[
Lists Home |
Date Index |
Thread Index
]
On Apr 5, 2004, at 12:00 PM, Bob Foster wrote:
> The question is, do you want to spend your time optimizing away the
> need for a <$100 memory chip?
>
That's one of the most profound questions the XML world faces, IMHO.
On the surface the answer is obvious - NO! It is cheaper to just buy
memory (maybe even for your customers!) than to spend time time/speed
optimizing code or bandwidth. But on the other hand ....
- Memory, bandwidth, and even processor speed are still precious on
mobile devices. Do you want to shut them out of your market?
Obviously that's not a consideration for Eclipse plugins :-)
- Batteries aren't covered by Moore's Law.
http://www.wired.com/wired/archive/12.04/start.html?pg=2 has a nice
rant on this subject.
- More generally, as the Wired piece implies, there's sortof a tragedy
of the commons here -- EVERYONE assumes that bandwidth/memory/power is
inexhaustible, exacerbating the problem in environments where it's a
limiting factor.
I think all this gets back to the advantages/disadvantages of
standardization we've been talking about. XML's text basis, Java's
virtual machine, Eclipse's loosely coupled architecture all have very
distinct advantages that are widely touted. They also have
disadvantages, mainly in the area of performance/resource overhead.
It's important not to oversell the advantages to people who are going
to be hurt by the disadvantages, and it's important to try to
ameliorate the disadvantages in a way that does not negate the real
advantages.
|
