Some notes:

Hardforks like Bitcoin ABC without a malleability fix are very unlikely to have payment channels, so the problem does not exist for those.

The designers of a hardfork which does have a malleability fix will probably know about payment channels, so they can just build a replay protection that allows the execution of old commitments. That needs some kind of timestamping of commitments, which would have to be integrated in the channel design. The easiest way would be to just write the time of signing the commitment in the transaction and the replay protection accepts old commitments, but rejects one's which were signed after the hardfork. These timestamps can essentially be one bit (before or after a hardfork) and if the replay protection in the hardfork only accepts old commitments for something like a year, then it can be reused for more hardforks later on. Maybe someone comes up with an interesting way of doing this without using space.

Nevertheless hardforking while having channels open will always be a mess as an open channel requires you to watch the blockchain. Anybody who is just not aware of the hardfork or is updating his client a few days too late, can get his money stolen by an old commitment transaction where he forgets to retaliate on the new chain. As other's can likely figure out your client version the risk of retaliation is not too big for an attacker.



2017-08-17 13:31 GMT+02:00 Bryan Bishop via bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org>:

---------- Forwarded message ----------
From: Christian Decker <decker.christian@gmail.com>
Date: Thu, Aug 17, 2017 at 5:39 AM
Subject: Re: [Lightning-dev] Lightning in the setting of blockchain hardforks
To: Martin Schwarz <martin.schwarz@gmail.com>, lightning-dev@lists.linuxfoundation.org


Hi Martin,

this is the perfect venue to discuss this, welcome to the mailing list :-)
Like you I think that using the first forked block as the forkchain's genesis block is the way to go, keeping the non-forked blockchain on the original genesis hash, to avoid disruption. It may become more difficult in the case one chain doesn't declare itself to be the forked chain.

Even more interesting are channels that are open during the fork. In these cases we open a single channel, and will have to settle two. If no replay protection was implemented on the fork, then we can use the last commitment to close the channel (updates should be avoided since they now double any intended effect), if replay protection was implemented then commitments become invalid on the fork, and people will lose money.

Fun times ahead :-)

Cheers,
Christian

On Thu, Aug 17, 2017 at 10:53 AM Martin Schwarz <martin.schwarz@gmail.com> wrote:
Dear all,

currently the chain_id allows to distinguish blockchains by the hash of their genesis block.

With hardforks branching off of the Bitcoin blockchain, how can Lightning work on (or across)
distinct, permanent forks of a parent blockchain that share the same genesis block?

I suppose changing the definition of chain_id to the hash of the first block of the new
branch and requiring replay and wipe-out protection should be sufficient. But can we 
relax these requirements? Are slow block times an issue? Can we use Lightning to transact
on "almost frozen" block chains suffering from a sudden loss of hashpower?

Has there been any previous discussion or study of Lightning in the setting of hardforks?
(Is this the right place to discuss this? If not, where would be the right place?)

thanks,
Martin
_______________________________________________
Lightning-dev mailing list
Lightning-dev@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/lightning-dev

_______________________________________________
Lightning-dev mailing list
Lightning-dev@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/lightning-dev




--

_______________________________________________
bitcoin-dev mailing list
bitcoin-dev@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev