[
Lists Home |
Date Index |
Thread Index
]
Yep. Good enough or not enough is sometimes not discovered
until run time. The cases that do need optimum speed and
absolute optimum size can't be discounted.
Standardization is a help and a straitjacket for reasons
everyone here likely understand. What one hopes a standard
prevents are reinvention, reimplementation, and terrorist patents.
OTOH, where the standard isn't sufficient or is encumbered
in some way, the engineers will find another solution. Just
another day at work. The approach of the XBRC so far looks
good in that it characterizes a solution space without attempting
to legislate a one-size fits all wire pushup.
len
From: Robin Berjon [mailto:robin.berjon@expway.fr]
Bullard, Claude L (Len) wrote:
> Yep. The problem is that the qualities one optimizes for tend
> to be competitive, that is, none make all as fast or as
> small as possible. It's too much like shoe and bra sizes:
> the trendy solution may not be comfortable or long lived
> so the classics keep on being used.
Yeah, but very much unlike shoes or bras in this case we'd really like
to all be wearing the same. The question isn't so much about the fact
that you can't get both the absolute optimal speed and the absolute
optimal size -- that's obvious. If however you include standardization
as a third optimization axis (which in many ways is sensical), you will
find many people agreeing that if it's standard they're happy with
"faster enough" and "smaller enough". And it's not hard to be both
faster and smaller than XML, the hardest part is getting a good feel for
the "enough" bit.
|
