Ian Clarke <[EMAIL PROTECTED]> writes:

> > How do you see Frost going in the medium term? I think we need some
> > convergence of Frost into Fproxy.
> 
> I disagree, Frost is fine as a separate application.  If they want to
> include a freenet.jar in their distro, then that would be fine, but it
> isn't appropriate for us to include their functionality in the Freenet
> platform.  Think of us as the Linux Kernel.  Frost is welcome to think 
> of themselves as RedHat, including our jar with their code if they want.  
> We should take a minimalistic approach towards client functionality in 
> Freenet - Fproxy is probably stepping over the line here, but I think we 
> should tolerate one exception.
> 
> Ian.
> 
In a perfect world, we can insist on other projects merely using the
core freenet provides as a building block to provide real
functionality, but frankly, freenet isn't that popular.  The core
freenet application is still changing enough that bundling it with an
application is going to lead to that application needing to be updated
as often as fred is, in order to keep up with the bug fixes and forced
upgrades.

Fproxy is definitely over the line in your view, but I see fproxy
being a *very* necessary step towards making freenet usable during a
time of constant change.  At this stage in freenet's development, we
(as the core development team) need to provide all the necessary
functionality for the average freenet user to get enough benefit out
of fred to justify installing it.  As fred makes the step to 1.0 and
beyond, these pieces of non-essential functionality should be split
off as seperate projects, but until then, the survival of the project
requires them being bundled with the current node.

Matthew Toseland writes:
> How do you see Frost going in the medium term? I think we need some
> convergence of Frost into Fproxy. For various reasons (mostly the fact
> that KSK-based boards are horribly floodable), I don't think we
> necessarily want a Frost-web-interface in the main freenet package
> yet... and we certainly would have second thoughts about packaging a
> GUI Swing client with Freenet. 
I agree we don't want KSK-based boards or Swing GUI.

> However, I think the following might be
> worth considering:
> * Signed, user-verifiable identities within Frost (but still based on
>   boards, of course).
> * Automatic insertion of summary files with all files inserted by a
>   given identity to a given board, or adopted by that identity (a
>   checkbox when downloading?). Summary file contains metadata etc for
>   each file, just as a message post would, but is cumulative. Is
>   inserted at SSK@<pubkey of identity>/<board name>/<serial number> or
>   some such standard place, with a well-known format.
> * An fproxy filter to recognize these keys and turn them into a pretty
>   HTML interface. Frost should have a link to them.
>
I really approve of this line of thought; it was one I went down years
ago when I first got involved in freenet.  The killer application I
see involves each freenet user maintaining (with a lot of automation
to make this easy enough for the average user) a pseudonymous
"link-list", with two types of links: 

1) links to actual content, hopefully with a bit of review/evaluation,
to indicate that that user "approves of", or in some way just
recommends people download that content.  This section/category would
automatically include all content inserted by that user, so that data
would be retrievable.  

2) links to other users' link-lists; sort of like a distributed TFE,
so that people who have similar interests can cluster links to each
others' lists, along the lines of the web-of-trust idea, but with
additional connotations of liking similar content, kinda like
webrings.

Of course this link list should be machine parsable, so that "spiders"
could index the content and calculate reputations for content based on
how many people link to it.  (More complicated algorithms would of
course be worth looking into of course, especially if people were able
to have anti-links, letting others know that certain content/
link-lists were "bad", in whatever way they deem appropriate.)

> Thus inserts lead to BOTH Frost AND Fproxy being able to get to new
> content. I'm not sure whether we want to directly support KSK-based
> boards on Fproxy, but one possibility would be to add extra sections to
> the summary files listing other identities that insert useful stuff,
> creating a web of trust... what do you think about all this?

If others aren't contrary to the idea, I'm all up for starting to
prototype the "summary file"/"link-list" format, to see how practical
this is going to be.

Thelema
-- 
E-mail: [EMAIL PROTECTED]                         Raabu and Piisu
GPG 1024D/36352AAB fpr:756D F615 B4F3 BFFC 02C7  84B7 D8D7 6ECE 3635 2AAB

_______________________________________________
devl mailing list
[EMAIL PROTECTED]
http://hawk.freenetproject.org:8080/cgi-bin/mailman/listinfo/devl

Reply via email to