CELKO (12/28/2012)

---

Surely there's some good SQL programmers out there doing exactly what CELKO says they do not do.

I have been at this for 25+ years now and I disagree. Can you give me a scenario where the physical insertion attempt count can have any meaning in a logical data model? How about the blink rate of the front panel of the disk drive?

Back in the original 16-bit minicomputer days on UNIX in the 1970's, the hardware was slow, people did not understand RDBMS and this was assumed to be a way to speed up access. It kept joins short, etc. All you had to give up was data integrity and portability. But we did not know about those things back then. People thought they would always be on one platform, there were no standards and very little theory. And who would have a Terabyte of data on a small machine??

Now move to the 21-st Century. 64-bit hardware, faster and bigger than all of the mainframes on Earth in 1970, SSD, ANSI/ISO Standards and decades of theory and research.

Confession time: when I learned FORTRAN, we had six letter variable names. I got good at inventing six-letter variable names, so I did not stop doing this when I got better FORTRAN compilers and modern languages. I did not realize it was weird and made maintaining code almost impossible for years. :Whistling: