walter@mwtech.UUCP (Walter Mecky) (12/05/89)
In the xenix-group there are some more or less hot debates to SCO in general and SCO UNIX 5.3 in special. I'll have to decide: Should I buy 386/ix or SCO UNIX 5.3 ? My question to all of you who have experience with the SCO UNIX and/or ISC 386/ix: 1. Does SCO UNIX have so many bugs as noted here ? 2. Is there an online manual in SCO UNIX 5.3 ? 3. How is the support from SCO ? 4. How do SCO UNIX and ISC 386/ix compare (documentation, support, bugs, device drivers, performance especially to disk IO) ? 5. What about software: X11, TCP/IP, VP/IX for *SCO UNIX* ? Thanks for any reply. Note: please email, I'll summarize.
mikes@NCoast.ORG (Mike Squires) (12/08/89)
In article <522@mwtech.UUCP> walter@mwtech.UUCP (Walter Mecky) writes: >In the xenix-group there are some more or less hot debates to SCO in >general and SCO UNIX 5.3 in special. I'll have to decide: > > Should I buy 386/ix or SCO UNIX 5.3 ? > >My question to all of you who have experience with the SCO UNIX and/or >ISC 386/ix: .stuff deleted. >3. How is the support from SCO ? It is at least there, and they seem to come up with workarounds for problems that cannot be fixed quickly. All of the important updates are available from their dial-up uucp site. My experience with other companies has been much worse. It's hard to compare disk access speed at this point - at least SCO UNIX will be (hopefully) changing in terms of its SCSI support. SCO and several other companies have been working on speeding up SCSI support for both XENIX and UNIX, and these changes will eventually wind up in both (I assume). One product that SCO has that makes them hard to ignore is the product that supports multiple CPU's in a single machine, such as the Compaq Systempro.
ken@metaware.metaware.com (ken) (12/09/89)
In article <522@mwtech.UUCP> walter@mwtech.UUCP (Walter Mecky) writes: >In the xenix-group there are some more or less hot debates to SCO in >general and SCO UNIX 5.3 in special. I'll have to decide: > > Should I buy 386/ix or SCO UNIX 5.3 ? > >My question to all of you who have experience with the SCO UNIX and/or >ISC 386/ix: >1. Does SCO UNIX have so many bugs as noted here ? I would have to say no. I found a few bugs that were very fatal for me. The floating point emulation did not work at all. Like AT&T, they give you 2 emulators with the Microsoft as the default one. The MS emulator is faster but less accurate and is missing support for many instructions. The AT&T emulator is completely broken, it fail immeadiately. The other fatal problem were the debuggers. I could not get code view to work. It only understands x.out files and if you switched virtual screens while running it, it would lock up that virtual console. Adb also had some bugs in it. I would like to mention that I think there installation process really stinks. >2. Is there an online manual in SCO UNIX 5.3 ? Yes. >3. How is the support from SCO ? I got good support but I am on some special developer list and thet are very close to me. At one point, The head of developer relations dopped of some fixes on the way home! >4. How do SCO UNIX and ISC 386/ix compare (documentation, support, bugs, > device drivers, performance especially to disk IO) ? I wish I had a recent ISC so I could find out myself. >5. What about software: X11, TCP/IP, VP/IX for *SCO UNIX* ? -Ken
caf@omen.UUCP (Chuck Forsberg) (12/09/89)
In article <1989Dec8.045152.28169@NCoast.ORG> mikes@ncoast.ORG (Mike Squires) writes:
:It's hard to compare disk access speed at this point - at least SCO UNIX
:will be (hopefully) changing in terms of its SCSI support.
A few numbers measured on a 33 mHz Micronics motherboard, CDC Wren V,
and Adaptec 1542A:
-rw-rw-rw- 1 caf omen 1141952 Dec 09 05:12 /tmp/foobaz
time /bin/wc /tmp/foobaz
33235 1818311141952 /tmp/foobaz
real 0m2.02s
user 0m1.12s
sys 0m0.52s
It took about 3.4 seconds to create foobaz with a cp command
from another file, and a sync immediately thereafter took
about .7 seconds.
time cp /tmp/foobaz /dev/null
real 0m1.54s
user 0m0.00s
sys 0m0.48s
The 500 kbytes/sec read rate for file system reads (not raw
disk!) seems pretty consistient every time I've checked it.
The buffer cache is 500k. You may need to have the Wren's
read-ahead buffer turned on with a special program available
from Adaptec to hit these top speeds.
Xenix was getting about 36000 bytes per second on similar
reads, but that was before I turned on the Wren's read-ahead.
I don't know how good the SCSI tape support is in SCO Unix,
but at least one SCSI disk (mine) is quite well supported.
Chuck Forsberg WA7KGX ...!tektronix!reed!omen!caf
Author of YMODEM, ZMODEM, Professional-YAM, ZCOMM, and DSZ
Omen Technology Inc "The High Reliability Software"
17505-V NW Sauvie IS RD Portland OR 97231 503-621-3406
TeleGodzilla:621-3746 FAX:621-3735 CIS:70007,2304 Genie:CAFwalter@mwtech.UUCP (Walter Mecky) (12/10/89)
Here are the emails I received for my posting to SCO UNIX experiences: From: mikros!gargoyle.uchicago.edu!tabbs!aris (Aris Stathakis) > 5. What about software: X11, TCP/IP, VP/IX for *SCO UNIX* ? Open Desktop should be released by SCO soon now. This will turn your 386 box into a (sort-of) Sun workstation type machine. Open Desktop is a package that includes the following: SCO UNIX TCP/IP Dos Merge (Run dos stuff under UNIX) X-Windows NFS (I think..) and a whole bunch of other stuff. All this is included in ONE package, and will cost only MARGINALLY more Than SCO XENIX. Really worth waiting for i think.. ------------------------ From: mikros!txsil.lonestar.org!bobh (Bob Hallissy) >2. Is there an online manual in SCO UNIX 5.3 ? Yes. But nroff is an optional product. All manual pages from SCO are already nroff'd & compressed. Other than no nroff in the base product, the manuals are really good except for one thing: they use the xenix section naming conventions, eg commands are section C, not section 1. >4. How do SCO UNIX and ISC 386/ix compare (documentation, support, bugs, > device drivers, performance especially to disk IO) ? Find a copy of _Unix_Review_ magazine from August 1989 -- there is a pretty in-depth comparison by Tom Yager starting on pg. 83. His "report card" results are: ISC SCO Installation B- B- Documentation C- A+ System Admin A C Performance A B+ Dev't Sys B A X Windows B+ B- Support B B+ Feature Set B A >5. What about software: X11, TCP/IP, VP/IX for *SCO UNIX* ? All are available now. VP/IX has been there for a while, X11 and TCP/IP were just (this week?) released. I saw them demo'd last night at an open-house from our vendor in Dallas. ------------------------ From: mikros!mergvax!udc!uiucuxc!n4hgf.gatech.edu!wht (Warren Tucker) > Should I buy 386/ix or SCO UNIX 5.3 ? > -----> SCO !!!!!!!! But I'm biased and here's why: My system: Compaq Deskpro 386/20, 4 Mb RAM, 155 Mb ESDI CDC Wren IV, Digiboard 8-port dumb serial card, Microcom 9600 baud modem, USR Courier 2400 baud modem, Wyse 60 (19200 baud), TAPR Amateur Radio TNC (4800 baud), Yaesu FT-980 HF transceiver (4800 baud), Motorola VME147 68030 connected via 4 9600 baud asynch ports [what a peripheral :-)]. > My question to all of you who have experience with the SCO UNIX and/or > ISC 386/ix: > 1. Does SCO UNIX have so many bugs as noted here ? I've only been running it for three weeks, but I've been through 100,000+ processes, heavy IPC and shared memory usage (200 IPCs/second), run about 20 meg through HDB UUCP. I've had three unscheduled shutdowns, all power failures -- it recovered nicely each time (no wierd disk problems with fsck, etc). The uptime has been 99%+. Every one of my XENIX 386 programs ran (as binaries) without flaws, with the exception of kmem readers (they merely had to be recompiled). One simple morse code speaker driver fit into the new kernel JUST by recompiling and gluing in. I got MUSH, smail 2.5 and pathalias going in an hour (no predjudice against MMDF, I just finally grokked a mail system and wanted to keep moving without learning a new Art overnight). The ONLY bug I've found is that the new vi marks a file dirty merely by reading it in IFF modelines are in use. I reported it via the sosco bbs and got a call the next afternoon. They're gonna fix it. > 2. Is there an online manual in SCO UNIX 5.3 ? Yes. So much that I had to clobber my DOS partition (:-) finally) to get it all on there. It is a very complete set of packed 'cat' pages not requiring you to have nroff. > 3. How is the support from SCO ? Haven't needed but 1 item other than what I just mentioned. I got a very good technical response in two days regarding the [non-]necessity for SIGUSR1/SIGUSR2'ing getty when stealing an in/out line (not exactly your front-end, phone-answering, coffee-drinker stuff). > 4. How do SCO UNIX and ISC 386/ix compare (documentation, support, bugs, > device drivers, performance especially to disk IO) ? The SCO installation went without a hitch. The documentation (about 20 pounds of it) is the BEST unix documentation I've ever seen. ^^^^ The only thing that bothers me is my lack of experience with the C2 trusted system security features (SCO value added). I've never been fond of Brown Shirts looking over my shoulder. But, there is a "Die Himmler Die" button (sysadmsh System->Configure->Security->Relax) that worked right away. C2 is nice, but not for my little fortress of secrets :-). As to disk I/O, I'm using the Acer Fast File System (default) and it seems to be as fast as XENIX 386 was. This is with 256k of buffers as opposed to the 307k I had with XENIX (reduced bufs 'cause the UNIX kernel is larger). I can do three makes at one time and still get reasonable response response from vi. The 386/20 CPU runs the Plum Hall benchmarks slightly faster than our Sun SPARCstations on all but the function call figures. I've seen two 9600 baud and one 2400 streaming ZMODEM file transfer sessions keep up with close to max theoretical transfer rates. And this is while watching the transfers with my curses-based siomon program in 0-delay mode running on a Wyse 60 at 19200. The siomon display queue numbers get blurry sometimes, so the update rate is good (read: 4 CPU and async I/O intensive processes getting scheduled very nicely). I do async for a living, and I'm really pleased with the performance I get with even the dumb 8-port card (which the kernel automagically recognizes, from a list of preconfigured cards half a page long!). The bothersome system hang you get sometimes when XENIX sync's the big one and stops the world for a few seconds is not present with UNIX. > 5. What about software: X11, TCP/IP, VP/IX for *SCO UNIX* ? I don't have any of those. Sorry. I've been a very satisfied customer of XENIX since 1986. One thing you get with SCO you don't get with ISC is GREAT Xenix compatibility, which means LOTS of BSD code lays right in. Being a Sun and Pyramid' hacker, I -like- that lots. ------------------------ That's all for now. Thanks to all who answered. Any comments ? Walter Mecky
larry@nstar.UUCP (Larry Snyder) (12/11/89)
In article <525@mwtech.UUCP>, walter@mwtech.UUCP (Walter Mecky) writes: > Open Desktop should be released by SCO soon now. This will turn your 386 > box into a (sort-of) Sun workstation type machine. Open Desktop is a > package that includes the following: Will this "Open Desktop" package replace their current Unix System V products? -- Larry Snyder, Northern Star Communications, Notre Dame, IN uucp: root@nstar -or- ...!iuvax!ndmath!nstar!root
aris@tabbs.UUCP (Aris Stathakis) (12/12/89)
From article <511077@nstar.UUCP>, by larry@nstar.UUCP (Larry Snyder): } In article <525@mwtech.UUCP>, walter@mwtech.UUCP (Walter Mecky) writes: }> Open Desktop should be released by SCO soon now. This will turn your 386 }> box into a (sort-of) Sun workstation type machine. Open Desktop is a }> package that includes the following: } } Will this "Open Desktop" package replace their current Unix System } V products? No, I don't think so - but they are marketing this thing HEAD ON with OS/2. Open desktop runs on TOP of UNIX. Their Xenix products will still be supported, but I have a feeling they'd like to phaze them out. I hear that SCO NFS will only be released for UNIX and not XENIX. This is a bit of a bummer. Hopefully a third party will close this gap. aris
karl@ddsw1.MCS.COM (Karl Denninger) (12/12/89)
In article <436@tabbs.UUCP> aris@tabbs.UUCP (Aris Stathakis) writes: >From article <511077@nstar.UUCP>, by larry@nstar.UUCP (Larry Snyder): >} In article <525@mwtech.UUCP>, walter@mwtech.UUCP (Walter Mecky) writes: >}> Open Desktop should be released by SCO soon now. This will turn your 386 >}> box into a (sort-of) Sun workstation type machine. Open Desktop is a >}> package that includes the following: >} >} Will this "Open Desktop" package replace their current Unix System >} V products? > >No, I don't think so - but they are marketing this thing HEAD ON with >OS/2. Open desktop runs on TOP of UNIX. Their Xenix products will still >be supported, but I have a feeling they'd like to phaze them out. I hear >that SCO NFS will only be released for UNIX and not XENIX. This is a bit >of a bummer. Hopefully a third party will close this gap. No kidding. No NFS is a big bummer. Our Xenix machine NEEDS to be connected to other systems now. We can't do it with SCO Xenix. We aren't going to spend a couple of thousand to >replace< perfectly functional software (Xenix 2.3), especially when there are other options out there that work and are less than half the price. SCO has half of it -- TCP/IP. So where's the NFS? SCO, you're dropping the ball on this one. I know you'd like to have everyone buy SCO UNIX 3.2, but some of us are aghast at the cost of that upgrade. As a dealer we could purchase BRAND NEW a 3.2 license from one of our distributors for the same money you want from us to upgrade our 2.3.2 copy to the new UNIX 3.2 release! This is rediculous! It comes about because your organization does not offer dealer discounts on upgrades -- thus we, just like everyone else, pay full price -- for the copy we will use to demo your software to our customers! What is our choice? Give up? It appears so. Although I have come to know and love your organization this is a real problem for us and (I suspect) many other dealers as well. Your policy of no dealer discounts on upgrades affects our customers when they want to update their software too. In the past we have referred them back to you directly -- because they >know< what the list price is and refuse to pay our markup (wouldn't you, knowing you can get it cheaper from the manufacturer?!) This makes more work for both you and us. Then there is the 2.3 OS version for the 286, which doesn't have a 2.3 compiler release. That is yet another problem, and a serious one for some of our current customers. I've got one right now who is rather miffed that we are unable to provide a 2.3 development system although he can get a 2.3 OS version! -- Karl Denninger (karl@ddsw1.MCS.COM, <well-connected>!ddsw1!karl) Public Access Data Line: [+1 708 566-8911], Voice: [+1 708 566-8910] Macro Computer Solutions, Inc. "Quality Solutions at a Fair Price"
jdm@gssc.UUCP (John David Miller) (12/13/89)
We've had a nightmare and a half getting SCO UNIX installed and working. to
be fair, INTERACTIVE is not much better (about half a nightmare ;-) but
they seem to be improving. ISC uses sysadm (leftover from AT&T 3B2's) but
SCO uses that funky sysadmsh, which is even more "black box," but it doesn't
work: we had installation scripts crash with no explanation leaving no known
state or forwarding address. I don't mind having administrative details
hiden - I actually use sysadm now on ISC boxes - but goddamit they better
work or their cure is worse than the disease. We had to install SCO 4 times
before we had a semi-working system.
On the positive side, our code developed on INTERACTIVE and AT&T UNIX 386
compiled and ran just swell, after specifying the AT&T compiler. Even some
of the binaries worked.
My beef then with the "big three" (AT&T, ISC, SCO) is that the system
administration and kernel configuration is all different. As a software
house trying to support all three platforms with an X Server, this
sucks. At that level they are all incompatable, not even considering the
differences in networking software, keyboard/mouse interfaces, and local
Xlib connections. The 386 will never be a viable UNIX workstation platform
as long as everyone is getting their jollies with this kind of (re-inventing
the wheel) "value-adding." Oh sure, you can run the UNIX of your choice and
hack to your heart's content, but you won't have the abundance of commercial
software available that you will on other UNIX platforms or, sadly, OS/2.
-- jdm
--
...!{tektronix!verdix}!sequent!gssc!jdm John David Miller
(503) 641-2200 Graphic Software Systems
* This space intentionally * 9590 S.W. Gemini Dr.
* left blank. * Beaverton, OR 97005