Hi all,

In some coinjoin implementations inputs are registered first because in
that way, if the user fails or refuses to sign the transaction the input is
banned and denial of service is made a bit more expensive, in the sense
that an attacker needs more and more utxos to keep the attack going.

Your proposal can work if you find an alternative mechanism for mitigating
the DoS attacks or when DoS attacks are not a problem (I can imagine there
are scenarios where it is not really important).

Best

- Lucas


On Mon, May 22, 2023 at 7:53 PM Ben Carman via bitcoin-dev <
bitcoin-dev@lists.linuxfoundation.org> wrote:

> The problem with using ALL|ANYONECANPAY is that you cannot verify
> beforehand that the other inputs are the inputs you want added to the
> transaction.
>
> Some examples of bad things that could happen:
>
>
>    - Coordinator adds its own inputs, you still get your outputs but
>    effectively paid fees for no privacy gain
>    - The inputs added could be paying at a lower fee rate than expected,
>    causing the tx to take longer than what you paid for
>    - Different input types or amount are added so you no longer have the
>    same uniformity across the inputs
>    - (if you care) An input from a sanctioned address is added, causing
>    you to get "tainted" coins.
>
> This is the code in ln-vortex that verifies the psbt on the client side if
> you are curious
>
>
> https://github.com/ln-vortex/ln-vortex/blob/master/client/src/main/scala/com/lnvortex/client/VortexClient.scala#L616
>
>
> Best,
>
> benthecarman
>
> ------------------------------
> *From:* bitcoin-dev <bitcoin-dev-bounces@lists.linuxfoundation.org> on
> behalf of alicexbt via bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org>
> *Sent:* Monday, May 22, 2023 7:51 AM
> *To:* Bitcoin Protocol Discussion <bitcoin-dev@lists.linuxfoundation.org>
> *Subject:* [bitcoin-dev] Coinjoin with less steps using ALL|ANYONECANPAY
>
> Hi Bitcoin Developers,
>
> I recently experimented with different sighash flags, PSBTs and realized
> ALL|ANYONECANPAY could be used to reduce some steps in coinjoin.
>
> Steps:
>
> - Register outputs.
> - One user creates a signed PSBT with 1 input, all registered outputs and
> ALL|ANYONECANPAY sighash flag. Other participants keep adding their inputs
> to PSBT.
> - Finalize and broadcast the transaction.
>
> Proof of Concept (Aice and Bob): https://gitlab.com/-/snippets/2542297
>
> Tx:
> https://mempool.space/testnet/tx/c6dd626591dca7e25bbd516f01b23171eb0f2b623471fcf8e073c87c1179c492
>
> I plan to use this in joinstr if there are no major drawbacks and it can
> even be implemented by other coinjoin implementations.
>
> /dev/fd0
> floppy disk guy
>
> Sent with Proton Mail secure email.
> _______________________________________________
> bitcoin-dev mailing list
> bitcoin-dev@lists.linuxfoundation.org
> https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev
> _______________________________________________
> bitcoin-dev mailing list
> bitcoin-dev@lists.linuxfoundation.org
> https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev
>