Optimze data exchange

[MaStr]

Hi, the current implementation downloads work like normal browser downloads. Is there a resume functionality- I assume not. Even the detection if the remote file is bigger, than my local one and then start to download is not very optimal.

Under heavy network-load or in disconnect situation the "download one file completly" is not very useful. What do you think of a DHT-torrent implementation for optimistic mode (and serving it)? (optional feature)

Matthias

[adulau]

There was a resume functionality in the past but I removed it when you have file with same name but different content it was an issue. Resuming different files ;-)

I think the block-oriented hash might solve two issues:

• The duplicate files with different names
• Downloading random block until the file is complete.

Implementing Kademlia could be a solution but that means Forban node search will be a mechanism for finding the nodes. Another approach would be to use a simple technique "à la ssdeep" where each file is represented by a set of "hashed" blocks. The index would be a simple list of files with a set of blocks. and then Forban could find the same content even if they have different name.

What do you think?

[MaStr]

After reading a few thing, I think we have to discuss that whole topic a bit more. I'm going to write more about that later.