[
Lists Home |
Date Index |
Thread Index
]
Actually, indirection is the least of the problems as the system
increases in complexity. It is the identification of shared
semantics and resolving these differences that makes it hard.
Indirection via metadata is the solution. Scaling will break
down anyway. That is why data translation is such a big
business.
My apologies for not having sufficient time for this debate.
I didn't come back from DC to a 'rest'. I came back to a
very deep stack, multiple overlapping tasks, and an impossible
schedule. One has to like chaos. But anytime I feel sorry
for myself, I watch CNN and the see the guys and gals welding
junk to their humvees. My job is easy.
len
On Wed, 15 Dec 2004 09:37:37 +0000, Christian Nentwich
<christian@systemwire.com> wrote:
> Roger,
>
> let's see:
> - The first approach does not scale - it will get too complex when
> you then want to reuse your invoice to build a yet larger model. Too
> much indirection to resolve.
|
