[Dxspider-support] hop_table.pl questions

Dirk Koopman djk at tobit.co.uk
Mon Jul 11 11:33:19 BST 2005


On Fri, 2005-07-08 at 17:36 -0400, Dave Hawes wrote:
> Dirk,
> 
> Thanks for the tidbit on the anomaly, but does this relate to KF3B's 
> question about the fact that an incoming protocol with H1 does not 
> expire when decremented to H0?  Asked in a different way, does this 
> mean that the node sending protocols to KF3B-2 that he wishes to 
> expire at KF3B-2 should send them with H2, not H1?
> 
> Thanks for your interest in this.  We would like to learn more, so as 
> to be better able to control those pesky PC messages.

These are deep questions and my answers may cause a degree of
controversy.

The first thing that has to be said is that the protocol, as it stands,
is broken in many ways - at least in the environment that it is used
today.

Hop counts are, by and large, meaningless. There are many reasons for
this but they include (in no particular order):

	* What right does X nine hops away have to determine that Y 
	  does not get the sentence (corollary: X one hop away to Y)? 

	* In a looped environment with everybody busily rehopping 
	  (nearly) everything, what use are hops anyway?

	* In a looped environment, where sentences cannot be uniquely
	  identified, how can hops do their job?

	* If you get something regarded as important, such as a spot,
	  and you get it with H1, you will accept that, stick it in
	  the deduping filter and then not pass it on. Unfortunately,
	  you will also fail to pass on the next 10 copies, all with
	  hop counts > 1.

Hop counts do work in DXSpider, but you have to be very careful and you
must not have any overrides of any kind present on a sentence. However,
having said that, I haven't tested it recently - largely because I
regard the subject as irrelevant. 

DXSpider does its (admittedly imperfect) best to reduce the need for hop
counts by trying to de-duplicate all sentences using a variety of
techniques. Unfortunately, they are not perfect and can be circumvented
by some things that occur. Hence we still have the occasional loop. The
deduping also can cause problems when sentences come in with H1.

I did start on a new protocol. Work has been stopped for several months
because I am up to my ears in (paid) work. I will carry on with it when
I have a bit more time. It does address these (and other) problems.

Dirk G1TLH




More information about the Dxspider-support mailing list