Johnson, Yeah, I do still see the issue. I think there are still some reasonable ways to mitigate it. I've started revising the extension block specification/code to coexist with mainchain segwit. I think the benefit of this is that we can require exiting outputs to only be witness programs. Presumably segwit wallets will be more likely to be aware of a new output maturity rule (I've opened a PR[1] which describes this in greater detail). I think this probably reduces the likelihood of the legacy wallet issue, assuming most segwit-supporting wallets would implement this rule before the activation of segwit. What's your opinion on whether this would have a great enough effect to prevent the legacy wallet issue? I think switching to witness programs only may be a good balance between fungibility and backward-compat, probably better all around than creating a whole new addr-type/wit-program just for exits. [1] https://github.com/tothemoon-org/extension-blocks/pull/16 On Mon, Apr 10, 2017 at 06:14:36PM +0800, Johnson Lau wrote: > > > On 6 Apr 2017, at 01:43, Christopher Jeffrey wrote: > > > > > >> This hits the biggest question I asked in my January post: do you want > >> to allow direct exit payment to legacy addresses? As a block reorg > >> will almost guarantee changing txid of the resolution tx, that will > >> permanently invalidate all the child txs based on the resolution tx. > >> This is a significant change to the current tx model. To fix this, you > >> need to make exit outputs unspendable for up to 100 blocks. Doing > >> this, however, will make legacy wallet users very confused as they do > >> not anticipate funding being locked up for a long period of time. So > >> you can’t let the money sent back to a legacy address directly, but > >> sent to a new format address that only recognized by new wallet, which > >> understands the lock up requirement. This way, however, introduces > >> friction and some fungibility issues, and I’d expect people using > >> cross chain atomic swap to exchange bitcoin and xbitcoin > > > > Yes, this issue is probably the biggest edge case in the proposal. > > > > I think there's two possible solutions: > > > > First solution: > > > > Like you said, add a maturity requirement for exiting outputs. Likely > > lower than coinbase's 100 block requirement. To solve the issue of > > non-upgraded wallets not being aware of this rule and spending early, > > have upgraded mempool implementations accept/relay txs that contain > > early spends of exits, but not mine them until they are mature. This way > > non-upgraded wallets do not end up broadcasting transactions that are > > considered invalid to the rest of the network. > > This won’t solve the problem. Think about the following conversation: > > Alice (not upgraded): Please pay 1 BTC to my address 1ALicExyz > Bob (upgraded): ok, paid, please check > > 10 minutes later > > Alice: received and confirmed, thanks! > > 5 minutes later: > > Carol (not upgraded): Please pay 0.5BTC to my address 3CaroLXXX > Alice: paid, please check > > 1 hour later: > > Carol: it’s not confirmed. Have you paid enough fees? > Alice: ok, I’ll RBF/CPFP it > > 2 hours later: > > Carol: it’s still not confirmed. > Alice: I have already paid double fees. Maybe the network is congested and I need to pay more….. > > Repeat until the lock up period ends. > > So this so-called “softfork” actually made non-upgraded wallet totally unusable. If failed to meet the very important requirement of a softfork: backward compatibility > > More discussion: > https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2017-April/013985.html > > > > > > Depending on how wallets handle reorgs, a non-upgraded wallet may put > > reorg'd spend chains from exits back into an unconfirmed state, when in > > reality they should probably delete them or mark them conflicted in some > > way. This may be an acceptable compromise as the wallet will still see > > the funds as unconfirmed when they really don't exist anymore, but maybe > > unconfirmed is good enough. Users are pretty used to dropping > > non-confirming txs from their wallet, and this is much better than > > legacy wallets seeing there funds as confirmed when they could be > > permanently reorged out at any moment. > > > > Second solution: > > > > Move all exiting outputs to the coinbase. This will enforce a 100 block > > maturity requirement and non-upgraded wallets will be aware of this. > > This is also unacceptable. > > When someone says "Please pay 1 BTC to my address 1ALicExyz”, no one anticipates being paid by a coinbase output. Some exchanges like btc-e explicitly reject coinbase payment. > > Such deterioration in user experience is unacceptable. It basically forces everyone to upgrade, i.e. a hardfork with soft fork’s skin > > > > > > > The first solution might require more implementation, but allows more > > flexibility with the maturity requirement. The second solution is > > possibly simpler, but sticks to a hard 100 block limit. > > > >> 1. Is it acceptable to have massive txid malleability and transaction > >> chain invalidation for every natural happening reorg? Yes: the > >> current spec is ok; No: next question (I’d say no) > > > > Answered above. > > > >> 2. Is locking up exit outputs the best way to deal with the problem? > >> (I tried really hard to find a better solution but failed) > > > > You've probably thought about this more than anyone, so I'd say yes, it > > may be the only way. Painful, but necessary. > > > >> 3. How long the lock-up period should be? Answer could be anywhere > >> from 1 to 100 > > > > I imagine having something lower than 100 would be preferable to users, > > maybe somewhere in the 5 to 15 range. A 15 block reorg on mainnet is > > seriously unlikely unless something strange is happening. A 5 block > > reorg is still pretty unlikely, but possible. The coinbase solution only > > allows for 100 blocks though. > > > >> 4. With a lock-up period, should it allow direct exit to legacy > >> address? (I think it’s ok if the lock-up is short, like 1-2 block. But > >> is that safe enough?) > > > > I think so. Adding a kind of special address probably creates more > > issues than it solves. > > > As I explained above, no legacy wallet would anticipate a lock up. If you want to make a softfork, all burden of incompatibility must be taken by the upgraded system. Only allow exit to a new address guarantees that only upgraded wallet will see the locked-up tx: > > https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2017-January/013490.html > > > >> 5. Due to the fungibility issues, it may need a new name for the > >> tokens in the ext-block > > > > I suppose the market will decide whether that's the case. > > > > It's worth noting, if segwit is not activated on the mainchain, it > > creates a much bigger incentive to use the extension block, and > > potentially ensures that users will have less of a reason to exit. > > > > I think it’s unacceptable if malleability is not fixed in main chain, for 3 reasons: > > 1. a solution is *already* available and tested for > 1 year. > > 2. the deactivation design (which I think is an interesting idea) makes the ext block unsuitable for long-term storage of value. > > 3. LN over main chain allows instant exchange of main coin and xcoin without going through the ugly 2-way-peg process. > > > -- Christopher Jeffrey (JJ) CTO & Bitcoin Menace, purse.io https://github.com/chjj