Author

Topic: Sunny King's view on OP_RETURN small data transaction and block pruning (Read 586 times)

sr. member
Activity: 448
Merit: 250
Gosh, I think I started to appreciate Sunny King's work recently,

I decided that I seem to like POS after all Smiley
hero member
Activity: 815
Merit: 502
sigmike's reply:
Original post at: https://www.peercointalk.org/index.php?topic=4086.msg39228#msg39228

I think the most important here is this:

Quote from: Sunny King on April 29, 2015, 07:42:31 am

"2) From bitcoin/peercoin core point of view, these probably can be pruned immediately from unspent set. Note pruning does not remove it from blocks. It's still on the blockchain and can be retrieved from blockchain at any moment albeit with higher access cost."


First pruning doesn't mean we shrink the blockchain. Some nodes will have to keep prunable data forever, whether there's a special flag or not. It is required so that new nodes can validate the blockchain. If there's anything missing in the blockchain you can't validate it.

When we talk about prunable data we mean that a client can delete these data from his own copy of the blockchain after he used them to validate a block. He won't need these data to validate future blocks. But then he's not a complete full node anymore, because he can't provide the full blockchain to others.

Second, anyone who doesn't use these prunable data and doesn't want to store the complete blockchain can delete the prunable data just after he received it. So a flag is not necessary: prunable data already means you can delete it as soon as you want.

This flag could be a hint, suggesting how much time nodes should keep it. But this decision is more a matter of how much they want to be a full node, and has nothing to do with the emitter or the data itself.
hero member
Activity: 815
Merit: 502
Original post at: https://www.peercointalk.org/index.php?topic=4086.msg39105#msg39105

Sunny King

My view on the issue at the moment:

1) There is no way to prevent ppl from stuffing data to the blockchain if they insist. This is somewhat due to script system (one might argue that even without the script system data can still be encoded in address or send amount). The only real discouragement of data on blockchain is the transaction fee. I am talking about general principles here, I have nothing against peerapps using the feature on peercoin. The idea is so long as fee is paid, it's considered okay. And users of this feature is actually being graceful to the network by not encoding data in obfuscated ways.

2) From bitcoin/peercoin core point of view, these probably can be pruned immediately from unspent set. Note pruning does not remove it from blocks. It's still on the blockchain and can be retrieved from blockchain at any moment albeit with higher access cost.

3) I would be very cautious modifying the bitcoin script system, due to the issue that maintaining such modification can be expensive long term. So one needs strong argument why its benefit is so great.

4) The proposed system of declaring different expiration to the network isn't quite consistent with my view that transaction fee is the only guard on data flood. It kinda suggests that fee alone is not enough, that additionally the network still expects you to be 'nice' and let nodes prune you more quickly.
Jump to: