[mnet-devel] Distributed Content Search

[Some Guy]

 --- Zooko O'Whielacronx <zooko at zooko.com> wrote:  
> [!@!@#$ don't print my email online!!!] asked a bunch of questions of icepick, but I'm replying
with my own answers.
> > What do you do with all the content trackers?  How do you implement a search engine in a DHT?
> 
> Mnet Record Keeper design doc
>
http://cvs.sourceforge.net/viewcvs.py/*checkout*/mnet/mnet_new/doc/record_keeper.html?rev=HEAD&content-type=text/html
> 
> mailing list: "open issues in record keeper design"
> http://sourceforge.net/mailarchive/forum.php?thread_id=2844317&forum_id=7702
> 
> mailing list: "open issues in record keeper, part 2: encryption"
> http://sourceforge.net/mailarchive/forum.php?thread_id=2844318&forum_id=7702
> 
> mailing list: "open issue in record keeper, part 3: efficiency"
> http://sourceforge.net/mailarchive/forum.php?thread_id=2844319&forum_id=7702
> 
> mailing list: "open issues in record keeper, part 4: public records"
> http://sourceforge.net/mailarchive/forum.php?thread_id=2844320&forum_id=7702
> 
> mailing list: "open issues in record keeper, part 5: load balancing"
> http://sourceforge.net/mailarchive/forum.php?thread_id=2844321&forum_id=7702

Ok, thanks.  I see now where you guys are at.  For freenet there is a application called Frost,
which uses freenet like a dumb DHT to post to boards and request stuff off them.  You poll under
keywords and if someone has requested one of your files, you insert it.

It leads to a bunch of stupid polling which they need to fix.  This guy Tom is proposing some TUK
kind of routeable object.  Which is supposed to let an administrator control access to boards, to
keep people from flooding them.  If you're interested you can read the last giant paragraph of
this doosy:
http://article.gmane.org/gmane.network.freenet.devel/7080

Personally, I've wondered if we shouldn't just attack the solution from the other end.  If we
assume that only a few people are nuts enough to fill in all the metadata, maybe we should let
them build big indexs which could then be downloaded and searched locally.  

This could ensure the metadata is accurate.  I could see people posting files under fake names in
order to get people to download commercials.

The guy that writes this kind of metadata, would be kind of the DHT equivalent of a DJ.  He'd go
around and listen to new stuff.  People would send him new stuff.  And he'd publish indexes and
maybe reviews of what he thought was cool.

When a newbie says "share these 10GB of files", if they are exact matches of other files in an
index, we could let MNet periodically publish them.

Of coarse it should be possible for anyone to build an index and share a link to it with his
friends.  We could also make a nice tool to merge indexes, to take out duplicates.  People will
probably want to pick thier indexes.  (not everyone will want the "giant nasty farm animal pron
index")

This was one idea we thought would have been good for a freenet search engine a while back.  Does
it seem acceptable for music?


Probably a 1GB metadata index of music, could be downloaded once and you'd have a big enough list
of stuff to search through.


__________________________________________________________________

Gesendet von Yahoo! Mail - http://mail.yahoo.de
Logos und Klingeltöne fürs Handy bei http://sms.yahoo.de


-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?   SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/
_______________________________________________
mnet-devel mailing list
mnet-devel at lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mnet-devel