pmarches / peercentrum-core Goto Github PK

Ensures the money on the server's side is used for a particular application (says P2PBlob) instead of H2PK. Otherwise we do not know what application should receive the money. The accounting money should be pushed to each application instead of the applications pulling from accounting.

When the DHT application starts, we need to populate the routing table it with the closest nodes to the local node. We currently grab a sample from the node database, but a lookup on our own node ID will yield closer matches.

Add a ever increasing nonce to each message to prevent replay attacks

Try to make everything more asynchronous with RXJava?

Does it provide benefits over secp256k1?

UPnP works fine for home networks, but UDP hole punching would be even better!
Maybe take the opportunity to implement these nifty features:

• Use Diffie Hellman for an ephemeral session key
• skip the TLS negociations & certificate exchange
• uTP has better congestion control for long running background processes like bit torrent, can we make use of this?

Another great way to showcase CCC, is to build a email client on top.
Will require at least these services:

• Queuing
• Blob storage
• Identity management/Key retrieval

The DHT and concensus DB applications currently have a client, transaction and server component. This makes it complicated to subclass these applications. Need to find a good paradigm for subclassing.

Keys can be rotated, upgraded , revoked. Many keys can be associated to an identity?

Should we include popular non-core applications like Blob, queues, h2pk or have them as a separate projects?

How can an application use services from another application?

Use OSGI?
How can we sandbox certain untrusted applications? Is the JVM security model robust enough to prevent our coins from being stolen?
Think about a marketplace for applications, eclipse-like.

This is to improve the chances other nodes will be able to connect to the local node.