Initially there was no communications from the DADS processor to the radio terminals or the mainframe(that is where I came in).
Jim Bentley documented and coded several screens that the dispatchers would use; Jim was an RPG/Cobol programmer and the intial screens were coded in Cobol and Fortran. DEC's compilers took up almost all the computer memory to run these two languages and there was little room left over for an application. The best that could be done , with high level computer languages, was generate dummy screens that didn't have an application behind them.
The Demo of All Demos
I witnessed an amazing feat my first week at FedEX. Jim would sit behind the computers and start screen 1, while Jim Moore would start discussing the future of the system and do a demo. After explaining the system a few minutes and talking about the first screen, Jim would give a queue to Jim Bentley, who would Stop the screen 1 application and start Screen 2 application; with synchronized timing...Jim would push a key and the other screen would pop up....jim would assign a dispatch number...and a manual message had been queued to go to an MDI terminal on the radio.
This was an amazing feat, given the limited amount of software,, but it did the job. Jim Moore was able to get out the message that sell the idea that this system would increase connectivity to the couriers and solve the problems in the growing cities.
There were at least 4 major issues that had to addressed to have a functioning DADS system
||Radio Task: A software interface needed to be developed that let the DEC computer talk to the radio terminals. MDI provided a black box called the (CC)CommController; so the DEC needed to talk to the CC
||SNA Task: A software interface was needed to talk to COSMOS; COSMOS ran on and IBM mainframe and IBM had recently upgraded there communications protocol from BISYNC to SNA/SDLC
||The Radios and frequencies needed to handle digitized data and still handle voice communications for the couriers. Also at the particular frequencies utilized, there could be lots of bouncing signals, fade and other issues which could cause poor data throughput (Jim Moore & Richard Dunn concentrated on this item.)
||An application had to be written to go behind the screens that Jim Bentley had developed. He defined the screens and data base elements; a more efficient programming language that would generate code that would fit in the PDP 11 memory was needed.
I had been offered a position at FedEx(by Ancel Hankins) in December of 1979; I wasn't able to start until Feb 1st, 1980 due to an internal ear surgery. I was hired in the Data Engineering department which consisted of Ancel Hankins(Mgr); Terry Cox, Dick Winter); and matrixed to Jim Moore's department.
My initial project was to evaluate the DADS system and figure out how to interface it to COSMOS.
DEC's Failure Story
Jim Barksdale took over Data Systems in 1979. He had several major projects to roll out including DADS, COSMOS, centralized call centers, and COSMOS IIa which was the first scanning application.
Digital Equipment had promised FedEx that it would deliver a communications interface that would allow the DADS computer to communicate with COSMOS. DEC discovered that this interface under development in their company would not work.
That is why FedEx started looking for someone who could write communications software.
In my first meeting with DEC and with Jim Barksdale, FedEx only had a one DEC computer and a vision. They also of course had an empty promise from DEC. Instead of being humble, the DEC salesmen began to push Barksdale to order at least 30 of the systems. If FedEx ordered these systems, then they would give the company a 5% discount. Instead of falling on their swords because of their failure, they proceeded to hard sell. Mr. Barksdale was not too kind to them.
Also at the meeting, they gave me the documentation on the software they had developed and would also provide me the source if it would help. I reviewed the documentation and discovered that they had started to develop a new IBM look alike product...that IBM had cancelled. That was the reason they had stopped its development.
COSMOS SDLC/SNA Interface
I reviewed specs for the COSMOS link and felt it would take about 3 weeks to do. The interface board(DUP-11) was ordered but it would take a while to arrive.
Jim Barksdale asked me if there was anything I needed to speed the development. I told him a week training class in DEC macro assembly language would help. There was a spending freeze, but he authorized two weeks training for both Jim Bentley and myself in Los Angeles.
Upon returning, I developed the files, database and macros so that Jim could continue developing the application interface that the dispatchers would use. DEC provided an operating system called RSX-11M and it was the ideal OSr to process thousands of messages an hour between COSMOS and the MDI Radio terminals.
I wrote the Radio task (MDI) in a few weeks and we could now send messages from the DADS PDP-11 computer to the radio terminals. I was able to add in some diagnostic and operations screens and in talking with the radio engineers there were going to be times of no communications (called NOACK)when the radios were in bad areas(like in between tall buildings) or out of the radio coverage area(about 30+ miles out). I put in variable timers for each courier so when they were out of range, we wouldn't waste radio time trying to communicate with those terminals. At the end of the NOACK timer or if any courier hit any button on the terminal, we would automatically start transmitting to them again.
The COSMOS LINK
The SDLC interface board finally came in, and I could start coding. I had already written the high level software, but the low level driver software would be the most difficult. After a week of despair, I finally asked for a budget for $500 for consulting time, to ask Digital Equipment experts what wasn't working. I finally got a hold of a DECNET programmer and sent him my software. He called back and told me that the documentation that DEC had provided had been printed wrong. I needed to add one blank byte at the beginning of my software. I re-compiled..and it worked.
Bob Higgins who was working on COSMOS in Colorado Springs came into town one day. I showed him what I was doing and that I was going to build a table to convert EBCIDIC to ASCII character and to map this into a 3270 screen. IBM used terminals that used 1920 characters. When applications sent messages, you had to convert it to something you could use for your terminals or printers.
Bob said something like...'well all you have to do is take the sixteen bit address, take the lst 6 bits of each byte, jam them together and you will have a number from 0 to 1920 that corresponds to the position on a screen...."
That was a lot of bits and bytes, but I went back and code this in a few lines of code and it worked, and it saved me hours of work.
At midnite that same day, the PDP-11 DADS computer started communications with COSMOS. It was a happy time.
We already had a minimal station set up in Jim Bentley's office on the 17th floor on Clark Tower. Either Jim or the Memphis dispatcher had been taking printouts and just typing in the dispatches into the comm controller to send to the couriers, just for test.
Jim Barksdale came down the hall with some guests and stepped inside Jim's office. He asked me how it was going, and when did I think I could get the COSMOS interface going. I said, 'Oh, I got that working at midnite yesterday'; I had a datascope monitoring the COSMOS - DADS link. I pointed to it and said, 'it's working pretty well, there is only one problem' He paused and said with concern, 'what is it'; I said, 'the dispatches are being transmitted to the couriers perfectly, but when I go to print them(for just backup), the first line is shifted over one character on the printer....' He had a rolled up set of papers in his hand, which he bopped me on the head with...and said 'don't ever scare me like that again.....and Congratulations, Great Work....he went out and told his guests...the bop on the head kind of hurt....but it was another happy day
We were now ready for a full Memphis test,,, around 10+ vehicles.
We could now automatically receive dispatches from COSMOS & into a file(queue) and let the dispatcher assign them. It would only take a few seconds to assign and transmit.
Testing the system was a crazy time. Jim Bentley and I hung out in in his office for weeks with the dispatcher waiting for something to happen. After completion of the days work, we would work at night to correct any issues, and start again the next day.
It was a good prototype test; and we were given the go ahead to prepare the system for a major city test.
Keyboard, no Keyboard, Keyboard again....
Jim Moore was over the Project, the demo's, interace with MDI, setting up plans for rollouts, and working on optimizing the radio channels for data and voice.
The MDI terminals was based on a Zilog Z80 micro. It had a full alpha numeric keyboard but MDI would customize it for FedEx since we would become their largest customer supposedly. Jim wanted simple operation and the terminal had some fixed buttons for STATUS, EMERGENCY and to alert the dispatchers what he was doing.(at LUNCH, DELivery, PickUp etc).
I heard Jim Moore and Jim Barksdale several times discussing whether to have a keyboard or not. The couriers were not typists. The company didn't want to confuse them with keys they would not use.
It was decided to take off the alpha keyboard and just have a numeric pad on the terminal plus the status keys. With numeric keys, the courier could eventually type in the airbill number, showing they delivered a package or picked one up.
So the keyboards came off, then they went back on again. It was decided it might be nice to have an alpha keyboard so the courier might be able to type the name of who the package was actually delivered to.
so keyboard, no keyboard, keyboard again