[freenet-dev] Bogus network construction function?? was Re: [freenet-cvs] r18322 - trunk/freenet/src/freenet/node/simulator

[Michael Rogers]

Are you checking for duplicate connections and self-connections? In my 
code that was done in the connect() method.

Cheers,
Michael

Matthew Toseland wrote:
> Hmmm, we really need to fix this, if our simulations are skewed...
> 
> Any hints, Oskar? The network construction function is right at the top of 
> this file.. could you have a look at it? Thanks.
> 
> http://emu.freenetproject.org/cgi-bin/viewcvs.cgi/trunk/freenet/src/freenet/node/simulator/RealNodeTest.java?rev=18322&view=markup
> 
> On Monday 03 March 2008 21:40, robert at freenetproject.org wrote:
>> Author: robert
>> Date: 2008-03-03 21:40:44 +0000 (Mon, 03 Mar 2008)
>> New Revision: 18322
>>
>> Modified:
>>    trunk/freenet/src/freenet/node/simulator/RealNodeTest.java
>> Log:
>> comments
>>
>>
>> Modified: trunk/freenet/src/freenet/node/simulator/RealNodeTest.java
>> ===================================================================
>> --- trunk/freenet/src/freenet/node/simulator/RealNodeTest.java	2008-03-03 
> 13:38:58 UTC (rev 18321)
>> +++ trunk/freenet/src/freenet/node/simulator/RealNodeTest.java	2008-03-03 
> 21:40:44 UTC (rev 18322)
>> @@ -29,6 +29,10 @@
>>  	
>>  	/*
>>  	 Borrowed from mrogers simulation code (February 6, 2008)
>> +	 --
>> +	 FIXME: May not generate good networks. Presumably this is because the 
> arrays are always scanned
>> +	        [0..n], some nodes tend to have *much* higher connections than the 
> degree (the first few),
>> +	        starving the latter ones.
>>  	 */
>>  	static void makeKleinbergNetwork (Node[] nodes, boolean idealLocations, 
> int degree, boolean forceNeighbourConnections)
>>  	{
>>
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> Devl mailing list
>> Devl at freenetproject.org
>> http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl