Question about checkpointing in practice

I did a presentation on peercoin yesterday, and someone told me that checkpointing consists of rollbacks. If an attack has to be countered by using the checkpoint system, would that be a rollback to a point in time for the entire network, also for valid transactions? Or would only the bad parts of the blockchain be filtered out?

Congrats on doing the Peercoin presentation, love to hear that type of news.

On the rare occasion some one is using a small number of nodes to deliberately try and fork the Peercoin blockchain, and it is a clear intent to artificially fork the network to gain some sort of advantage (or to simply create multiple forks all over the place).

Sunny can release an official signed checkpoint, which basically tells all nodes which chain is the more authoritative one.

Any valid transactions that were processed on the “bad fork”, get reorganized into a new block on the good block chain.

Valid signed transactions aren’t compromised when checkpointing occurs. (Only new created blocks for mining and minting)

We’re about to find out more info about this in the upcoming interview with Sunny.

But the point is, valid transactions are safe, so there is no worry there.

Of course, I could be wrong, so I’ll wait for mhps or Ben to confirm.

That is very good to hear, thanks!

If Sunny King pushes a checkpoint that is not part of the main chain, we would all switch to a new chain.

It is a rollback if this new chain only includes transactions up to a certain block.
But Sunny King can also build a new chain with only the bad parts filtered out.
That’s up to him.

In any case you could send again your transactions if they were not included in this new chain.

We discussed that here:
Since then I discovered that the client may actually send these “back to unconfirmed” transactions automatically: