multiple routes

[Bdale Garbee <bdale@col.hp.com>]

> This issue of multiple ax.25 paths is, I hate to remind people,
> [no, no, tell the truth -- I don't hate it at all] is one
> part of the problem that I tried to raise, and address, several
> months ago, which was, shouldn't we be able to handle multiple
> physical paths to a single IP address and choose the best one
> at any given time (or, really, for a time interval delta-t).

Doc, I think many of us agreed with this part of your assertion, but not to
the extent that anyone was willing to write the code to do anything about it.

What I, at least, found unpleasant in your original proposal was the mechanisms

discussed to discover the routes.  Brian's suggestion of just caching the last
couple of learned routes would have substantially less impact on network
performance than the flooding algorithms you were proposing, and so might be
worth investigating.

Personally, I think that just allowing the program to replace an existing AX.25
route with a new one when heard would solve in excess of 90% of the perceived
problem.  And this should not be hard code to write.

But I'm not going to write it, at least not today.  :-)

Bdale