On Saturday 20 Apr 2013 20:26:30 Arne Babenhauserheide wrote:
> Hi G. Nitesh Bharadwaj,
> 
> Am Samstag, 20. April 2013, 12:21:37 schrieb Nitesh Bharadwaj:
> > Please reply if you find this an interesting project for GSOC. Also, you
> > could give suggestions to improve my idea.
> 
> Your proposals sounds pretty interesting, I think, though for starters 
> especially the part where Freenet runs on Android.
> 
> If you would be able to make Freenet run as an Android app, that could be 
> quite an interesting project. 
> 
> Freenet poses special challenges for mobile development, though, for example 
> high CPU, storage and bandwidth requirements, which would have to be 
> restricted to places with Wifi access - or to wifi-direct connections - as 
> well as to times when enough power is available. You would have to find ways 
> to deal with those in a user-friendly way.

Right. IMHO the way to deal with this is:
- Port Freenet to Android.
- Add a special "mobile" mode. This does not (usually) route. It keeps a 
connection to a "main" Freenet node, over FNP, and to its friends; it uses the 
same connectivity tricks that Freenet does.
- Add an easy way to exchange noderefs between 2 phones in close physical 
proximity. This would establish a connection between their respective "main" 
nodes, not the phone nodes.
- Add a means to access Freenet via the main node. For some things 
(particularly Sone and long-term downloads) we want the "main" node to do 
everything, and we just receive the data, so it's effectively a proxy, but for 
browsing, small files etc we might want to route and send requests through the 
main node's darknet peers. But since we're not routing requests, there's no 
plausible deniability (although that's always been dubious) - so it only works 
if they have a high enough trust rating.
- Possibility of actual routing as something you can turn on e.g. when you have 
generous wifi. But the assumption is this is going to be rare.
- Look into transferring larger amounts of data between phones when they are in 
close proximity. Firstly, to fill up datastores with stuff that's in your 
friends' datastores. Secondly, to fulfill nearby requests (Bloom filter 
sharing). Thirdly, pub/sub and long-term request routing (similar to 
sneakernet).

IMHO the first 3 items are more than enough to make a mobile client useful.
> 
> Wifi-direct connections themselves might be possible to realize as transport 
> plugins.

I don't think this is related to transport plugins. We don't want to talk to 
treat nearby nodes as "opennet" peers, a la Haggle, do we?

Attachment: signature.asc
Description: This is a digitally signed message part.

_______________________________________________
Devl mailing list
[email protected]
https://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to