Cataloguing our first manuals and punched cards at Ingenium

About a month ago, I was back at Ingenium in Ottawa with a crew of helpers (Shout out to Laura, Dan and Piérre!). We catalogued the various manuals that were part of the GE-115 collection, and we scanned about 3,000 cards using an Epson RR-600W scanner. It was a tremendous effort of teamwork to make this happen. Of course, thanks to David Pantalone, the Curator of Science and Medicine at Ingenium for facilitating this access and for his kind encouragement.

Laura and I catalogued about 20 manuals. Most of them were GE manuals, some Honeywell, one Olivetti and some miscellaneous authors. These manuals are a treasure trove documenting the DOS commands, disk organization, Fortran, COBOL, Sort Program Generator, LOGEL 3, Extended APS and several other topics. As nearly as I can determine, these many be the only copies of these manuals in existence after 50 years!! There is so much valuable information insofar as improving the quality of the emulation software. Obviously, there is a great deal more work to be done on this! We are discussing the possibility of digitizing the bulk of the manuals so that they will be available to researchers and the nostalgic.

One huge surprise for me was to discover a textbook from the University of Waterloo called Programming with COBOL / WATBOL. This was the textbook that was used in the first year COBOL course at UW. I opened the cover to discover my name written inside! I must have given the book to Tim Crawford after I finished first year (I did work in Barrie in the fall of 1974 after first year and before resuming studies in January 1975). Tim was always interesting in how they were teaching COBOL and data processing principles. He wrote several textbooks in his lifetime on these topics.

On our next trip, I will spend more time with these manuals, as there are specific things I want to pull out next (DOS commands, program file structure, etc.). On this trip, we were primarily focused on the punched cards.

There are perhaps 8,000 punched cards in the GE-115 collection. We did a quick triage on the card decks, and selected 45 decks containing 3,000 cards from the A list. We set up a production line to scan the cards through the Epson reader. Dan loaded a section of the deck (the Epson will only handle about 50 or 60 cards at a time .. one deck was over 200 cards) into the scanner. I started the scanning process. Laura took the scanned cards and reconstructed the original deck and added an index card with a pre-printed catalogue number and then rebound the deck (I think she did the most work!). Piérreis the official photographer of the team and documented the process. However, he also played a critical role in quality control, and identified a few problems in sequencing which we were able to correct because the error was caught in time.

The card decks were bound in cloth tape, which ages much better than the elastic bands which would have originally held the decks together. Dan would remove the cloth tape before loading the segments into the scanner. Dan and Laura performed the "end of file" ritual of passing the cloth tape to Laura so that the segments stayed together in the original deck. It was a clever and innovative solution. The team also worked together to diagnose a problem were it appeared cards were being skipped. We discovered (thanks to Dan and Piérre) that there was a scanner setting which allowed the scanner to delete "empty" scans. For some deck, people had used the reverse side of white cards, often mark sense cards, since we had to pay for our own cards in those days. Once we fixed the setting, everything worked magnificently.

We scanned the 3,000 cards in about 90 minutes of production time, excluding setup and creating our "procedures". This is much slower than the 100 cards per minute of the CRZ100, or the 1000s of cards per minute of later card readers, but it was quite impressive for our first effort at scanning so many cards.

 When I got home, I got to work converting the scanned images to text files. Since I am a programmer, I wrote a program to facilitate this, and the conversions became easier and easier as I added new features to assist in recognizing the alignment of the cards and the presence of the hole punches. There were programs of all kinds, APS, Fortran, COBOL and SPG, as well as a lot of production decks for class scheduling and report card generation, which the administration of the school used in the 70s. There were also a number of loader files, which contain binary load files (basic machine language). I did some work to reverse engineer these load files into APS programs, and discovered some very interesting things. One of these things was an undocumented operation code, 47. The normal jump sequence in machine code would be 43 x0 yyyy, where x is a hexadecimal condition code and yyyy is the address to jump to. In the load files, I found several examples of 47 F0 yyyy which appeared to do the same thing. I will keep an eye open to see if there are some special features to this jump.

Thanks for reading through this far. There is so much detail, it is hard to be brief! 

Comments

Popular posts from this blog

How it got started

The Student Computer Operator

Fortran compiler update and back to the hunt for information