Have compression (was Re: Standards (was [BitTorrent] Back to

Justin Cormack justin at street-vision.com
Thu Feb 10 09:44:30 EST 2005


> 
> 
> > > I agree with the testing part.  Olaf's got a clean, simple
> > > implementation of Merkle hash trees.  The only major departure from it
> > > that I see might be dynamic piece size.  I'd be up for testing out if
> > > you want, just to see how it goes.
> > 
> > I will look into making a BT simulator. I dont have enough spare machines
> > to make a decent size cluster at the moment, so I will look at how to 
> > simulate low bandwidth links between processes on a single Linux box. Not
> > sure how soon I can get that running though. It would be good to have 
> > replicatable realistic simulations with measures of overhead and timing.
> 
> Some simple code in the client could make the simulation fairly
> realistic.  In particular, most clients (including mine) have an upload
> bandwidth limit option.  A download bandwidth option would be pretty
> easy, and a series queue between the socket and the message queue could
> emulate latency.  However, so far, I haven't found latency to be a
> significant issue, so long as you queue up a few requests when unchoked.

I thought of that, the problem I saw was that they bandwidth limitations
might be enforced in a bursty way, and it might interfere with the
clients own choking algorithm. But it would do as a start.
 
> One of the two main reasons I wrote a client was to have a test-bench
> that I'm very familiar with to try out P2P ideas.  I feel very strongly
> about coding and testing extensions as part of the process of making
> extensions.

I am currently rewriting my client so its much more modular and the bits
of functionality are entirely isolated. It actually leads to a much better
design anyway.
 
> Not that trackers are hard to come by, but I could set up a tracker and
> seed from home specifically for testing BT extensions if that would
> help.  I already do that for my own testing (using Azureus).

I have tended to set up a tracker myself or use random fedora/gentoo/BSD
torrents. One thing that might be useful is a machine that hosts several
different trackers and clients for interoperability testing, but thats
less important for extension testing.

Its good to put up test cases for hash functions for sanity checking I
think, for Merkle trees.

j


 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/BitTorrent/

<*> To unsubscribe from this group, send an email to:
    BitTorrent-unsubscribe at yahoogroups.com

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/
 





More information about the BitTorrent mailing list