On Wed, May 03, 2000 at 10:13:57AM -0500, Jim Duey wrote:
> What are the disadvantages of using versioning for the data modification
> function?  The attraction I see is that you can always go back to
> recover previous versions of the data, until those versions are purged. 
> I believe that the node software would be less complex to implement than
> some kind of replacement scheme.

I agree with this.  Versioning would also make a replacement attack far less
effective.  Because someone could make a new version of a file, but can't
delete old versions of the file, if someone realizes that the new version is
bogus then the bogus version can be replaced with a newer version based off
an older version.  Adding versioning would require that the protocal be
changed so that new version messages would be transmitted from a node to
other nodes telling the other nodes that a new version is availiable.
Without such a change, other nodes would use an older version of the file
because they don't know that a new version exists.

(On another note, I'm currently writing a Freenet client module in Python.)

> Ian Clarke wrote:
> > 
> > Firstly, it is great to see that people are implementing inter-node
> > encryption, and hopefully the other types of encryption we need will
> > follow shortly.
> > 
> > One thing I am personally a-little unclear on is what encryption
> > negociation mechanisms people hope to use (ie. a:"I talk blowfish and
> > twofish" b:"I talk XOR and twofish" a:"lets talk twofish" b:"ok"), and
> > how these will integrate into handshaking, and also how these will
> > integrate into the current mechanism (ie. maintain backward
> > compatability - which I think we should do if possible).
> > 
> > With that out of the way, probably the most exciting think that I am
> > looking forward to seeing is data modification.  There are several
> > interesting aspects to this:
> > 
> > - What PKK crypto do we use?
> > 
> > - How do we ensure that update messages are propogated to all or most
> > nodes holding the data?
> > 
> > - What form should the updates take? (ie. a simple "replace the entire
> > data with this / delete the data", or a more sophisticated "diff" style
> > update).
> > 
> > Updatable data will open the door to many cool new features, such as
> > useful hyperlinking, and such-like.
> > 
> > One thing I was thinking of is placing Java class files on Freenet, so
> > requesting "com.sun.whatever.Class" will return that class file.  Java
> > could then be hacked to run Java straight off Freenet.  The problem here
> > is that there is nothing to prevent others from placing whatever they
> > want under "com.sun", so perhaps some form of chaining where "com.sun"
> > will actually be an index of classes under "com.sun.*".  Obviously this
> > needs to be thought out further...
> > 
> > Ian.
> > 
> > _______________________________________________
> > Freenet-dev mailing list
> > Freenet-dev at lists.sourceforge.net
> > http://lists.sourceforge.net/mailman/listinfo/freenet-dev
> 
> _______________________________________________
> Freenet-dev mailing list
> Freenet-dev at lists.sourceforge.net
> http://lists.sourceforge.net/mailman/listinfo/freenet-dev

-- 
Travis Bemann
And yes, I know my mailing software is misconfigured.  My email address
should really be bemann at execpc.com instead of bemann at bemann.

_______________________________________________
Freenet-dev mailing list
Freenet-dev at lists.sourceforge.net
http://lists.sourceforge.net/mailman/listinfo/freenet-dev

Reply via email to