We have until the 22nd to rank these, finalize our decision on how many
projects to accept (at most), and assign them to mentors. Currently we
have in the top 4:
- Jerome Flesch's super-FUQID app
- Michael Rogers' proposed work on congestion control and load balancing
- Nextgens' Installer, GCJ, packaging proposal
- Michael Bendersky's search engine

Personally I think Freemail should be up there somewhere, however Ian is
against this for reasons which I am not yet certain of.

On Wed, May 10, 2006 at 03:47:11PM +0100, Matthew Toseland wrote:
> SoC proposals so far:
> 
> FUQID clone: Is basically a matter of coding to a design. There will be
> some queries on the global queue interface, there may even be some
> things that need changing, but not much, as all the functionality is
> already there and used by FUQID. There are at least 3 apps for this. One
> of them is more of a generic filesharing app, talking about indexes and
> index plugins.
> 
> Freemail: I think would be good. It's important from the political angle,
> it's also important from the point of view of dogfood (being self
> hosting). It would require considerably less supervision than FUQID, and
> has been extensively discussed already. Dbkr is offering to do it, and
> we can trust him as he's worked on Freenet and has already done much of
> the work for Freemail.
> 
> Congestion control and load balancing is a controversial one. I think
> this would be really useful. I'm not happy about shipping 0.7 with the
> current load balancing scheme. Firstly I'm not sure it works - at least
> with inserts. I get reports fairly frequently from people with all their
> nodes backed off and so on. Secondly, it is grotesquely insecure. And I
> don't think this would be fully fixed by premix routing. However, this
> would require detailed supervision. It's something that I'd like to do
> for 0.7.0 anyway, and I'd welcome having that level of help; although
> it's not self-contained, it gets my vote.
> 
> OS/X bundle: This is a self-contained project which requires essentially
> no supervision. It would however require a good deal of interaction with
> the installer; nextgens wrote the installer, but if none of his apps
> succeed he will not be available over the summer. Probably more of our
> users use linux now than use OS/X but nonetheless it'd be good to have
> proper OS/X support as many people who'd consider using freenet use
> OS/X.
> 
> JFKi: This is a fairly self-contained project to improve the security of
> the transport layer. Right now, it's ephemeral Diffie-Hellman). It will
> soon be something like STS (signed DH). The proposal is to implement
> JFKi, which is an industrial strength, rather complex, connection setup
> protocol, which is strongly resistant to DoS and in which almost
> everything can be precomputed. This would be wrapped in our existing
> symmetric crypto layer to prevent traffic profiling, just as we wrap DH
> now. I think this is fairly low priority, the main reason being that
> only those who have a noderef can DoS the node (on CPU or memory), and
> very few will on darknet. On opennet it's a different matter, but still,
> it doesn't seem that important to me.
> 
> Freesite spiders: A good spider would support both metadata and content
> based indexing. It has been proposed for spiders to automatically
> publish index sites a la 0.5; this may be useful, but publishing
> searchable indexes is probably just as important if not more so. There
> are people working on spiders already for this purpose, but that doesn't
> necessarily disqualify this. One of the proposee's has experience
> working on distributed search. One at least proposes an external search
> app; a plugin might be better, or both could be implemented. I think a
> good spider would be a reasonable contribution anyway.
> 
> Untrusted plugins: This would allow for dynamic web content and so on
> over Freenet; pages like Slashdot, or a wiki. The web is important IMHO,
> even if it's not the best UI in the world, it's the most used UI in the
> world! Would be quite a difficult project to get completely right, and
> might require significant supervision. I think it might be worth it; Ian
> doesn't. I think maybe we should postpone this one for next year.
> 
> Local messaging and file sharing: IMs, chat boards, searchable file
> archives, on your direct peers, or nodes only a few hops away. This
> would encourage people to connect to their friends, help to find stuff,
> and build the darknet community. However, it might be seen as too close
> to "filesharing". I think this would be useful, but maybe not as useful
> as some of the other proposals.
> 
> Installer etc: Nextgens has proposed a grab bag of fixing the installer,
> creating a systray icon widget, creating an uninstaller, fixing the
> serious flaws in the php on the website, splitting up freenet-ext.jar,
> fixing the rest of the GCJ problems and packaging freenet for linux
> dists. I think this is an interesting proposal. We absolutely must have
> an uninstaller, there are serious issues with the website, an icon would
> be good, and sorting out free JVMs and packaging would be really nice
> for the large part of the community that runs freenet and the much
> larger group of people who would run freenet if it was well-behaved and
> didn't require sun java. Wouldn't require much supervision. This is by
> nextgens, just as dbkr's is by dbkr; obviously these are people we know,
> so we have to be careful not to show favouritism.
> 
> STUN/UP&P/etc: STUN would be useful, although it relies on public
> servers (these are used by lots of other apps too). UP&P would also be
> useful, although there are security issues with it, and also deployment
> issues (it's often either disabled or not working). The basic objective
> here is to 1) find our current IP, and 2) open the port if possible.
> ARKs will go a long way towards fixing these problems however. This is
> another middle-ranked application IMHO.
> 
> Transport plugins: This would be really cool, but is probably too
> advanced for *this* summer. I don't personally see why the interface and
> a few really basic plugins can't be written in 2 months, but the proposer
> (nextgens) isn't convinced. It would probably require detailed
> supervision. Obviously carrier pigeon etc will have to wait; high
> latency transports will require major core changes which we're not ready
> for yet, probably in 0.8 (passive requests maybe). So another middle
> ranking app; and Ian will probably say no.
> 
> FCPv2 libs: There is a proposal to write FCPv2 libraries in a number of
> languages. This will be useful; the main downside is that FCPv2 is quite
> simple, so many people trying to write apps will not be prevented by the
> lack of a library.
> -- 
> Matthew J Toseland - toad at amphibian.dyndns.org
> Freenet Project Official Codemonkey - http://freenetproject.org/
> ICTHUS - Nothing is impossible. Our Boss says so.



> _______________________________________________
> Tech mailing list
> Tech at freenetproject.org
> http://emu.freenetproject.org/cgi-bin/mailman/listinfo/tech

-- 
Matthew J Toseland - toad at amphibian.dyndns.org
Freenet Project Official Codemonkey - http://freenetproject.org/
ICTHUS - Nothing is impossible. Our Boss says so.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: 
<https://emu.freenetproject.org/pipermail/tech/attachments/20060510/90e6f73e/attachment.pgp>

Reply via email to