Re: Serializability of Transactions and Automatic (Number) Generators
Date: 30 Nov 2004 04:22:05 GMT
Message-ID: <312ardF34m0dgU2_at_uni-berlin.de>
Clinging to sanity, "Marshall Spight" <mspight_at_dnai.com> mumbled into her beard:
> "Jan Hidders" <jan.hidders_at_REMOVETHIS.pandora.be> wrote in message news:Xxspd.655$nN1.57401_at_phobos.telenet-ops.be...
>>
>> If you think of the generated values as abstract identifers (and
>> that is actually what they are supposed to be) then the semantics
>> of your transaction become non-deterministic functions (aka binary
>> relations), although in a very limited sense because the only thing
>> that is not determined is the exact value of the generated values.
>
> This suggests that there might be some value in a data type for
> abstract identifiers. It is a little weird that we use integers for
> keys so often, when there is nothing otherwise integer-like going
> on.
Indeed.
The thing about integers is that:
- They're discrete, unlike floats
- They provide unambiguous ordering, unlike alphanumeric values. (Between locales, handling of case, and EBCDIC...)
- They make use (pun intended) of every available bit of storage
Even in modern days of tens of GB of RAM, speed of systems is bounded by how much you can fit into (registers|cache), and integer (mod 2^32 or mod 2^64) provide _perfectly_ dense utilization of that space.
-- "cbbrowne","_at_","gmail.com" http://linuxfinances.info/info/multiplexor.html We all live in a yellow subroutine, a yellow subroutine, a yellow subroutine...Received on Tue Nov 30 2004 - 05:22:05 CET