> Actually, a server (probably a separate server from the Freenet
> server) could load all the metadata files and index them, and respond
to
> queries.  I don't think we should bolt on search-engine functionality
to
> Freenet, this might bog things down a bit.  There are a lot of
problems
> with this idea, but its just random fluff from my head.

Good idea.  There's no search built into HTTP but Google and Altavista
do a pretty good job, and they'd do an even better job if there was more
metadata on the web.

There are a lot of different ways to do indexing and searching, and most
of them will be "bolt-on".  That doesn't mean there shouldn't be a
standard index/search built into Freenet, just that whatever it is, it
shouldn't interfere with the ability to do it some other way externally.
But as long as you can insert & request data by key directly, and you're
not forced to use some kind of search request, or forced to provide all
kinds of metadata or talk to an index server when you insert, there
shouldn't be a problem with indexing and searching however you want.

> 
> ...ooh, also, I'd like to say that if we have searching, we should
still
> allow documents to exist that can't be searched for.  If we force the
> person doing the insert to submit his metadata to a search-server,
than he
> can opt out and keep his document hidden from search.
> 

Exactly!  I think straight key-based insert and request operations
should always be supported, and interfacing with searching and indexing
mechanisms should always be optional.


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

Reply via email to