Re: Temporal database - no end date

On Jan 24, 4:26 am, Frank Hamersley <terabitemigh..._at_bigpond.com> wrote:
> Marshall wrote:
>
> >http://en.wikipedia.org/wiki/MODCOMP
>
> > They used to have some of them at Lawrence Berkeley Labs
> > where I worked in the early eighties.
>
> > Oh, hell: just imagine three base-40 digits packed in to a
> > 16 bit quantity. Fits nicely: 40*40*40 = 64000, not much left
> > over at all. Each character consumes five and one third bits.
>
> Interesting - this smells like the sort of system that a interplanetary
> spacecraft designer would feel compelled to build.

Heh. Agreed.

> That said...did this gadget perform direct string/character operations
> on the triplets? I assume that single character strings cost 16 bits to
> store? Was it possible to directly address each of the 3 characters?

It was a long time ago and I was just a kid, but I would not be surprised
to find that all the limitations and problems that you imagine would apply.

> Personally I suspect that Darwin got the better of this gem as evidenced
> by the rather short wiki noting its demise (by adsorption).

Again, I'm pretty sure that's right.

One less-lame possible use of this partial-bit technique is to pack decimal digits into binary digits as an integer representation. Math in general gets slower but converting binary to and from human readable format becomes table-driven and hence very fast. For some narrow set of applications that might be a good tradeoff. I hesitate to mention that however for fear of exciting the everything-is-a-string camp.

Marshall Received on Wed Jan 24 2007 - 17:55:21 CET