Re: WWW/Internet 2009: 2nd CFP until 21 September
Date: Mon, 10 Aug 2009 19:35:11 GMT
Walter Mitty wrote:
> Then, one time, I had a student who had worked on the original SAABRE, or
> one like it. He corrected me. He said they had spent months trying to
> figure out how to detect overbooking, until finally the marketing people had
> decided that overbooking was a feature, and not a bug! They had not used
> relational, or even CODASYL databases in their work. And they had rolled
> their own transaction control.
I believe that most of the time a PARS application transaction involved a single physical record. And there was only a handful of record types, some of them pretty big. I heard of one financial version (system was based on a variant called PARS/F) that had some hundreds of fields in one record (without even counting all the repeating group fields). The networks of the 1960's and even the 1970's mostly ran at speeds of 110 baud, later a few of them reached 300 baud. Speed was improved by about 25% by using six-bit character codes. This was improved further by very short message codes which spawned careers for many airline people because there were so many brief mnemonics that it was considered a valuable skill to be able to enter them quickly at terminals, kind of like the keypunch operator I knew who used to correct certain errors in our programs.
At many airlines once the computers eliminated some of the manual entry, these unionized people became the business analysts who were charged with specifying and overseeing the development of the new-fangled database management systems, resulting in a number of remarkable blunders, such as the system feature that promised real-time solutions for the travelling salesman problem, re-arranged to handle flight segments from and to any of the 6,000 or so airport codes in the world. Received on Mon Aug 10 2009 - 14:35:11 CDT