Good news after some abstinence from me.
We decided to totally reorganize the udp and udp reliable part. That took of course time to thoroughly investigate after coding was done.
Anyway we come up now with 150 lines less code than before and - more important - it is more straight forward and cleaner than before. We also improved the logging speed (if many transactions per seconds appear this is somewhat not irrelevant, however, the logging will be reduced once it will have to deal with more people). We cleared one memory waste but we will seek for not freed memory with 0.6 more thoroughly.
If 0.59.3 runs stable until the planned eta of 0.6 (to be announced) then the next version will be 0.6 instead of 0.59.4
Also, if it runs stable for the next days then the launch early june seems realistic to us. No major feature is missing, however, we need the proxy part to be done to be more safe against early DDoS attacks. But that is no rocket science so we should get in ready early enough. Features like standing orders can be implemented later on without major troubles as well as the decentral inclient marketplace.
Just created 500 transactions in ~2.5 minutes which translates to 3.3 transactions per second (but note: this speed is not limited by the network but it is limited by my local computer's speed). More important here is not the speed but: all 500 transactions got confirmed and integrated in the block. Then, a new client, was able to download that big block without problems. And that's the major difference to 0.59.2: Even with 0.59.2 we were able to create 2000 transactions (should be still possible
) _but_ a new client was not always able anymore to get that block because of its size. Not always translated to: "a node with very good connection to the network had a good chance to get that block, others not". A block with 500 transactions will be splitted to roughly 250 single udp packets/ethernet frames and if one of them was lost on the way then the 0.59.2 client requested all parts again.
With 0.59.3 we not only split those big blocks into small *reliable* packets - if any packet gets dropped then only that one will get queried again - but with our redesign every udp message in the program that is considered to be too big will automatically splitted into reliable ones. So the reliability aspect moved completly into the inner-udp class. But enough said ---
Right now 4 nodes are up, running and synced. Please report if you have troubles syncing beyond block #14 or #15 (but wait to get it synced first). It should not but who knows....
I'd just like to make a final check on all nodes and make a final test in the other chain.
So I will come back here in about ~2hours and I really hope that we can deliver the update then to you
best regards exocoin
