Re: To 'bit' or not to 'bit'...
From: --CELKO-- (jcelko212_at_earthlink.net)
Date: 12/29/04
- Next message: David Gugick: "Re: UDF problem"
- Previous message: Mike Morella: "Re: Capturing database size on a schedule and graphing?"
- In reply to: Anith Sen: "Re: To 'bit' or not to 'bit'..."
- Next in thread: Anith Sen: "Re: To 'bit' or not to 'bit'..."
- Reply: Anith Sen: "Re: To 'bit' or not to 'bit'..."
- Messages sorted by: [ date ] [ thread ]
Date: 29 Dec 2004 08:50:51 -0800
>> Replace the word bit above with any other SQL datatype and frame the
question again. <<
The ANSI X3J programming languages have rules for converting internal
SQL data types to their native data types as part of the Standards.
NULLs are handled by indicators. However, there is no Standard for BIT
in higher level languages that do not support them (remember COBOL? 78%
of the world's programs are written in it and it works at the character
level).
But more than that, there is no standard or reasonable way I can see to
have a BYTE whose third BIT is a NULL. Your choice is to get all of
the X3J languages to agree that BIT and BYTE are both atomic numeric
types with "funny math" operators.
I find the idea that bytes are not made from bits strange. I also find
that doing byte operations with a NULL bit in the BYTE to be really
strange -- What is an exclusive OR on NULLs? What is a two-complement
addition? A ones-complement addition? How did you define shift, carry
and other basic bit level operations?
I will not even get into the old Russian 3-stage computers, early
base-ten hardware and the Burroughs BCD Cobol machines, in which the
hardware had no bits at all.
Again, bits and bytes are too low level to be used in SQL. You are
mixing physical implementation with the logical model.
- Next message: David Gugick: "Re: UDF problem"
- Previous message: Mike Morella: "Re: Capturing database size on a schedule and graphing?"
- In reply to: Anith Sen: "Re: To 'bit' or not to 'bit'..."
- Next in thread: Anith Sen: "Re: To 'bit' or not to 'bit'..."
- Reply: Anith Sen: "Re: To 'bit' or not to 'bit'..."
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|