Funny

[Steve Rickaby]

At 10:27 -0500 31/3/06, Ridder, Fred wrote:

>It's interesting to ponder how fast some of today's applications might run if developers still had the skills, tools, and inclination to write efficient code. Ever-increasing processor power and clock speeds have allowed many programmers to write evermore convoluted and bloated code. It's really refreshing to work with teams of developers doing signal processing and telecommunications software, where processes need to operate in real time and latency is evil.

Interesting that you should mention this. The very first ever 'real' programming job I did, back in '79, was 'real' real-time: a thread of code that controlled and synchronized groups of processing modules that ran at 8, 16, 32 and 64 Hz, in such a way that they all continued to communicate with each other correctly even when processing sequences were switched. Other parts of the system had detailed design specs that ran for many pages: 'my' bit just said 'The bla-bla module will sequence other parts of the system to ensure correct running and change-over of processing in the [various] operating modes'.

To do it, I had literally to learn what was for me then a whole new way of thinking. It took me several months to get my head around the task, and then a couple of weeks to write and test what was in the end about 8k of Assembler. It was the hardest programming job I've done, before or since.

The point of this story? Fortunately, no-one told me how hard a job it was, and I was too young and naive to realize. If they had, I probably would have funked it. A similar situation to that which Tracey Kidder describes in 'The Soul of a New Machine', about how Data General played catch-up with DEC in the '70s to get a 16-bit mini to market in six months: if you want something impossible done, don't let the people doing it know that it's impossible.
-- 
Steve