GTL (Georgia Tech Language) compiler

[John Covert]

GTL was a fascinating extension to B5500 Algol which implemented string processing and some LISP functions.

As those of you who programmed in ALGOL on the B5500 know, String Processing was done in Stream Procedures, which were incredibly dangerous.  At GaTech, the ability to compile a stream procedure was restricted to people who were on the ACL for some special file (I've forgotten which); the compiler would attempt to open that file, and if the open failed, it didn't let you compile stream procedures.

But in GTL, the string processing was safe and also very, very easy.  There was this horrendous program that a bunch of grad students had tried to write that was intended to produce an index to the minutes of the meetings of the Board of Regents.  Once a year, this program would be submitted for a run, and it would usually take the better part of 24 hours with no other jobs allowed in the mix.  And since that was longer than the MTBF of the hardware, it would sometimes take several days to get it to run.

One day I got fed up with this situation, and sat down overnight and rewrote the whole program in GTL (without bothering to look at the Algol version which was a mess).  The next day I got the data deck for the previous year's run and loaded it to disk, and then ran my program under the TSSMCP with 20 other users logged in.  In 45 minutes it was done.  I was a hero.

Georgia Tech definitely gave the compiler (which, of course, was written in ALGOL) to Burroughs.  Do the folks who worked with Unisys to get the software we have here think there is any chance that we might be able to get the source of GTL.  I'm pretty sure that copies of the manuals would still be available from the GaTech library.  GaTech also submitted its extensive patches to everything else to Burroughs, and it would be very cool to get those as well.

Since this is my first post here, I should introduce myself.  I started at GaTech as a freshman in the Fall of 1968, and one evening one of my fraternity brothers went into the computer center to pick up his ICS 151 Introduction to Programming assignment.  It was the first time I had ever seen a real computer, and it was love at first sight.  Eventually the B5500 was transferred from the Rich Electronic Computer Center to the School of Information and Computer Science, and it became basically my personal toy, to keep it running for everyone on campus who used it.  Those were the days.  Seven years of B5500.  Seven instead of four because I spent more time with the B5500 than with the EE coursework I needed to be doing.

Regards/john

[Paul Kimpel]

Thanks for your post and recollections, John. I know that Georgia Tech was a very active B5500 site during the time you mentioned, but we have very little material that we can attribute to them. The only items I can think of off hand are a listing of Gary Kildall's APL interpreter that came through GT, and GT's user manual for it. 

We also have a large number of scanned listings on bitsavers.org, most of which appear to have come from UC Santa Cruz, and it's possible some of the GT patches are in those listings.

I'd be very interesting in learning more about GTL, and even more interested in getting a copy and trying to make it run again.

It's pretty clear by now that Burroughs/Unisys itself saved NOTHING from that time, unless it's in the archives they donated to the Charles Babbage Institute at U Minnesota. The only reason that we are able to run B5500 software today is that Sid McHarg had three Burroughs release tapes from 1971 that sat on his bookshelf for 40 years, miraculously he found someone with a working 7-track drive, and even more miraculously, the tapes proved to be flawless. All the other software we have is in the form of listings, which must be transcribed -- an ugly, error-prone process.

Listings are better than nothing, however, and we'll take whatever we can get.

On Friday, March 27, 2015 at 7:25:45 PM UTC-7, John Covert wrote:

GTL was a fascinating extension to B5500 Algol which implemented string processing and some LISP functions.

<snip> 

Georgia Tech definitely gave the compiler (which, of course, was written in ALGOL) to Burroughs.  Do the folks who worked with Unisys to get the software we have here think there is any chance that we might be able to get the source of GTL.  I'm pretty sure that copies of the manuals would still be available from the GaTech library.  GaTech also submitted its extensive patches to everything else to Burroughs, and it would be very cool to get those as well.

<snip>

[Paul Kimpel]

I forgot to mention that in poking around today, I found this album of early-computing pictures from Ga Tech, including a good shot of the B5500 console and tape drives.

https://www.flickr.com/photos/ccgatech/6918379668/in/album-72157629420584004/

[Dick Hamlet]

Looks like a KSR35 teletype SPO!
_Dick

[Larry Futrell]

Hi John, It's been a long time ...

 

And Paul, I got your request for the Ga. Tech Chess Program via linkedin, but I am not allowed to send messages to other people there unless I pay to upgrade my account.

 

I think I actually do still have the source for the GTCP, but only as an old deck of cards and a printed listing.  But it was written using some of the GTL extensions so one would need the GTL compiler in order to compile and run it.  I have always wondered how that program would do if I took the time to convert it to C and ran it on a current computer.

 

And while we are talking about GTCP, I want to give credit for that program to Mike Day, a good friend and student at Emory at the time, who designed and wrote a chess program in Fortran and allowed me to use it as the starting point for GTCP.  Back in those days I knew nothing about giving credit for software authorship, and his name should have appeared alongside mine as an author of the program.

 

Larry Futrell

[Paul Kimpel]

Larry,

Thanks for replying to my LinkedIn connection request -- that was the only way I could find to contact you, and you can now ignore it if you'd like.

On Saturday, March 28, 2015 at 1:04:04 PM UTC-7, Larry Futrell wrote:

<snip>

And Paul, I got your request for the Ga. Tech Chess Program via linkedin, but I am not allowed to send messages to other people there unless I pay to upgrade my account.

I completely understand about not wanting to upgrade the account -- I'm not interested in that, either. It's actually better that you responded on the forum.

I think I actually do still have the source for the GTCP, but only as an old deck of cards and a printed listing.  But it was written using some of the GTL extensions so one would need the GTL compiler in order to compile and run it.  I have always wondered how that program would do if I took the time to convert it to C and ran it on a current computer.

I'd still be interesting in seeing the program, even if it's only in the form of a listing. If you have no objection, I'd like to add it to our collection for possible future transcription. I'd also suggest that either the listing or a good scan of it should eventually be forwarded to Al Kossow at bitsavers.org.

Who knows? Some day the source for GTL may surface and we'll be able to resurrect it and run your program. Stranger and more unlikely things have already happened on this project.

I'll contact you privately to see if we can arrange a way for me to get a copy of the program.

<snip rest>

[Nigel Williams]

On Sunday, March 29, 2015 at 3:44:01 AM UTC+11, Paul Kimpel wrote:

I'd be very interesting in learning more about GTL, and even more interested in getting a copy and trying to make it run again.

There is a good chance that Al Kossow at Bitsavers might have the tape with some (or all) of GTL, checkout the listing of the CUBE tape:

http://bitsavers.trailing-edge.com/pdf/burroughs/B5000_5500_5700/listing/CUBE_Library_Listing.pdf

If you look under the INST column (for institution) there are several entries labeled GA TECH that also include GTL in the title, partially transcribed here for future reference (excerpted line of text deliberately incomplete):

GTL    /L200015 GA TECH LANGUAGE(GTL), SEE GTL/L200016 FOR DOCUMENTATION. JAN 72
GTL    /L200016 GTL DOCUMENTATION. JAN 72
GTL    /L200017 ALGOL/ERRORS, GTL ERROR MSG FILE JAN 72
ERRIN  /Y200007 ERROR/INPUT, DATA FOR ERRSRC/Z100011 JAN 72
ERRSRC /Z100011 GTL ERROR MSG FILE GENERATOR. JAN 72


 

[John Covert]

On Saturday, March 28, 2015 at 6:53:01 PM UTC-4, Nigel Williams wrote:

There is a good chance that Al Kossow at Bitsavers might have the tape with some (or all) of GTL, checkout the listing of the CUBE tape:

That's very promising. Thanks, I've left a voicemail message for Al Kossow.

Another possible source: I just got off the phone with Bud Pass who says that we might be able to contact Martin Alexander, if we could find him, possibly through the GaTech Alumni Association.  He was the author of GTL, and very well might have saved a listing or even a card deck.

 

[John Covert]

We had the Smith-Corona SPO, but since it would jam all the time, we also had an ASR33 connected to a Datacomm port, and everytime we would H/L, we would immediately enable the Remote SPO and disable the main SPO to keep it from breaking yet again:

BS1/0

USSPO

We could still enter SPO commands on the Smith-Corona, especially when output to eh ASR33 would get backed up.

Eventually, we put an Inktronic 1200 bps printer on Datacomm port 1/0, and would still type the same two commands right after any H/L, and do all input on the Smith-Corona.  Since there was never any output to the Smith-Corona during normal operations, from anywhere in the room, if you heard the Smith-Corona start typing, you knew that what was on it was a crash message.

[Nigel Williams]

On Tue, Mar 31, 2015 at 8:08 AM, John Covert <cov...@covert.org> wrote:
> That's very promising. Thanks, I've left a voicemail message for Al Kossow.

Don't be disappointed if you don't get a reply from Al, he is
generally very busy and focused and often flooded (overwhelmed) with
good-intentioned inquiries and unable to respond to them all. In an
ideal universe CHM or the software preservation community would hire
Al a secretary.

> Another possible source: I just got off the phone with Bud Pass who says
> that we might be able to contact Martin Alexander, if we could find him,
> possibly through the GaTech Alumni Association. He was the author of GTL,
> and very well might have saved a listing or even a card deck.

Regarding other sources, for those that have contacts from that era it
would be worth putting the word out for anyone who kept any CUBE
tapes. We know at least one person in Oregon who can read 7-track
tapes.

[jhha...@earthlink.net]

I took care of the machine at UCSC from the time it arrived until the scrap dealer
came for it.  I'm pretty sure I gave the tapes all to the Computer History Museum
along with some other documentation.  What didn't go to CHM went to Cal Poly
Pomona, where they were running a B5500 after the one at UCSC shut down.

We had GTL but nobody ever seemed interested in using it - we mostly ran
the (XALGOL?) that was supposed to be more or less compatible with
B6500 ALGOL and provided string handling in the way that machine did.

I'm not a programmer, but I did write a stream procedure in ALGOL once, for
a professor who had a tape from a CDC6600 and needed the data translated
into something he could use.

[John Covert]

I have discovered that I have a copy of the GTL manual, dated December 1971 and referencing Mark XIII.

Now if only we could find the source code.

I see that it is listed as GTL/L200015 on the Cube Tape listing at http://bitsavers.trailing-edge.com/pdf/burroughs/B5000_5500_5700/listing/CUBE_Library_Listing.pdf

[Nigel Williams]

Good find John! would you be able to get the manual scanned so we can
have a reference manual once we get the CUBE tape recovered? it is
only a matter of time before the CUBE tape from CHM is processed and
we will all be able to examine the GTL compiler and try it out.

[John Covert]

It's loose bound and in good enough shape that it should go through an automatic scanner.  I'll see if I can get access to one that will scan and maybe even index the PDF in the process.

[Nigel Williams]

On Monday, April 11, 2016 at 6:45:14 PM UTC+10, John Covert wrote:

It's loose bound and in good enough shape that it should go through an automatic scanner.  I'll see if I can get access to one that will scan and maybe even index the PDF in the process.

Hi John, just wondered if you had any success getting the GTL manual scanned? we can offer to help if you find that useful; several of us have ready access to fast scanners, so the original manual can be returned to you quickly.

 

[Paul Kimpel]

The CUBE tapes from 1972 are in the process of being recovered, and it appears that GTL is present on one of the tapes. It would be wonderful if we could get a copy of the language document. I hope to be able to post the source code and tape images in the next few weeks.

[Al Kossow]

we also appear to have it
http://www.computerhistory.org/collections/catalog/102660328
http://www.computerhistory.org/collections/catalog/102727201

[Al Kossow]

On 5/15/18 4:54 PM, Al Kossow wrote:
> we also appear to have it

done..

http://bitsavers.org/pdf/georgiaTech/GTL_Programmers_Reference_Manual_for_the_Burroughs_B_5500_Aug1974.pdf

[Paul Kimpel]

Fantastic -- thanks very much, Al.

Paul Kimpel
San Diego, California