you have an adequate system now, > with a small change in software you could find out that you've doubled > your minimum and comfortable system levels soon after. I don't see this. Have you experienced this problem or are you just imagining it. I can see how 1 MEG of RAM couldn't hurt, but I don't see how UNIX would require an extra 20-40 Meg of disk space. > My current experience leads me to believe that a system developer > can get away with using a much less capable system than an end user. My experience is the opposite. I have seen *lots* of end-users with 256K 2-drive IBM-PCs. When I have to use a system that small I have to start juggling disks, putting the editor and compiler on seperate disks, and sometimes even having to copy a program onto a blank diskette to compile it. But it's fine for 123 or dBase. > He may *choose* to have a bigger system, but he could easily get by > without. If the above technique can be called "getting by", I guess. I can do real work on a DG-1 with 512K and 2 720K microfloppies, but not on a Compaq with 256K and 2 standard drives. I can also do real work on an HP-Integral with 512K and two 720K floppies. And have a less frustrating time doing it. > >> of running UNIX requires 1) hard disk(s) and 2) memory management. Both > > > > 1) False. HP put out a very nice little machine called the Integral > > that not only doesn't need a hard disk, but doesn't really work > > well with one. > > Maybe. Hard to define "nice". Nice means that it is competitive with the other luggables, and allows real software development in a UNIX environment. > > 2) True, but... I don't want to do serious work on a machine without > > an MMU any more. It's so frustrating... even sickening... to watch > > a lost pointer or bad copy of foocalc blow away everything & write > > garbage over a:\*.* > > OS-9 checks CRC before executing. It hasn't got anything to do > with *needing* an MMU (I'm not saying MMU's don't help ... ). A program can have a good CRC (meaning the author thought it was good) and still have a bad bug that can blow everything away. Also, with an MMU you can't write a Killer Prolok. > Other protection schemes can also be added. Gimix OS-9 has protection. How does it work? > Take the Tandy 6000 for an example. With the same hardware, an > OS-9 would allow me more disk space and more user memory space. But is an OS/9 available? And how much extra is it going to cost? > Some > applications might even run a bit faster (unlikely that any would run > slower). They could. What sort of disk cache does OS/9 use? > For a small time user like me (1 man law-firm) and up to > say a 20 man lawfirm, I can't see what advantage Unix would have. How about compatibility? How about UUCP? Or are you running Usenet on OS/9? How about the ability to upgrade to a bigger machine without having to learn yet another operating system... > Lawyers are *not* made of money. Other end users would probably prefer > OS-9 too if they knew what the real choice was. Don't you think > it's time you guys got out of the ivory towers and thought about > what end users need? I'm in an ivory tower, now? I believe that end-users need something better than Messy-Dos and other obsolete and incomplete operating systems. I agree that OS/9 is a good choice for such a situation. All I'm saying is that UNIX is too. My experience has been that, on similar hardware, UNIX is actually smaller and faster than other operating systems of equivalent capabilities. I would like a nice, cheap, floppy-based home computer with a 68000 and OS/9. Or with a 68000 and UNIX. Or a 32032 and UNIX. Or a Z8000 and UNIX, ... But in the meantime I think I'll get an AMIGA.
peter@graffiti.UUCP (Peter da Silva) (10/16/85)
microware!kim: We don't need a bunch of flames. I was just trying to point out that UNIX is perfectly viable on low-end systems. Don't assume I'm attacking your baby. I'm not. All I'm doing is defending mine :->. > That HP machine is a single-user machine. The OS was an HP in-house > clone of UNIX. Not even in the same class as OS-9. The O/S *is* UNIX: it's a very carefully enhanced port. The only reason the machine is billed as single-user is that it is extremely low on disk space. There is no reason you can't run a shell out a serial port & run two users. It'll just be real frustrating... > > 2) True, but... I don't want to do serious work on a machine without > > an MMU any more. It's so frustrating... even sickening... to watch > > a lost pointer or bad copy of foocalc blow away everything & write > > garbage over a:\*.* > > > The MMU does a little more than protect memory. It allows manipulation of a > process' logical address spaces. In fact, the UNIX fork system call CANNOT > operate without an MMU (or some type of segment/base register allocation > scheme [ala 80*86]). If you are careful about the code your compiler generates there's no reason you can't. That's all the UNIX-es for the PC do. > > Sure can. Xenix, Venix, and PC/IX all run without an MMU, apparently quite > > well. HPUX runs without a hard drive, and I'd use it if the Integral didn't > > have a typical Hewlett-Packard price tag. Neither seem to be necessary for > > UNIX. > Sorry. Those machines use Intel processors with their segment register > on-chip MMUs. In this case the MMU is free. The Intel segment registers qualify for on-chip MMUs about as well as I qualify as a race-car driver. If you generate properly position-independant code, which should be possible for the 68000 or the 6809, there's no reason you can't port UNIX to it.
steve@wlbr.UUCP (Steve Childress) (10/18/85)
I too did not like Microware's Module CRC scheme (in level I, I assume
that level II is likewise encumbered). The idea was to validate that
ROMs and EPROMs existed at boot-time. I believe that the CRC was intended
to detect ROMd modules (Mot's "Silcon Software" crusade), and
not to catch memory or I/O errors. But someone got carried away
and used the CRC scheme for modules and programs fetched from disk.
I say it was a mistake because disk files themselves have error checking.
When I NOP'd the call to CRC check LOAD'ed modules I noted a large speed-up
in hard disk systems. I sent the patches for the APPLE, CoCo, and SS50
OS9 level I's to '68 Micro Journal (magazine) to publish for the users
but they never printed..
Regards,
Steve Childress
Eaton IMS R&D Group MS 43
31717 La Tienda Drive
Westlake Village, CA 91360
{trwrb, scgvaxd, ihnp4, voder, vortex} !wlbr!steve
or ...wlbr!wlbreng1!stevejimomura@lsuc.UUCP (Jim Omura) (10/20/85)
In article <379@wlbr.UUCP> steve@wlbr.UUCP (Steve Childress) writes: > >I say it was a mistake because disk files themselves have error checking. > >When I NOP'd the call to CRC check LOAD'ed modules I noted a large speed-up >in hard disk systems. I sent the patches for the APPLE, CoCo, and SS50 >OS9 level I's to '68 Micro Journal (magazine) to publish for the users >but they never printed.. > > Regards, > Steve Childress > Eaton IMS R&D Group MS 43 > 31717 La Tienda Drive > Westlake Village, CA 91360 > {trwrb, scgvaxd, ihnp4, voder, vortex} !wlbr!steve > or ...wlbr!wlbreng1!steve Uh. mmmm. Steve, what error checking? At write time? Doesn't catch everything. I've had corrupted files. At read all it does is say that it seems to have gotten what seems to have been written. Part of the problem is boarderline latency. A write may verify at the time it's done, but later, the magnetic pattern may not be as clear. Also, poor storage techniques of magnetic environmental 'noise' can cause this. I'd keep the CRC if I were you. Especially for executable binary files. Cheers! -- Jim O. -- James Omura, Barrister & Solicitor, Toronto ihnp4!utzoo!lsuc!jimomura Byte Information eXchange: jimomura Compuserve: 72205,541 MTS at WU: GKL6
davidsen@steinmetz.UUCP (Davidsen) (10/23/85)
People who assume that UNIX must have huge quantities of {disk, memory, cpu
cycles, etc} are simply repeating the misinformation they have heard
elsewhere. I recently helped someone set up a small office system with the
following confiuration: 1 640k IBM PC clone ($2200), two terminals ($800),
one copy of PC/IX ($720). This included enough word processing, editors,
etc to run an office which does lots of word processing. At some time in
the future some additional software will be installed, but the system
including printer was < $5k, and runs in 640k for three users. The nroff
processor is set up as a queue to keep too many jobs from killing the
machine.
I think I could do this for < $10k using an AT&T 7300 (or 3b1) which
would give me more memory, 68k processor, virtual memory, and access to
many packages in the future. Even that package would be only 20Mb (or at
most 40) and 1 Mb of memory. UNIX is not always a hog of anything in
particular, although the use of processors with verbose instruction sets
(like the 68k) can increase the memory and disk somewhat. 8086 programs
run 30-50% smaller, since many instructions are only 1 byte.
Note that the software includes all packages, including uucp, accounting,
and even games! This takes about 6Mb of disk. The clone they got runs a
20Mb, so there was no need to pare it. When the load increases they will
buy a second machine and link with uucp (no need for shared databases in
this case).
Let a few hard facts shine on the matter.
Bill Davidsen
"It seemed like a good idea at the time..."jimomura@lsuc.UUCP (Jim Omura) (10/29/85)
In article <286@steinmetz.UUCP> davidsen@kbsvax.UUCP (Davidsen) writes: > >People who assume that UNIX must have huge quantities of {disk, memory, cpu >cycles, etc} are simply repeating the misinformation they have heard >elsewhere. I recently helped someone set up a small office system with the >following confiuration: 1 640k IBM PC clone ($2200), two terminals ($800), >one copy of PC/IX ($720). This included enough word processing, editors, >etc to run an office which does lots of word processing. At some time in >the future some additional software will be installed, but the system >including printer was < $5k, and runs in 640k for three users. The nroff With 640K under OS-9 I could support 6-8 for similar work (Level II OS-9 6809 version--it's unclear to me what the overhead of 68K version OS-9 really is at this time. On the one hand, the code should be smaller due to direct addressing instead of mmu bank juggling (2K or 4K banks are swapped in Level II depending on the computer), but there are some added features in 68K OS-9 which probably increase the size again). Welcome to the *real* world of OS-9! >processor is set up as a queue to keep too many jobs from killing the >machine. > >I think I could do this for < $10k using an AT&T 7300 (or 3b1) which >would give me more memory, 68k processor, virtual memory, and access to >many packages in the future. Even that package would be only 20Mb (or at >most 40) and 1 Mb of memory. UNIX is not always a hog of anything in >particular, although the use of processors with verbose instruction sets >(like the 68k) can increase the memory and disk somewhat. 8086 programs >run 30-50% smaller, since many instructions are only 1 byte. Curious. Most people I talk to tell me that programs under 68K are generally smaller (except for very trivial routines) because it takes fewer instructions to accomplish anything with a 68K. > >Note that the software includes all packages, including uucp, accounting, >and even games! This takes about 6Mb of disk. The clone they got runs a >20Mb, so there was no need to pare it. When the load increases they will >buy a second machine and link with uucp (no need for shared databases in >this case). > >Let a few hard facts shine on the matter. > >Bill Davidsen > >"It seemed like a good idea at the time..." Cheers! -- Jim O. -- James Omura, Barrister & Solicitor, Toronto ihnp4!utzoo!lsuc!jimomura Byte Information eXchange: jimomura Compuserve: 72205,541 MTS at WU: GKL6
steve@wlbr.UUCP (Steve Childress) (10/31/85)
In article <860@lsuc.UUCP>, jimomura@lsuc.UUCP (Jim Omura) writes: > In article <379@wlbr.UUCP> steve@wlbr.UUCP (Steve Childress) writes: > > > >I say it was a mistake because disk files themselves have error checking. > > > >When I NOP'd the call to CRC check LOAD'ed modules I noted a large speed-up > >in hard disk systems. I sent the patches for the APPLE, CoCo, and SS50 > >OS9 level I's to '68 Micro Journal (magazine) to publish for the users > >but they never printed.. > > > > Uh. mmmm. Steve, what error checking? At write time? Doesn't > catch everything. I've had corrupted files. At read all it does is > say that it seems to have gotten what seems to have been written. > Part of the problem is boarderline latency. A write may verify at > the time it's done, but later, the magnetic pattern may not be as > clear. Also, poor storage techniques of magnetic environmental > 'noise' can cause this. > > I'd keep the CRC if I were you. Especially for executable binary > files. > > James Omura, Barrister & Solicitor, Toronto > ihnp4!utzoo!lsuc!jimomura I've gotta rebut your claim that OS9's CRC will protect you from "magnetic environmental 'noise'". If your disk flux's get hosed up due to magnetic fields, your disk controller will discover same when it is read. The probability of an undetected garble of data using the popular CRC-16 scheme is tiny. Did you know that many inexpensive computers do no error checking on their memory and I/O busses? And cheap floppies and winchesters do NOT have a read-after-write check as you assume. All I can say is that I ran OS9 with CRC defeated for two or more years and never needed to reformat my hard disk. The only corrupted files I saw were caused by wayward programs (and a lack of protection by OS9). Regards, Steve Childress Eaton IMS R&D Group MS 43 31717 La Tienda Drive Westlake Village, CA 91360 (818) 889-2211 X2148 {trwrb, scgvaxd, ihnp4, voder, vortex} !wlbr!steve or ...wlbr!wlbreng1!steve
peter@graffiti.UUCP (Peter da Silva) (10/31/85)
> In article <286@steinmetz.UUCP> davidsen@kbsvax.UUCP (Davidsen) writes: > > > >People who assume that UNIX must have huge quantities of {disk, memory, cpu > > ... > >the future some additional software will be installed, but the system > >including printer was < $5k, and runs in 640k for three users. The nroff > > With 640K under OS-9 I could support 6-8 for similar work (Level II OS-9 But how many people would trust a multiuser system with no memory management? For single-user use, yes, but multiuser? -- Name: Peter da Silva Graphic: `-_-' UUCP: ...!shell!{graffiti,baylor}!peter IAEF: ...!kitty!baylor!peter
jimomura@lsuc.UUCP (Jim Omura) (11/06/85)
In article <375@graffiti.UUCP> peter@graffiti.UUCP (Peter da Silva) writes: >> In article <286@steinmetz.UUCP> davidsen@kbsvax.UUCP (Davidsen) writes: >> > >> >People who assume that UNIX must have huge quantities of {disk, memory, cpu >> > ... >> >the future some additional software will be installed, but the system >> >including printer was < $5k, and runs in 640k for three users. The nroff >> >> With 640K under OS-9 I could support 6-8 for similar work (Level II OS-9 > >But how many people would trust a multiuser system with no memory management? >For single-user use, yes, but multiuser? >-- >Name: Peter da Silva Well, first of all, as I've said before, a Gimix Level II system (which is the one I'd probably use) has memory management, and dedicated IO processors for that matter. But memory management isn't that much an issue when you consider we're talking about word processing and possibly a bit of database stuff -- *not* development work. End users running say Databank and Stylograph don't need to worry about memory management. It simply wouldn't help even if you had it. By the way, I also should have commented on the fact that the original message stated that he'd only need 6 meg. of disk for system overhead. *Only* 6 meg.? An equivalent OS-9 could probably be done in under 2 meg. Keep in mind that the system described only had something like 20 meg. of disk in total (I can't remember exactly what he said here) leaving 14 meg. for user memory. As a business person, I think I'd need about 4-5 meg. myself. Some of this could be common documents and data, but I wonder how his 3 users are going to manage? At 20 meg., the OS-9 equivalent would yield 18 meg. for users (this is why I said before that you don't get close to equivalent user value in 10 meg. systems before) The ratio of 7/9 is just starting to even out the effect of overhead. Someone else once pointed out that I wouldn't have Unix mail on OS-9/ This is true. I only have Unix mail through lsuc. Still, check my address list again. Unix mail is nice. It's also not the only way to go. (Of course if I *really* wanted to, I could write a 'mail' utility, but I wouldn't bother). Cheers! -- Jim O. -- James Omura, Barrister & Solicitor, Toronto ihnp4!utzoo!lsuc!jimomura Byte Information eXchange: jimomura Compuserve: 72205,541 MTS at WU: GKL6
peter@graffiti.UUCP (Peter da Silva) (11/08/85)
> Well, first of all, as I've said before, a Gimix Level II system > (which is the one I'd probably use) has memory management, and dedicated > IO processors for that matter. But memory management isn't that much > an issue when you consider we're talking about word processing and possibly > a bit of database stuff -- *not* development work. End users running > say Databank and Stylograph don't need to worry about memory management. > It simply wouldn't help even if you had it. (1) There are no bug-free programs. (2) I *AM* talking about development work. If all you're doing is running canned software you might as well stick with CP/M... it's a lot cheaper than either OS/9 or UNIX. -- Name: Peter da Silva Graphic: `-_-' UUCP: ...!shell!{graffiti,baylor}!peter IAEF: ...!kitty!baylor!peter
jimomura@lsuc.UUCP (Jim Omura) (11/12/85)
In article <417@graffiti.UUCP> peter@graffiti.UUCP (Peter da Silva) writes: >> Well, first of all, as I've said before, a Gimix Level II system >> (which is the one I'd probably use) has memory management, and dedicated >> IO processors for that matter. But memory management isn't that much >> an issue when you consider we're talking about word processing and possibly >> a bit of database stuff -- *not* development work. End users running >> say Databank and Stylograph don't need to worry about memory management. >> It simply wouldn't help even if you had it. > >(1) There are no bug-free programs. >(2) I *AM* talking about development work. If all you're doing is running >canned software you might as well stick with CP/M... it's a lot cheaper than >either OS/9 or UNIX. >-- >Name: Peter da Silva >Graphic: `-_-' >UUCP: ...!shell!{graffiti,baylor}!peter >IAEF: ...!kitty!baylor!peter Peter my man, you know not of what you speak. I run *both* OS-9 and CP/M (I have *2* CP/M systems) and CP/M is not appreciably cheaper. Even if it were, there are definite advantages to OS-9 memory management (the terminology has been getting sloppy here and I'm in part to blame) which handles memory in 2K or 4K pages and arranges them in variable allocations as needed. Cheers! -- Jim O. -- James Omura, Barrister & Solicitor, Toronto ihnp4!utzoo!lsuc!jimomura Byte Information eXchange: jimomura Compuserve: 72205,541 MTS at WU: GKL6
peter@baylor.UUCP (Peter da Silva) (01/18/86)
> Peter my man, you know not of what you speak. I run *both* OS-9 > and CP/M (I have *2* CP/M systems) and CP/M is not appreciably cheaper. >[Jim Omura] Well, could you please inform me how I can purchase a complete OS/9 system for $500 or less (I already have a terminal, so there are several CP/M systems I can get for under $500). If not, I stand by my guns. If you can, please include an address & other ordering info, because I would be VERY interested in getting such a beast. -- -- Peter da Silva -- UUCP: ...!shell!{baylor,graffiti}!peter; MCI: PDASILVA; CIS: 70216,1076
jimomura@lsuc.UUCP (Jim Omura) (01/22/86)
You want cheap OS-9 (under $500.00)? OK, we got it, but when I tell
you a bit more, you may prefer to spend a bit more--that's life.
1. From the immortal pages of 68 Micro Journal, November '85, p. 52 we
have Sardis Technologies, (604) 225-4485, 2261 E. 11th Ave., Vancouver
B.C., Canada, V5N 1Z7.
The new ST-2900 system --a complete 64K small business or hobbyist
computer ...
Small enough to hold in your hand (Eurocard size: 3.9" * 6.3")
3 board system for versatility
CPU Board -- powerful 6809E processor, 16K or 64K RAM, 1K-32K EPROM,
2 RS-232 serial ports with software programmable baud rates, 16 bit counter/
timer. Run the CPU board all by itself, or plug your own custom board or
our FDC board and/or RAM-512 board into the expansion connector
FDC Board -- double-sided/double-density floppy disk controller
with adjustment free digital data separator and write precompensation,
2 8-bit parallel ports, 2 16-bit counter/timers, prototyping area
RAM-512 Board -- 512K RAM on a 4.15" * 6.3" board. Low power.
Includes RAM-Disk software for Flex or OS-9.
OS-9 Conversion Package lets you use the low cos Radio Shack CoCo
version of OS-9 on our ST-2900 system. Save $131 off Suggested list of
OS-9.
No programming is involved. Supports CoCo OS-9, standard OS-9, and
MIZAR OS-9/68K disk formats. Compatible with PC-XFER to let you read/write/
format MS-DOS disks.
CPU bare board plus EPROM $45.00
FDC bare board $38.00
***CPU + FDC board set assembled and tested $329.00***
RAM-512 board assembled and tested $485.00
OS-9 Conversion package $49.00
(Prices are US funds -- $5.00 shipping and handling, Visa OK)
The Shack's OS-9 is under $60.00 in the US as I recall, so there's
*one* possible basic system.
Another route is to buy a 64K Color Computer and disk controller/drive
set up. These are going fairly cheaply right now. I'd include the price
of the RS-232 card and Multi-Pak interface to the estimate though, because
I don't feel the standard CoCo is adequate without these extras. Also,
wait about a month or so for the new Version 2.00.00 OS-9. Getting the
current Version 1.01.00 would be a waste of your time converting later.
The difference is sufficient to warrant the upgrade.
But Peter, my best advice at this minute is not to buy either of
these. Spend a $800.00 - $900.00 and buy a 68000 Emerald Computers
OS-9 system. There isn't as much software for OS-9 68K, yet, and it
tends to cost a bit more, but I think it would be worth it.
Emerald Computers Inc. (503) 620-6094, Telex 311593, 16515 S.W. 72nd
Avenue, Portland, Oregon, 97224.
ESB-II, 68000 (8 mHz.) w/128K DRAM, 2 serial ports, 24 bit parallel
port, 1 - 24 bit timer, 1 - 16 bit timer, floppy controller for up to 4
5 1/4" drives, card size is 5 1/4" drive footprint.
All this for $595.00 (I expect the OS-9 68K to cost about $250.00
extra, but it's unclear in the add whether it's included or not).
This board will take up to 1 meg. DRAM and up to 128K ROM. It
can be chip upgraded to up to 12 mHz. and a 68010.
OS-9 68K is significantly better. Everyone who has used it has said
so. I won't go into why.
Anyway, there you have it. Cheap OS-9. The choice is yours.
Cheers! -- Jim O.
--
James Omura, Barrister & Solicitor, Toronto
ihnp4!utzoo!lsuc!jimomura
Byte Information eXchange: jimomura
(416) 652-3880doug@terak.UUCP (Doug Pardee) (01/22/86)
> > and CP/M (I have *2* CP/M systems) and CP/M is not appreciably cheaper. > > Well, could you please inform me how I can purchase a complete OS/9 system > for $500 or less... Radio Shack. But what does this have to do with 68000's? Wouldn't it be more appropriate for net.micro.cpm, net.micro.6809, and/or mod.os.os9? -- Doug Pardee -- CalComp -- {hardy,savax,seismo,decvax,ihnp4}!terak!doug