Wednesday, December 05, 2012

Color to Grayscale (2):

This morning I weighed 224.6 pounds. (My pretty darned accurate scale reports weight to fifths of a pound.)

Yesterday I spoke about the scary risks of developing software on punch cards. I'm thinking about one specific contract my company undertook. Briefly, our typical system offered the customer a number of seats that operators could sit at to issue commands and watch factory status on a CRT. Our typical system sensed hundreds of inputs and manipulated hundreds of control outputs to help manage a factory process.

This particular customer wanted our standard system, but with a switch: a video switch that would enable each of 16 operators, anywhere in the plant, to see what was on any of the 16 CRTs. This was a color system, and they had specific TV studio hardware in mind to do the video switching.
We developed this system, but at the start of the design process, we agonized over one detail for a whole week. We knew this issue would not be noticeable to the customer, but it would have an enormous affect on our software:

Were we developing a system with 16 CRTs, each displaying informative output, that 16 operators could connect to? Or were we developing a system with 16 operator workstations, each of which could see any CRT display they wanted to see?

This question was important because our pre-existing software assumed that each operator workstation and its CRT had specific capabilities that were designed for each operator. In order to develop our new system, we were going to separate each workstation's data and capabilities from its display capabilities. And in the process, we would have to decide, as Lewis Carroll's Humpty Dumpty said, Who was Master? The operator controls, or the critical displays?

Eventually we decided that our new system would have 16 workstations that could see any CRT. And I think we made the right decision, because those little surprises (in terms of work order changes and unexpected clarifications from the customer) never threw us.

Tomorrow I will talk about one small feature of this project that we paid a subcontractor $20,000 to develop (that's in 1975 dollars); and how upset we were at the price, when the developer delivered it in one day.

No comments: