Oh and to add to the idea, perhaps with a mapreduce implementation of nutch it could lend itself to a style of processing much like [EMAIL PROTECTED] in the way that you could duplicate the segments that are sent down for fetching as a map job, run the tasktracker on the remote client, process the fetch and submit the processed segment up and take it from there (atleast offloading some of the fetching needs to your remote nodes)
i can't imagine processing requests in a p2p fashion in the speed that search requires, but it may be a nice implementaiton of a global information repository that is searchable, just not a search engine itself. -byron --- Byron Miller <[EMAIL PROTECTED]> wrote: > the mod_estraier is an interesting concept, albeit > expensive i believe :) > > I would think it would be easier just to post > process > your apache logs like the google perl scripts for > google sitemaps. > > > > --- Alexandre Dulaunoy <[EMAIL PROTECTED]> wrote: > > > Hi Guys, > > > > I was wondering if any of you has already checked > > the P2P approach in Hyper > > Estraier ? > > > > > http://hyperestraier.sourceforge.net/nguide-en.html > > > > Another nifty feature using the P2P approach is > the > > mod_estraier apache > > module : > > > > http://modestraier.sourceforge.net/ > > > > If you have any feedback compared to Nutch, I > would > > be very happy to hear > > them. > > > > Have a nice day, > > > > adulau > > > >
