Update on Test Net Release
Update on what’s going on for the ark code.
With the last testing of the code, it has become clear that the event management is a complete mess and variables are interacted by different actors, triggered by network events. So, we are not even sure that “last Block” variable remains the same during some of the processing.
This enables many improvements in the protocol. Things in the works on the pipeline:
- Relay node and forging node are clearly separated
- Much faster rebuilding (I am now able to mount network interface for when rebuilding)
- Broadcast block header only. Since majority of transactions are likely already received by all nodes before a block is forged.
- Better block processing (no need to remove unconfirmed transactions that are included in the block)
I will release the new version next week, but only for internal testing with tharude. I am progressing really well by the way. Things are completely clear now in my mind.
Also, a change in the protocol as follows:
- No maximum block size
- Only limit will be: As many transactions as the forging node can verify and include in less than block time/10. So basically, the network will scale with the power of the forging node hardware
- ARK will never broadcast full block. Only block header + transactions IDs. Reason is that most likely, a node has already received all transactions, and needs to combine with the block header to reconstruct the block.
I will integrate the nonce later on when we will have hundreds of tx per block so compression will be 30:1 or so.
If you wanna check out the code as it’s being updated for fixcrypts internal test before testnet. This is the link to that github branch https://github.com/ArkEcosystem/ark-node/commits/newtestnet
Update from Fixcrypt