On Tue, 23 May 2000, Brandon wrote:
> > > Anyway, I've implemented the separate metadata (not attached to
> > > files) insert and request.
> > 
> > What is this good for? There is no need to keep MetaData seperate from the
> > data, since the MetaData is irrelevant without the data itself. 
> 
> I thought it had been agreed that there should be a separate metadata
> space which holds incriminating metadata, for the purpose of searching
> Freenet without knowing what's on any particular node. I don't yet have a
> searching solution, but here's an implementation of a metadata store for
> the searching to operate on.

OK, but that this has to be kept in a seperate datastore in this case, since
the closeness relation will be different from data routing.

> > > I also have a really neat structured query parser that returns a list of
> > 
> > And what in the world is this good for? I'm assuming this is in the client?
> 
> I don't have an application for this at the moment, but I figure somebody
> might want to use it or cannabalize some of it because it's pretty
> neat. Structured searches would be nice, where you could search for
> something based on a field in the metadata. But then, of course, we need a
> searching methodology to use that.

I think we should keep as little info as possible in the data that is searched
on. Just a set of search terms and references really. Possibly a signature.

> > > data insertion and retrieval, and write a file server node to keep
> > > information in Freenet.
> > 
> > Why? What is a "file server node"?
> 
> Something which periodically requests specified items from Freenet and, if
> it can't find them, inserts them, so as to always keep them around. I'm
> not sure how useful this will be in practice, but people keep asking about
> it and if I implement it, they will stop asking.

It doesn't work. If the node keeps requesting it, chances are it will keep it
alive in the node that it has a refernce to, but not in the node that is the
"epi-center". 

This is not a node anyways, it is a client. A special purpose automated client,
but a client non the less.

> > AFAIK Scott already fixed support for meta-data fields preceding the actual
> > data. That is where the ContentType should go - using it is just a client 
> > issue.
> 
> Yes, I'm going to work on the client classes so that they process trailing
> field metadata so that the www interfaces can add ContentType trivially.

That is good. The client classes need work, unfortunately I have been afraid to
change them now that we have Freeloader et al depending on them. 

Among other things, I would like to merge the "Simpleclient" that the GUI
authors wrote code with that used for a better single commandline client.

-- 

Oskar Sandberg
md98-osa at nada.kth.se

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

Reply via email to