Apparently I need to put out some more of this stuff
I seem to be getting email and a lot of searches and hits on bit masking (or bitmasking), so I guess I'll put out some more on this topic. I think it is an interesting technique and apparently others do as well. I first saw this in action in the RDBMS in the late 90's during the period when Match.com was moved from SFO to Dallas. Don Bishop, who I affectionately refer to as my mentor, is great at the technique and showed me many of the ins-and-outs while we worked together at Match and One And Only. Essentially, bitmasking, whether it be in the database or in a procedural language, can be in one or two forms: 1) bit cramming, and 2) traditional representations of values. For the sake of not giving you a lousy definition of it, go to Wikipedia and see what they have to say.
Give me a week or two and I'll see what I have. If you don’t like the technique, send mail and let me know. If you like it, send mail. I’m not for or against anything in particular until I see how it is used. In a data warehouse with a billion rows, no I probably wouldn’t use any masking. Other implementations maybe, and most of the time probably not. Why do it then, you ask, or waste time “bit fiddling” [1]? ‘Cause it rulz!
Lee
-------------------------------------
Guess I better brush up a tad on it. Not sure if I remember how...
[1] Joe Celko (nd). (Search the web, I can’t remember where he said this).
c829c4ff-0854-4e6d-8025-d1993d646cae|0|.0