Atomic coin swapping?
6 messages, 22 Sep 2012. Started by Jeff Garzik.
Jeff Garzik (2), Jorge Timón (2), Mike Hearn (2)
Forum URL: https://bitcointalk.org/index.php?topic=112007.0
gmaxwell was talking about colored coins[1] in IRC recently. They are potentially interesting in the context of distributed bonds[2], which I am currently pursuing with pybond[3].
Here is the problem I am trying to solve, does the crowd have an answer?
1. Alice transfers a 1-satoshi colored coin to Bob. 2. Bob transfers 100 BTC to Alice. May be restricted to 1 txout, if that eases implementation details. 3. Steps #1 and #2 happen as an atomic unit, all-or-none. 4. Alice and Bob must both approve this atomic transfer of coins, with appropriate signatures.
Is this possible within the current bitcoin system? As far as I can see, the answer is "no" but maybe I'm missing something.
My best guess to the answer is "possible, but requires a new SIGHASH_* type"?
[1] https://bitcointalk.org/index.php?topic=106449.0 [2] https://bitcointalk.org/index.php?topic=92421.0 [3] https://github.com/jgarzik/pybond
I'm very interested in this. I was expecting transitive/multi-hop transactions (Ripple) with colored coins, and I don't understand why is not possible.
From https://en.bitcoin.it/wiki/Contracts
--- SIGHASH_ALL: This is the default. It indicates that everything about the transaction is signed, except for the input scripts. Signing the input scripts as well would obviously make it impossible to construct a transaction, so they are always blanked out. Note, though, that other properties of the input, like the connected output and sequence numbers, are signed; it's only the scripts that are not. Intuitively, it means "I agree to put my money in, if everyone puts their money in and the outputs are this". ---
Why "Signing the input scripts as well would obviously make it impossible to construct a transaction"? I don't understand that part. I think a new SIGHASH_* type that doesn't pay attention to that "obviously" is needed to achieve what we want.
Say we want the following transaction:
A 1 satoshi -> B 1 satoshi -> C 100 btc -> A
It would be necessary to sign the following:
Inputs: from srcA, from srcB, Outputs: 1 satoshi to destB, 1 satoshi to destC, 100 btc to destA
"from srcC" is not really necessary.
This same scheme can be used for n-hops.
What am I missing?
On 9/22/12, Jeff Garzik <jgarzik@exmulti•com> wrote:
Quoted text (41 lines)
> Forum URL: https://bitcointalk.org/index.php?topic=112007.0 > > gmaxwell was talking about colored coins[1] in IRC recently. They are > potentially interesting in the context of distributed bonds[2], which > I am currently pursuing with pybond[3]. > > Here is the problem I am trying to solve, does the crowd have an answer? > > 1. Alice transfers a 1-satoshi colored coin to Bob. > 2. Bob transfers 100 BTC to Alice. May be restricted to 1 txout, if > that eases implementation details. > 3. Steps #1 and #2 happen as an atomic unit, all-or-none. > 4. Alice and Bob must both approve this atomic transfer of coins, with > appropriate signatures. > > Is this possible within the current bitcoin system? As far as I can > see, the answer is "no" but maybe I'm missing something. > > My best guess to the answer is "possible, but requires a new SIGHASH_* > type"? > > [1] https://bitcointalk.org/index.php?topic=106449.0 > [2] https://bitcointalk.org/index.php?topic=92421.0 > [3] https://github.com/jgarzik/pybond > > -- > Jeff Garzik > exMULTI, Inc. > jgarzik@exmulti•com > > ------------------------------------------------------------------------------ > How fast is your code? > 3 out of 4 devs don\\\'t know how their code performs in production. > Find out how slow your code is with AppDynamics Lite. > http://ad.doubleclick.net/clk;262219672;13503038;z? > http://info.appdynamics.com/FreeJavaPerformanceDownload.html > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development >
Perhaps I missing something obvious about the definition of coloured coins, but this appears to be very simple. Just create a transaction that transfers 300 coins and have an unsigned input connected to the coloured output. send to the owner of the coloured output, they sign it and broadcast. On Sep 22, 2012 11:10 AM, "Jorge Timón" <timon.elviejo@gmail•com> wrote:
Quoted text (98 lines)
> I'm very interested in this. I was expecting transitive/multi-hop > transactions (Ripple) with colored coins, and I don't understand why > is not possible. > > >From https://en.bitcoin.it/wiki/Contracts > > --- > SIGHASH_ALL: This is the default. It indicates that everything about > the transaction is signed, except for the input scripts. Signing the > input scripts as well would obviously make it impossible to construct > a transaction, so they are always blanked out. Note, though, that > other properties of the input, like the connected output and sequence > numbers, are signed; it's only the scripts that are not. Intuitively, > it means "I agree to put my money in, if everyone puts their money in > and the outputs are this". > --- > > Why "Signing the input scripts as well would obviously make it > impossible to construct a transaction"? > I don't understand that part. I think a new SIGHASH_* type that > doesn't pay attention to that "obviously" is needed to achieve what we > want. > > Say we want the following transaction: > > A 1 satoshi -> B 1 satoshi -> C 100 btc -> A > > It would be necessary to sign the following: > > Inputs: from srcA, from srcB, > Outputs: 1 satoshi to destB, 1 satoshi to destC, 100 btc to destA > > "from srcC" is not really necessary. > > This same scheme can be used for n-hops. > > What am I missing? > > On 9/22/12, Jeff Garzik <jgarzik@exmulti•com> wrote: > > Forum URL: https://bitcointalk.org/index.php?topic=112007.0 > > > > gmaxwell was talking about colored coins[1] in IRC recently. They are > > potentially interesting in the context of distributed bonds[2], which > > I am currently pursuing with pybond[3]. > > > > Here is the problem I am trying to solve, does the crowd have an answer? > > > > 1. Alice transfers a 1-satoshi colored coin to Bob. > > 2. Bob transfers 100 BTC to Alice. May be restricted to 1 txout, if > > that eases implementation details. > > 3. Steps #1 and #2 happen as an atomic unit, all-or-none. > > 4. Alice and Bob must both approve this atomic transfer of coins, with > > appropriate signatures. > > > > Is this possible within the current bitcoin system? As far as I can > > see, the answer is "no" but maybe I'm missing something. > > > > My best guess to the answer is "possible, but requires a new SIGHASH_* > > type"? > > > > [1] https://bitcointalk.org/index.php?topic=106449.0 > > [2] https://bitcointalk.org/index.php?topic=92421.0 > > [3] https://github.com/jgarzik/pybond > > > > -- > > Jeff Garzik > > exMULTI, Inc. > > jgarzik@exmulti•com > > > > > ------------------------------------------------------------------------------ > > How fast is your code? > > 3 out of 4 devs don\\\'t know how their code performs in production. > > Find out how slow your code is with AppDynamics Lite. > > http://ad.doubleclick.net/clk;262219672;13503038;z? > > http://info.appdynamics.com/FreeJavaPerformanceDownload.html > > _______________________________________________ > > Bitcoin-development mailing list > > Bitcoin-development@lists•sourceforge•net > > https://lists.sourceforge.net/lists/listinfo/bitcoin-development > > > > > -- > Jorge Timón > > > ------------------------------------------------------------------------------ > How fast is your code? > 3 out of 4 devs don\\\'t know how their code performs in production. > Find out how slow your code is with AppDynamics Lite. > http://ad.doubleclick.net/clk;262219672;13503038;z? > http://info.appdynamics.com/FreeJavaPerformanceDownload.html > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development >
Quoted text (3 lines)
> > Why "Signing the input scripts as well would obviously make it > impossible to construct a transaction"?
As it states in the source code, signatures cannot sign themselves. If scriptSigs were included in the data that is being signed, the act of inserting the newly calculated signature for one input would break the signatures for all the others.
On Sat, Sep 22, 2012 at 5:10 AM, Jorge Timón <timon.elviejo@gmail•com> wrote:
Quoted text (8 lines)
> I'm very interested in this. I was expecting transitive/multi-hop > transactions (Ripple) with colored coins, and I don't understand why > is not possible. > > From https://en.bitcoin.it/wiki/Contracts > > --- > SIGHASH_ALL: This is the default. It indicates that everything about
Yes, SIGHASH_ALL was the crucial piece I was missing.
I have updated the post to add a concrete example with as much detail as necessary for programmers to recreate. Review requested: https://bitcointalk.org/index.php?topic=112007.msg1212356#msg1212356
On 9/22/12, Mike Hearn <mike@plan99•net> wrote:
Quoted text (9 lines)
>> >> Why "Signing the input scripts as well would obviously make it >> impossible to construct a transaction"? > > > As it states in the source code, signatures cannot sign themselves. If > scriptSigs were included in the data that is being signed, the act of > inserting the newly calculated signature for one input would break the > signatures for all the others.
I see. By "input scripts" I thought you meant scriptPubKey rather than scriptSigs. We only need to sign scriptPubKey. Right, Jeff?
Quoted text (1 line)
> Yes, SIGHASH_ALL was the crucial piece I was missing.
Great, there's no need for an additional SIGHASH. I guess you're implementing the simple case you describe first. Do you plan to implement the more general case with n participants instead of only 2 (a Ripple transaction)? That would be awesome.