Re: Larry Breed and APL\1130 (was Re: Hello Group!)
139 views
Skip to first unread message
David Cortesi
Jul 20, 2011, 10:22:40 AM7/20/11
to ibm...@googlegroups.com
James C Field <james...@cooptel.net> Jul 20 12:55AM +0100
^
I believe that Larry M Breed (q.v.) wrote most or all of the APL\1130
implementation while associated with IBM.
You believe incorrectly. APL\360 was developed at the Phil. Science Center where Ken Iverson worked. However it was *maintained* by a group in the Data Processing Division located in Palo Alto, CA. Incidentally this same group was responsible for maintaining CALL 360 while it existed, and a few years later, did use the APL\360 code base to create VS/APL.
That Palo Alto DP group got the job of creating APL\1130 as a product. They were in close touch with the Philadelphia people but were not managed by them, indeed were in completely different divisions (very significant in IBM's corporate world!). Note that at this time, "close touch" meant postal mail, telephone and telex, as this was in the last few years before IBM had an internal network that allowed email!
I joined this group (my first programming job after several years as a hardware CE) when it had been working for some months. My now-spouse was also a member (so I can call on another witness!) and I remember the names of most of the others. I remember the code being designed and written, from scratch, by our group, and who wrote what.
There is an article linked from the Wikipedia article on Larry Breed, "How We Got to APL\1130"
http://archive.vector.org.uk/art10001190
which details some purported history that is absolutely bizarre to me! Breed claims that the first version of APL\1130 was created by re-targeting an interpreter he wrote for a failed product called Elsie, using an Elsie emulator in the 1130 and gradually replacing its code with 1130 native code.
The article goes on to say, "It wasn’t hard to persuade IBM Marketing to fund development of Version 2, which added multi character identifiers and most of the other APL\360 features." If the group I was part of had started with any code model other than APL\360, I think I would have been told that during the many technical conversations we had. I do not recall any mention of such a starting code base. Of the Philly Science Center people, Adin Falkoff was much more of a presence to us as a code and design guru than Breed, and of course Iverson as a distant near-deity.
There are some technical peculiarities in that article, for example it says of the 1130, "The console keyboard was adapted from a Model 029 key punch, with only one alphabetic case.." That certainly isn't true; the console was based on a Selectric and with the famous APL typeball in place, entered all characters just fine. What machine is Breed recalling that had an 029 console?
I would very much appreciate if anyone here can clarify this article, or Breed's claim to have developed the first APL\1130.
Thanks,
Dave Cortesi
That Palo Alto DP group got the job of creating APL\1130 as a product. They were in close touch with the Philadelphia people but were not managed by them, indeed were in completely different divisions (very significant in IBM's corporate world!). Note that at this time, "close touch" meant postal mail, telephone and telex, as this was in the last few years before IBM had an internal network that allowed email!
I joined this group (my first programming job after several years as a hardware CE) when it had been working for some months. My now-spouse was also a member (so I can call on another witness!) and I remember the names of most of the others. I remember the code being designed and written, from scratch, by our group, and who wrote what.
There is an article linked from the Wikipedia article on Larry Breed, "How We Got to APL\1130"
http://archive.vector.org.uk/art10001190
which details some purported history that is absolutely bizarre to me! Breed claims that the first version of APL\1130 was created by re-targeting an interpreter he wrote for a failed product called Elsie, using an Elsie emulator in the 1130 and gradually replacing its code with 1130 native code.
The article goes on to say, "It wasn’t hard to persuade IBM Marketing to fund development of Version 2, which added multi character identifiers and most of the other APL\360 features." If the group I was part of had started with any code model other than APL\360, I think I would have been told that during the many technical conversations we had. I do not recall any mention of such a starting code base. Of the Philly Science Center people, Adin Falkoff was much more of a presence to us as a code and design guru than Breed, and of course Iverson as a distant near-deity.
There are some technical peculiarities in that article, for example it says of the 1130, "The console keyboard was adapted from a Model 029 key punch, with only one alphabetic case.." That certainly isn't true; the console was based on a Selectric and with the famous APL typeball in place, entered all characters just fine. What machine is Breed recalling that had an 029 console?
I would very much appreciate if anyone here can clarify this article, or Breed's claim to have developed the first APL\1130.
Thanks,
Dave Cortesi
Robert A. Lerche
Jul 20, 2011, 11:36:11 AM7/20/11
to ibm...@googlegroups.com
> There are some technical peculiarities in that article, for example it says
> of the 1130, "The console keyboard was adapted from a Model 029 key punch,
> with only one alphabetic case.." That certainly isn't true; the console was
> based on a Selectric and with the famous APL typeball in place, entered all
> characters just fine. What machine is Breed recalling that had an 029
> console?
> of the 1130, "The console keyboard was adapted from a Model 029 key punch,
> with only one alphabetic case.." That certainly isn't true; the console was
> based on a Selectric and with the famous APL typeball in place, entered all
> characters just fine. What machine is Breed recalling that had an 029
> console?
I cannot speak directly to the origin of the code base but I can
assure you that the console printer and keyboard were two completely
separate devices. The keyboard looked almost identical to an 029 and
generated "card code" (the same bit patterns as the 1442 card
reader). The keyboard was not connected directly to the console
printer -- the CPU had to read from the keyboard and translate to
"rotate/tilt" code to print.
Jürgen Winkelmann
Jul 20, 2011, 4:15:51 PM7/20/11
to ibm...@googlegroups.com
Hi Dave,
I'm really excited that my original post reveals so much diverse information about APL's history!
I've also read the article you mention and of course took everything written there for granted, given the fact that the article is cited in so many places on the web including Catherine's site about creating an APL documentation. As I just came out of Kindergarden age when IBM started marketing the IBM 1130 I don't claim to have any first hand know how of this machine and the software that was available at that time. But from playing around with the IBM 1130 simulator and with APL\1130 in it, I think that there are some facts that at least partly support the "How We Got to APL\1130" article:
- On bitsavers I found the User's Manual for the second release of APL\1130 which came out in 1969: http://bitsavers.org/pdf/ibm/1130/lang/1130-03.3.001_APL_1130_May69.pdf. From what I tested, the APL\1130 that is currently available on the IBM 1130 simulator is exactly that version. The PDF file contains a scan of the original IBM internal "Program Distribution Form" for a Type III program, signed off by Stephen Raucher and mentioning explicitely L.M. Breed as co-author (page 3 of the PDF file). The "Acknowledgements" on page 12 of the PDF file (page 15 of the scanned manual) do also prominently display L.M. Breed and the fact that APL\1130 is based on APL\360.
- The documentation of the IBM 1130 simulator on IBM1130.org contains a detailed map of the character encodings used by the different peripherals and states clearly, that the console keyboard produced 029 card punch codes. This doesn't contradict at all to the fact, that the console printer used a typeball, which could be exchanged by the special APL typeball for APL users. But it confirms, that for APL there where no lowercase characters: The 029 card punch was able to encode 64 characters with it's keyboard. By using the "2" and "3" keys to provide two different "shift" positions APL\1130 was able to get exactly 96 characters out of that keyboard. The keyboard layout for APL\1130 is depicted in the above mentioned User's Manual. I cross checked by testing all possible encodings that APL\1130 has exactly 96 character positions assigned. These produce on the APL typeball uppercase characters, digits, punctuation and special APL characters. None of these 96 codepoints do produce lower case characters, even not on the standard typeball. But "normal" programs, for example FORTRAN running under DM2, can of course print lowercase characters using a standard typeball or printer equipped with an appropriate print chain.
So, for me that's a bit weird... I don't doubt that you remember correctly what at that time was going on in Palo Alto. Could it be possible that an APL\1130 "Program Product" has been developped at that time in parallel to the Type III contributed program we're talking about in the IBM 1130 simulator context? If so, I'd of course be interested to hear more of it...
Best, Juergen
James C Field
Jul 20, 2011, 6:26:06 PM7/20/11
to ibm...@googlegroups.com
On 20/07/2011 3:22 PM, David Cortesi wrote:
> James C Field<james...@cooptel.net> Jul 20 12:55AM +0100
> ^<#1314685aead9dbd0_digest_top>> James C Field<james...@cooptel.net> Jul 20 12:55AM +0100
> fund development of Version 2, which added multi character identifiers and
> most of the other APL\360 features." If the group I was part of had started
> with any code model other than APL\360, I think I would have been told that
> during the many technical conversations we had. I do not recall any mention
> of such a starting code base. Of the Philly Science Center people, Adin
> Falkoff was much more of a presence to us as a code and design guru than
> Breed, and of course Iverson as a distant near-deity.
>
> There are some technical peculiarities in that article, for example it says
> of the 1130, "The console keyboard was adapted from a Model 029 key punch,
> with only one alphabetic case.." That certainly isn't true; the console was
> based on a Selectric and with the famous APL typeball in place, entered all
> characters just fine. What machine is Breed recalling that had an 029
> console?
>
> I would very much appreciate if anyone here can clarify this article, or
> Breed's claim to have developed the first APL\1130.
>
> Thanks,
>
> Dave Cortesi
>
having worked with KEI, LMB, RDM, EBI, RBE and many others in the APL
community.
I will certainly (re)read the cited Vector article now. Larry NEVER
represented himself to me or anyone I know that he "developed" [by this
I mean invented from original thought the product].
He built it from reduced specifications from the great men who invented
it for a particular (tiny) machine.
regards, JCF
Catherine Lathwell
Jul 21, 2011, 8:55:49 AM7/21/11
to ibm...@googlegroups.com
I think the issue is timing... though I'm in the gathering stage of my documentary (not the fact checking, which gets more rigorous later).
I *think* that Larry's 1130 implementation preceded the advent of the Palo Alto development group at IBM, and also, in fact, our collective migration from New York to Philadelphia with the opening of the Philadelphia Scientific Centre.
Thank you for surfacing the controversy! And I could be wrong too, but I have an idea! Let's ask Larry!
Catherine
John Doty
Jul 21, 2011, 4:40:08 PM7/21/11
to ibm...@googlegroups.com
On Jul 20, 2011, at 9:36 AM, Robert A. Lerche wrote:
> I cannot speak directly to the origin of the code base but I can
> assure you that the console printer and keyboard were two completely
> separate devices. The keyboard looked almost identical to an 029 and
> generated "card code" (the same bit patterns as the 1442 card
> reader). The keyboard was not connected directly to the console
> printer -- the CPU had to read from the keyboard and translate to
> "rotate/tilt" code to print.
Yep. One of the peculiarities of the 1130 was the diversity of character encodings. A substantial fraction of the DM2 subroutine library consisted of functions to translate codings for specific peripherals to/from EBCDIC.
John Doty Noqsi Aerospace, Ltd.
http://www.noqsi.com/
j...@noqsi.com
Larry Breed
Jul 22, 2011, 11:45:19 PM7/22/11
to ibm...@googlegroups.com
I don't remember the names James C. Field or David Cortesi,
but if it'll help clear up misunderstandings I'm glad to talk or
email with either of them. Here's comments about the several messages on this topic.
1. I wrote very little of the Elsie interpreter (that was Phil
Abrams) or of APL/1130 version 1 (that was Charles Brenner and Alan
Nemeth.)
2. The discussion topic changed from APL\1130 to APL\360 so abruptly
that I wonder if the writer simply mistook one name for the other.
In any case, Ken's group developed APL\360 at IBM Research in
Yorktown, NY starting 1966. It was distributed as program products
5734-XM1 and 5736-XM6 around 1969-1970. Development continued at the
Philadelphia Scientific Center after Ken's group moved there from Research.
3. Science Research Associates in Chicago contracted with an IBM
software development group in Palo Alto around 1966 to write an APL
for the 1800. (This sounds like the group Cortesi might have been
in.) The 1800 was a larger machine similar to the 1130, and SRA
envisioned an offering for the educational market. Names I recall
from that group were Hans Jeans (mgr), Jon Peterson (mgr or
programmer) and Everett Allen (programmer). The weak linkage between
that group and Ken meant many non- APL characteristics found their
way in. I will guess that the effort and costs were underestimated,
and the project came to a halt. It was revived, unofficially, by
three bright summer students at SRA: Harold Driscoll, Scott
McMurchie, and one whose name I forget. They plunged into the code,
rewrote it and completed it and gave SRA something to show off by the
end of the summer.
4. I don't know what happened later with that code; it had no
connection with APL\1130. Philip Abrams' work in Palo Alto on Elsie
APL overlapped in time, and was funded by SRA. Perhaps someone is
conflating one SRA effort with the other.
5. APL\1130 development, after Yorktown's "overnight" exploit
described in my paper, became contract work at I. P. Sharp Associates
in Toronto. I left IBM for STSC about the same time, so I might
never have heard of other APL\1130 work in IBM's DP division, in Palo
Alto.
6. On the other hand, IBM's Palo Alto Scientific Center ("PASC";
Creasy, Hassitt, Lyon) was actively writing their own APL, which got
internal IBM use as APL/CMS. They probably did read the Yorktown
assembler source code, and may have borrowed some operator execution routines, but their data structures and freespace
management were quite different.
7. The APL program product was maintained and extended in Palo Alto
in (I think it was called) the General Products Division of IBM --
but the core of the group was most of the programmers from the
Philadelphia Scientific Center. I rejoined that group under Alex
Morrow and Jim Brown, following 5 years at STSC.
8. The group to which PASC handed off their code may well have
descended from the group I mentioned in (3) above.
They worked next door to the group described in (7) ; in the late
1970s their target project was called VSPC. It was led by Irv Vago,
and Miriam or Marian Cortesi was one of the managers.
9. The 1130 console terminal used a keypunch keyboard (only one
alpha case), and an 1050 typeball printer.
Now that I've written this much, I see how close and intertwined the
paths were through APL\360, APL\1130, and VSAPL, and I also see how
one misread product name could lead to David Cortesi's disbelief and
indignation. If you find any discrepancies in my recollections --
well, it was a long time ago.
Larry Breed
but if it'll help clear up misunderstandings I'm glad to talk or
email with either of them. Here's comments about the several messages on this topic.
1. I wrote very little of the Elsie interpreter (that was Phil
Abrams) or of APL/1130 version 1 (that was Charles Brenner and Alan
Nemeth.)
2. The discussion topic changed from APL\1130 to APL\360 so abruptly
that I wonder if the writer simply mistook one name for the other.
In any case, Ken's group developed APL\360 at IBM Research in
Yorktown, NY starting 1966. It was distributed as program products
5734-XM1 and 5736-XM6 around 1969-1970. Development continued at the
Philadelphia Scientific Center after Ken's group moved there from Research.
3. Science Research Associates in Chicago contracted with an IBM
software development group in Palo Alto around 1966 to write an APL
for the 1800. (This sounds like the group Cortesi might have been
in.) The 1800 was a larger machine similar to the 1130, and SRA
envisioned an offering for the educational market. Names I recall
from that group were Hans Jeans (mgr), Jon Peterson (mgr or
programmer) and Everett Allen (programmer). The weak linkage between
that group and Ken meant many non- APL characteristics found their
way in. I will guess that the effort and costs were underestimated,
and the project came to a halt. It was revived, unofficially, by
three bright summer students at SRA: Harold Driscoll, Scott
McMurchie, and one whose name I forget. They plunged into the code,
rewrote it and completed it and gave SRA something to show off by the
end of the summer.
4. I don't know what happened later with that code; it had no
connection with APL\1130. Philip Abrams' work in Palo Alto on Elsie
APL overlapped in time, and was funded by SRA. Perhaps someone is
conflating one SRA effort with the other.
5. APL\1130 development, after Yorktown's "overnight" exploit
described in my paper, became contract work at I. P. Sharp Associates
in Toronto. I left IBM for STSC about the same time, so I might
never have heard of other APL\1130 work in IBM's DP division, in Palo
Alto.
6. On the other hand, IBM's Palo Alto Scientific Center ("PASC";
Creasy, Hassitt, Lyon) was actively writing their own APL, which got
internal IBM use as APL/CMS. They probably did read the Yorktown
assembler source code, and may have borrowed some operator execution routines, but their data structures and freespace
management were quite different.
7. The APL program product was maintained and extended in Palo Alto
in (I think it was called) the General Products Division of IBM --
but the core of the group was most of the programmers from the
Philadelphia Scientific Center. I rejoined that group under Alex
Morrow and Jim Brown, following 5 years at STSC.
8. The group to which PASC handed off their code may well have
descended from the group I mentioned in (3) above.
They worked next door to the group described in (7) ; in the late
1970s their target project was called VSPC. It was led by Irv Vago,
and Miriam or Marian Cortesi was one of the managers.
9. The 1130 console terminal used a keypunch keyboard (only one
alpha case), and an 1050 typeball printer.
Now that I've written this much, I see how close and intertwined the
paths were through APL\360, APL\1130, and VSAPL, and I also see how
one misread product name could lead to David Cortesi's disbelief and
indignation. If you find any discrepancies in my recollections --
well, it was a long time ago.
Larry Breed
Bob Flanders
Jul 22, 2011, 11:48:15 PM7/22/11
to ibm...@googlegroups.com
Larry ...
Welcome to the group, and thanks for your comments!
Bob
Catherine Lathwell
Jul 28, 2011, 7:23:23 PM7/28/11
to ibm...@googlegroups.com
When I asked my dad (Richard Lathwell) about his involvement with APL\1130 last week, this is what he replied:
The 1130 wasn't around until after I got to IBM. Charlie Brenner's claim to fame is the implementation of APL\1130. The 1130 console keyboard was pretty lame, and required a bizarre triple shift scheme to be able to type APL. So I came up with the idea of building an interface for a typewriter terminal (IBM 1050 or 2741) using the 1130's paper tape punch interface for output and the paper tape reader interface for input. Joe Vanginderen, a technician at the Research Centre, designed the circuitry and we built a few. I have a copy of the invention disclosure that IBM published. If you found one of those interfaces that would be truly astounding!
Bill Rubenstein
Jul 28, 2011, 7:56:49 PM7/28/11
to ibm...@googlegroups.com
Sort of interesting (at least to me)...
In the early days, IBM's interest in invention disclosures was mainly for the purpose of keeping others from patenting something which they were already considering or using. Plant people kept dated notebooks and turned them in regularly. Important developments were published regularly -- can't remember the name of the publication -- the St Louis public library had a subscription and I'd frequently go there on my lunch hour to read it. Field people didn't have to keep notebooks because, as far as the company was concerned, we didn't develop anything worth patenting.
Anyway, last year we were on a 30 day cruise and one of the passengers we became friendly with (we are Bill and Nancy, as were he and his wife) was a recently-retired IBM patent lawyer. Things have changed there, and now their patents, at least according to him, are a major source of revenue for the company.
Bill
In the early days, IBM's interest in invention disclosures was mainly for the purpose of keeping others from patenting something which they were already considering or using. Plant people kept dated notebooks and turned them in regularly. Important developments were published regularly -- can't remember the name of the publication -- the St Louis public library had a subscription and I'd frequently go there on my lunch hour to read it. Field people didn't have to keep notebooks because, as far as the company was concerned, we didn't develop anything worth patenting.
Anyway, last year we were on a 30 day cruise and one of the passengers we became friendly with (we are Bill and Nancy, as were he and his wife) was a recently-retired IBM patent lawyer. Things have changed there, and now their patents, at least according to him, are a major source of revenue for the company.
Bill
Catherine Lathwell
Jul 29, 2011, 8:24:21 AM7/29/11
to ibm...@googlegroups.com
That is very interesting to me too, Bill. In fact, it explains why my dad kept detailed journals the APL/360 implementation. He's kind of a journal writer anyway, so I never really thought about why.
I have many thoughts on the subject in general, now that you mention it, and you've just tipped me in the patent protection direction. Which, I guess is, of course, obvious. If you were here in Toronto, you'd hear the proverbial pin drop in the room. I better move the conversation to my blog so I don't annoy everyone with a segue. But I hope you kept in touch with Bill!
Jürgen Winkelmann
Oct 31, 2011, 5:14:11 PM10/31/11
to ibm...@googlegroups.com
Hi Catherine,
While reading a bit around the group I just rediscovered your post from July concerning the 2741 attachment:
Am Freitag, 29. Juli 2011 01:23:23 UTC+2 schrieb Catherine Lathwell:
When I asked my dad (Richard Lathwell) about his involvement with APL\1130 last week, this is what he replied:The 1130 wasn't around until after I got to IBM. Charlie Brenner's claim to fame is the implementation of APL\1130. The 1130 console keyboard was pretty lame, and required a bizarre triple shift scheme to be able to type APL. So I came up with the idea of building an interface for a typewriter terminal (IBM 1050 or 2741) using the 1130's paper tape punch interface for output and the paper tape reader interface for input. Joe Vanginderen, a technician at the Research Centre, designed the circuitry and we built a few. I have a copy of the invention disclosure that IBM published. If you found one of those interfaces that would be truly astounding!
I only now realize that the interface your dad invented is exactly the same I've now resurrected on the IBM 1130 simulator by completing Brian's initial 2741 code fragment from 2005. So, if your dad ever wants to use "his" interface again please point him to http://wotho.ethz.ch/APL-1130/2741_aplpreview.zip containing the IBM 1130 simulator preconfigured to run an APL\1130 session on an emulated 2741.
Cheers, Juergen
Reply all
Reply to author
Forward
0 new messages