Re: Impossible Database Design?

From: Bob Badour <bbadour_at_pei.sympatico.ca>
Date: Wed, 17 May 2006 21:29:44 GMT
Message-ID: <cBMag.9011$A26.227530_at_ursa-nb00s0.nbnet.nb.ca>


Marshall wrote:

> mAsterdam wrote:
>

>>Marshall wrote:
>>
>>>Seconded. Why not state a more realistic requirement, such as:
>>>support events up to 1000 years in the future? 1000 is much,
>>>much less than infinite, and so has the advantage of being actually
>>>possible.
>>>
>>>Also remember Scott McNealy's comment: "Most software
>>>has the shelf life of a banana." Your software probably won't
>>>still be in use in even 10 years.
>>
>>Also remember that knowing that people actually built
>>software with that assumption led to the millennium paranoia.

>
>
> It is a fair point. But that was people building systems with less
> than 50 years of headroom. It turns out some systems last
> that long. I am fairly confident that no software built during
> my lifetime will be in use 100,000 years from now; probably
> not even 1000 years from now.
>
>
>>Arbitrary limits where none are logically
>>necessary have quite a track-record (640Kb anyone?).

>
> If you don't put any limit in the dates you can specify, you
> still have the physical limits of the computer, as Bob pointed
> out. I think there are a lot of fine compromises; there are a
> *lot* of distinct values in a 64 bit number.

And there are just as many distinct (ie. non-overlapping) ranges the entire size of a 64 bit number in a 128 bit number. And a 128 bit number   occupies a minute fraction of the memory available in a modern laptop computer while the entire state machine defined by the largest computer ever built is still finite.

I strongly suspect some people come to the newsgroup not to learn but because they are enamoured with their latest creation and think everyone else should be just as enamoured. Received on Wed May 17 2006 - 23:29:44 CEST

Original text of this message