Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Where did System 1022 go?

495 views
Skip to first unread message

1022 guy

unread,
Jan 10, 2010, 11:32:35 PM1/10/10
to
I've been wondering who might have the sources?

System 1022 was the leading DBMS for 10/20's. It was developed at Software
House. SH was bought by its largest customer (CompuServe) and became
Compuserve Data Technologies. CDT was sold to CCA (Computer Corp of Amer).
It seems to me they may have only been interested in 1032 (for Vax's). I've
tried at CCA, but haven't gotten a response if they have 1022.
Since Compuserve's Consumer Info Service (CIS) was heavily reliant on 1022,
maybe CS kept 1022 when it sold CDT to CCA. I see now on Wikipedia that CIS
continued on CS till last year.

CS was as far as I know the biggest latest user of 10 hardware (System
Concepts clones), Does anyone know how late they ran them (with heavily
modifed TOPS-10)? Does anyone have a contact at CS that might be able to
lead me to where to where 1022 went? I'd like to revive it and make
available for the community. Thanks...

Pat Farrell

unread,
Jan 11, 2010, 10:00:05 PM1/11/10
to
1022 guy wrote:
> I've been wondering who might have the sources?

Wow, that is an interesting question. While I never officially saw any
sources to 1022, I sure patched it a lot in DDT.


> System 1022 was the leading DBMS for 10/20's. It was developed at
> Software House.

Wasn't it "the only" if you didn't count the CODASYL crocks?
AMS was the second biggest customer of SH, and we supported a vendor
who claimed they were writing a more relational competitor/replacement.
West Coast folks, they got it fairly far along, but we (AMS) lost
interest because our customers really did not care about the Relational
religion,

Accent R was the product name of the competitor

1022 was fast and reliable once we stopped using it for updates.

While the 1022 report writer was highly touchy, and lacked nearly all
modern programming constructs, when what you wanted to do was in its
domain, you could really crank out some reports in no time.

> SH was bought by its largest customer (CompuServe) and
> became Compuserve Data Technologies. CDT was sold to CCA (Computer Corp
> of Amer). It seems to me they may have only been interested in 1032 (for
> Vax's). I've tried at CCA, but haven't gotten a response if they have
> 1022.

Do you have any idea of System 1032 was considered a success by Andy and
Charlie? I know it sold, to a lot of folks who got departmental Vaxen
when their KL was too overloaded and obsolete. I was on the board of the
System 1022/1032 Users Group for years, and I never felt that the Vaxen
folks truly drank the koolaid like a lot of commercial KL houses did.

> Since Compuserve's Consumer Info Service (CIS) was heavily reliant on
> 1022, maybe CS kept 1022 when it sold CDT to CCA. I see now on
> Wikipedia that CIS continued on CS till last year.

I thought that the PC/Modem service was sold to AOL in the late 90s.

Just a little time before Steve Case proved he was one of the smartest
businessmen in the century.

> CS was as far as I know the biggest latest user of 10 hardware (System
> Concepts clones), Does anyone know how late they ran them (with heavily
> modifed TOPS-10)? Does anyone have a contact at CS that might be able
> to lead me to where to where 1022 went? I'd like to revive it and make
> available for the community.

That would be cool.

The guy from Compuserv, Dave Eastburn or Easterlake or ? would have been
the contact between CompuServ and Software House.

I do remember the sale. There was a tax change that was taking effect on
January first, and Andy and Charlie decided to take the money under the
old tax treatment. They contacted many of their major customers,
including AMS. We considered it, but by then, post death of the Jupitor
and all hopes of a continued viable business, we passed. CompuServ
jumped at it.

We were sending about $15,000 a month to Charlie and Andy, I think
CompuServ was sending three or four times as much.

Trivia: at the time, where was no final "E" on CompuServ, but sometime
in the late 80s, they added it.

--
Pat Farrell
http://www.pfarrell.com/

Rich Alderson

unread,
Jan 14, 2010, 6:12:36 PM1/14/10
to
"1022 guy" <s1022...@Use-Author-Supplied-Address.invalid> writes:

> I've been wondering who might have the sources?

> System 1022 was the leading DBMS for 10/20's. It was developed at Software
> House. SH was bought by its largest customer (CompuServe) and became
> Compuserve Data Technologies. CDT was sold to CCA (Computer Corp of Amer).
> It seems to me they may have only been interested in 1032 (for Vax's). I've
> tried at CCA, but haven't gotten a response if they have 1022.

I spoke to them about 15 years ago, at which time they still had the sources on
backup tapes but no hardware. They were amenable to a discussion with XKL of
selling the product to us, for inclusion in the suite of software we would make
available to our customer base. It fell through for political reasons within
XKL.

> Since Compuserve's Consumer Info Service (CIS) was heavily reliant on 1022,
> maybe CS kept 1022 when it sold CDT to CCA. I see now on Wikipedia that CIS
> continued on CS till last year.

They did not hold back 1022 from the sell-off.

> CS was as far as I know the biggest latest user of 10 hardware (System
> Concepts clones), Does anyone know how late they ran them (with heavily
> modifed TOPS-10)? Does anyone have a contact at CS that might be able to
> lead me to where to where 1022 went? I'd like to revive it and make
> available for the community. Thanks...

Good luck. I don't think there's any there, there.

--
Rich Alderson "You get what anybody gets. You get a lifetime."
ne...@alderson.users.panix.com --Death, of the Endless

chuckylo...@gmail.com

unread,
Feb 21, 2019, 7:25:49 PM2/21/19
to
I stumbled across this thread. It's now 2019 - 9 years after the posts! Anyway, I remember System 1022. I was a college work-study student at the Transportation System Center (TSC) under the Department of Transportation at Kendall Square in Cambridge, Mass.

They had 1022 on a PDP-10 and I ran it on ASR 33/34/35(?) - the ASCII version and on the newly introduced TI Silent 700.

I recall one day my boss/mentor was out and I was working in the office alone. The Budget Director came in and had a meeting in 10 minutes. He needed some budget data. He gave me the parameters - contracts for last year between $10,000 and $50,000 (or something similar). Piece of cake with 1022.

I vaguely recall:

Open xxx [filename]
find zzz where aaa >= 10000 and aaa <= 50000
list contract date amount

Something along those lines. SO simple and easy to use. I loved it. Of course, I was also running 96-column cards on an IBM System/3 so timesharing was close to magic.

Great stuff. I just hope this is not lost to history.

Chuck Gage

fishtoprecords

unread,
Feb 21, 2019, 10:44:08 PM2/21/19
to
On Thursday, February 21, 2019 at 7:25:49 PM UTC-5, chuckylo...@gmail.com wrote:
> SO simple and easy to use. I loved it. Of course,

Yes, it was simple and very easy to do several kinds of stuff.
There was a great DECUS presentation by the two Software House guys who were managing 1022 in the early 80s about what you wanted a DBMS to do. I think one was "Marty". It described a 1022-type query language, and compared it against SQL.

The feature with SQL was that it was an academic design, that created a "basis set" of operations that could do anything. The problem was (and still is) that there is no intermediate feedback and it quickly gets arcane.

You could teach a smart secretary how to do useful work with 1022, like your example. Teaching SQL is orders of magnitude more difficult, perhaps even impossible.

chuckylo...@gmail.com

unread,
Feb 22, 2019, 12:09:54 AM2/22/19
to
================
Thanks for the reply! Glad to see someone is around after 9 years!

Chuck Gage


David Todd

unread,
Feb 22, 2019, 6:27:14 AM2/22/19
to
I keep hoping that someone will release 1022 for distribution from the DEC-10/20 archives. It was a great system to get real work done quickly, as you noted, Chuck. It scaled well enough to run the information system operations of a moderate-size university, with secretarial staff able to learn it and generate reports. And it was used by major research organizations (Pfizer, for example) for data collection and analysis.

IBM promoted SQL-based systems, and the abstract elegance of SQL was a major attraction for professional database programmers. But to just get data into the system and be able to examine and report on it in a variety of ways quickly, nothing beat 1022.

While it didn't lead to a world revolution in data-handling, 1022 did have a significant impact in promoting the concept of the value of data and databases. It would be a shame if it were lost in some corporation's vault and not archived for posterity.

1022 guy

unread,
Feb 22, 2019, 8:10:03 AM2/22/19
to
On Sunday, January 10, 2010 at 11:32:35 PM UTC-5, 1022 guy wrote:
I had traced it to a guy at CCA who had the source, but my appeals to his management never got a reply. So sad!

Paul Lambert

unread,
Jun 11, 2019, 5:11:27 PM6/11/19
to
Hi - I was an early and long-time employee at CompuServe. The original name was "compu-serv network services, inc" (no caps), and was changed to "CompuServe Incorporated" in the early 80s, prior to our being acquired by H&R Block

pl2...@gmail.com

unread,
Jun 11, 2019, 5:14:09 PM6/11/19
to
On Sunday, January 10, 2010 at 11:32:35 PM UTC-5, 1022 guy wrote:
The last Systems Concepts clone was run well after the AOL acquisition of CompuServe Information Services from Worldcom (WCOM bought the whole company, then a day or so later sold CIS to AOL, getting AOL's Advanced Network Services in return, which WCOM merged into CompuServe Network Services).

Paul Lambert

unread,
Jun 11, 2019, 5:26:47 PM6/11/19
to
btw - prior to its relationship with System House, CompuServe had put substantial resources into developing its own relational database system, which we called "Info/Plus." The schema and command language was much more sophisticated than the 1022 equivalents, and a few internal applications were built in it. I don't recall it every being released to customers.

Info/Plus was allowed to die because CompuServe began pursuing federal government contracts that at the time were being hosted at Online Systems, another PDP-10 based timesharing company in Pittsburgh. The apps were already written in Fortran with 1022 as the database, so the easiest way to win the contracts and ease the conversion was for us to use 1022 as well.

No sense supporting two database systems at that point, so we cancelled the Info/Plus project.

daeas...@gmail.com

unread,
Jun 11, 2019, 5:35:00 PM6/11/19
to
On Sunday, January 10, 2010 at 11:32:35 PM UTC-5, 1022 guy wrote:
The last SC-40's, built by CompuServe under license from System Concepts, were retired in January of this year (2019) after running nearly non-stop for many years in an AOL computer room in Dulles, where they were moved when the CompuServe headquarters complex in Upper Arlington, OH, was sold. Verizon acquired AOL assets a few years ago, which included those CompuServe systems. And, yes, they were still running 1022 up until the day they were decommissioned. I'm not sure of the last time I ran 1022 on those systems, but it was probably sometime in 2018.

1022 could do things we only dream of doing in ACCESS and other "modern" database systems :>)

fishtoprecords

unread,
Jun 11, 2019, 6:34:44 PM6/11/19
to
On Tuesday, June 11, 2019 at 5:26:47 PM UTC-4, Paul Lambert wrote:
> btw - prior to its relationship with System House, CompuServe had
> put substantial resources into developing its own relational database
> system, which we called "Info/Plus." The schema and command language
> was much more sophisticated than the 1022 equivalents, and a few
> internal applications were built in it. I don't recall it every
> being released to customers.

AMS also invested in an alternative to System 1022. It was a pure relational, with SQL query language being developed out of The Valley. I think it was called Accent R (with the R standing for Relational). The developers got fairly far, but we had a large customer base using System 1022, and they did not want to switch. 1022 was much simpler to use, even if the query language was not as technically advanced. Of course, the death of Jupiter made it all moot.

Paul Lambert

unread,
Jun 12, 2019, 8:24:21 AM6/12/19
to
The Verizon acquisition of AOL actually reunited the band, bringing both the Network Services and CIS divisions of CompuServe under the same umbrella. Detail of the 1998 deal not many know: because both the CIS and Network Services businesses depended on the PDP-10 technology (the PDP-10s performed AAA and DNS functions for the network) the deal was structured so that AOL owned the PDP-10 hardware and WCOM owned the operating system, guaranteeing that the two company had to work together going forward.

David Todd

unread,
Jun 12, 2019, 11:50:50 AM6/12/19
to
Does anyone have a contact at AOL who might be able to retrieve 1022 and add it to the archives? It would be a shame to lose that system to history.

fishtoprecords

unread,
Jun 12, 2019, 3:34:13 PM6/12/19
to
On Wednesday, June 12, 2019 at 11:50:50 AM UTC-4, David Todd wrote:
> Does anyone have a contact at AOL who might be able to retrieve 1022
> and add it to the archives? It would be a shame to lose that
> system to history.

I don't have a current one. About a decade ago, I was talking to Dick Ouilette (sp?) who was at CompuServ in the early 80s, and was a board member of the System 1022/1032 User Group. He was heavily involved with 1022 on their 36-bit systems. He said they still (circa 2008) had a KL out in a data center room at the AOL HQ building in Dulles/Ashburn VA, specifically to run 1022 to help in patent legal matters.

I thought it was really cool that a KL was still turning power into heat.

I did not get the feeling that they were interested in releasing the software, but that topic did not come up. I do remember discussions about patent trolls....

Paul Rubin

unread,
Jun 12, 2019, 3:43:24 PM6/12/19
to
fishtoprecords <pat2...@gmail.com> writes:
> I thought it was really cool that a KL was still turning power into heat.

Anyone know if they still run any -10's? Someone mentioned that they
powered off the last physical ones quite recently, but I wonder if
that's only because emulators are so much more efficient now. Even a
decade ago I'm pretty sure that x86-based emulators were beating KL's.

daeas...@gmail.com

unread,
Jun 12, 2019, 3:56:32 PM6/12/19
to
Actually, the last machines running 1022 at AOL were SC-40 clones built by CompuServe which were clones DEC machines only much faster. The last two machines were shipped to the Living Computer Museum in Seattle with hopes of getting them up and running once again. Earlier, when AOL decommissioned a number of other SC's, some were purchased by 3rd party collectors. The Computer History Museum in California also has a PDP-10 running standard TOPS-10 (I think).

Pat, those machines being used for patent litigation defense are the same ones I mentioned here. Sandy Trevor and I (aka, Nuvocom Incorporated) were under contract with AOL as custodians of the CompuServe Archive, including those systems, and help coordinate the transfer of those assets to the LCM.

daeas...@gmail.com

unread,
Jun 12, 2019, 3:59:03 PM6/12/19
to
Not sure who Dick Ouilette was -- in the early 80's I was still heading up the Database and Graphics product teams at CompuServe and was on the board of the SH Users Group. Around that time is when I handed it off to Bob Root and moved full-time into the consumer information service.

chathamwo...@gmail.com

unread,
Jun 14, 2019, 12:01:23 AM6/14/19
to
On Wednesday, June 12, 2019 at 1:56:32 PM UTC-6, daeas...@gmail.com wrote:

> Pat, those machines being used for patent litigation defense are the same ones I mentioned here. Sandy Trevor and I (aka, Nuvocom Incorporated) were under contract with AOL as custodians of the CompuServe Archive, including those systems, and help coordinate the transfer of those assets to the LCM.

David,

Do you know, or could you suggest how to find out, who actually *owns* System 1022 these days. Is it AOL, and by parentage, thus Verizon. Or did AOL transfer it to Time Warner or someone else over the years?

I think there are two issues: does anyone actually have source code for it that could recompiled, and who actually owns it and could make it public, legally?

This probably really is a rabbit hole, but it might be worth contacting the CEO of Verizon Media Group, K. Guru Gowrappan, of which AOL is a group member, if AOL is in fact the owner.

David

David Todd

unread,
Jun 14, 2019, 12:03:50 AM6/14/19
to
On Thursday, June 13, 2019 at 10:01:23 PM UTC-6, chathamwo...@gmail.com wrote:
> On Wednesday, June 12, 2019 at 1:56:32 PM UTC-6, daeas...@gmail.com wrote:

Sorry, posted that from an association account. Sent by hdtodd@gmail, David Todd.

boblel...@gmail.com

unread,
Aug 10, 2019, 8:49:55 PM8/10/19
to
On Thursday, February 21, 2019 at 10:44:08 PM UTC-5, fishtoprecords wrote:
I remember P Duff at SOftware House, we used 1022 at Lehman Brothers in NYC - all of their fixed income global indexes started with this DBMS, it was so great to use!

BobL

Paul Galbraith

unread,
Jan 6, 2020, 10:51:03 PM1/6/20
to
On Wednesday, 12 June 2019 15:56:32 UTC-4, daeas...@gmail.com wrote:
> Actually, the last machines running 1022 at AOL were SC-40 clones built by CompuServe which were clones DEC machines only much faster. The last two machines were shipped to the Living Computer Museum in Seattle with hopes of getting them up and running once again. Earlier, when AOL decommissioned a number of other SC's, some were purchased by 3rd party collectors. The Computer History Museum in California also has a PDP-10 running standard TOPS-10 (I think).
>
> Pat, those machines being used for patent litigation defense are the same ones I mentioned here. Sandy Trevor and I (aka, Nuvocom Incorporated) were under contract with AOL as custodians of the CompuServe Archive, including those systems, and help coordinate the transfer of those assets to the LCM.

Does that mean that older compuserve systems are actually preserved somewhere? Someone's posted a dump of the drives from one of those last SC-40's (the legality no doubt is questionable!), and I doubt this contains the older software that would be of more interest for nostalgia's sake. It sure would be neat to see some of the old compuserve stuff running again, though.

https://archive.org/details/2015-05-compuserve-raw-disks

David Todd

unread,
Jan 7, 2020, 6:42:35 AM1/7/20
to
Does anyone know what the format is of the ".dsk" files stored on archive.org and how they might be read? I haven't yet tried to download and mount any of those .dsk files, but I noted that they're much larger than an RP07 (one is 8GB!), so I have a feeling they're not mountable by my Panda DEC20.

David Todd

unread,
Jan 7, 2020, 9:30:44 AM1/7/20
to
Well, the naive approach of just trying to mount one of the .dsk files as an RP07 didn't work. Adding
devdef dsk3 rh1.1 rp type=rp07 format=dbd9 path=compuserve-st32550n-04050852-584-2074m-acc.dsk
to the .ini file gave the boot-time error messages:
SJ 0: Device error on Channel 1 Drive 1
SJ 0: Drive cannot be used
SJ 0: Unit has bad home block

No surprise there, but I thought I'd save others the effort of trying ... and perhaps someone can suggest a different set of parameters for the devdef. I didn't try running CHECKD on that "drive" -- didn't seem like a good idea since I doubt that the basic format parameters are right to start with.

Ideas?

Paul Galbraith

unread,
Jan 7, 2020, 1:32:12 PM1/7/20
to
This is definitely out of my realm of expertise, but there is a post here about someone else's efforts to untangle this. (It's impressive, really, considering he started with zero knowledge.) I think he concluded they are RP20 disks, if that's means something to you. It also seems he was able to contact the person who posted the disks, so someone dedicated to this may want to reach out and follow up.

https://medium.com/@mpnet/trying-to-make-sense-of-compuserve-server-hard-disk-images-posted-on-archive-org-b1c62ce6012b

Rich Alderson

unread,
Jan 7, 2020, 3:57:32 PM1/7/20
to
David Todd <hdt...@gmail.com> writes:

> On Tuesday, January 7, 2020 at 6:42:35 AM UTC-5, David Todd wrote:

>> Does anyone know what the format is of the ".dsk" files stored on
>> archive.org and how they might be read? I haven't yet tried to download and
>> mount any of those .dsk files, but I noted that they're much larger than an
>> RP07 (one is 8GB!), so I have a feeling they're not mountable by my Panda
>> DEC20.

No, they're not.

> Well, the naive approach of just trying to mount one of the .dsk files as an
> RP07 didn't work. Adding
> devdef dsk3 rh1.1 rp type=rp07 format=dbd9 path=compuserve-st32550n-04050852-584-2074m-acc.dsk
> to the .ini file gave the boot-time error messages:
> SJ 0: Device error on Channel 1 Drive 1
> SJ 0: Drive cannot be used
> SJ 0: Unit has bad home block

> No surprise there, but I thought I'd save others the effort of trying ... and
> perhaps someone can suggest a different set of parameters for the devdef. I
> didn't try running CHECKD on that "drive" -- didn't seem like a good idea
> since I doubt that the basic format parameters are right to start with.

> Ideas?

Yeah, don't try mounting not TOPS-20 disks on a TOPS-20 system.

The Compuserve file system is based on that of Tops-10, which is radically
different from that of TOPS-20. If CHECKD would even touch the file system
(which it shouldn't), it could only destroy it.

I suspect that Tops-10 would have issues with it as well.

A Compuserve disk image needs to be mounted by the Compuserve monitor, and no
other operating system.

--
Rich Alderson ne...@alderson.users.panix.com
Audendum est, et veritas investiganda; quam etiamsi non assequamur,
omnino tamen proprius, quam nunc sumus, ad eam perveniemus.
--Galen

jeff gunter

unread,
Jan 8, 2020, 5:52:14 AM1/8/20
to
My recollection of the originally posted -584 files is that they’re TOPS10, based on 584 byte blocks (not 576) and also are missing blocks and data and so have tons of errors (but still some readable shorter files). The extra bytes include checksum and block# info; this is how I came to believe some missing or corrupt blocks cause errors; istr some images are better than others and don’t recall specifically about ‘acc’. I think the alluded to tape images would be a better way to go but can’t help wondering how welcome decoding accounting files would be in a more modern safety conscious world.

titan

unread,
Jan 9, 2020, 3:31:11 AM1/9/20
to
"compuserve-st32550n-04050852-584-2074m-acc.dsk"

You may not have used enough seagate drives to recognise that st32550n is a seagate part number. The SC-40's used SCSI disk I don't know how they virtualised the odd sector sizes. https://www.seagate.com/support/disc/manuals/scsi/c8930d.pdf

Neither TOPS-20 or TOPS-10 is going to know what to do on the real hardware and I don't know how likely the Compuserve monitor used a compatible file system

jeff gunter

unread,
Jan 9, 2020, 7:45:22 AM1/9/20
to
I recognize the seagate disk ids. We used them too but ours were formatted with 576 bytes while (these) compuserve disk images used 584. Have a few 576 in the basement. Not every brand actually works well with non512 byte sectors.
Not so sure they were virtualized; the disk actually had 584 Byte blocks that actually read as 584 bytes. Non cis monitors don’t know what to make of the disks because the 128 words set aside for each 576 byte block overflow and data gets shoved progressively further along. If you somehow massage the 584 byte records on the -584 files and remove the extra bytes you will indeed get “dbd9” files suitable for klh10 (not sure of other simulator details); they’re just a continuous trail of 576 bytes that fit perfectly into pdp10 blocks. Unfortunately simply removing the bytes in that manner doesn’t solve the problem (for me at least) because there is a repeating pattern of missing blocks with an occasional additional random dropout. The volume ends up with (mostly undetected) data errors (TOPS10 checksumming is a little simplistic) and Rib errors when the missing block is a prime rib. Some Rib damage can be corrected with the spare ribs but missing data is harder to fix. Some intact blocks can be moved to the correct location thanx to the crc which allows identifying the correct block number data. Oops forgot to mention there is also a magic value byte there too somewhere which helps identify blocks.
-jfg
0 new messages