ietf
[Top] [All Lists]

Re: Trees have one root

2002-07-29 05:25:30
On Mon, Jul 29, 2002 at 10:12:20AM +0200, Codogno Maurizio wrote:
From: Brian E Carpenter 
This said, I still think that having multiple roots is a 
Bad Thing, because it gives much more hassles than advantages: but
I don't think it is fair to give an answer hiding workable
solutions.

<mild rant>

Maybe I'm just getting older or else the caffiene hasn't kicked in yet 
but this is something that has really started to bug me over the past 
few years.  While it may not be the case in the lower layers where
the laws of physics hold more sway, at least in the applications 
area the world is suffering from a bad case of "any solution that 
accomplishes the task is just as valid as any other" syndrome. Just 
because you might be able to figure out how to bend over backwards 
fifteen times and end up making alternate roots work for a few cases 
doesn't mean its a viable solution. Given enough money and time you can 
turn _any_ combination of random hairbrained ideas into a 'workable' solution.

IMHO, simply saying something is 'workable' is so far away from not being
sufficient that to use it as a point in a debate is specious at best. 
We're not here to just make something work. We should be here 
rigorously applying sound engineering principles that have been found to 
work for large heterogeneous networks. And one of those principles is that
infrastructure level hierarchical namespaces require one root.

In other words, its also been proven that using carrier pigeons as a 
transport layer is also workable. Does that mean that workable is sufficent?
No. Instead it means that being a 'workable solution' is meaningless...

</mild rant>

-MM

-- 
--------------------------------------------------------------------------------
Michael Mealling        |      Vote Libertarian!       | urn:pin:1
michael(_at_)neonym(_dot_)net      |                              | 
http://www.neonym.net



<Prev in Thread] Current Thread [Next in Thread>