Bitcoinjs-lib: How to calculate fees in segwit multi-sig
Hello
I'm doing a multi-sig transaction and i'm not sure how i should calculate the transaction fee.
I want 1 party to sign a tx and pass the transaction to another party to sign and then broadcast.
If i understand this correctly, i need to call getVirtualSize to calculate the TX Fee, However the witness field isn't completed yet. Because i'm waiting for the other party to sign the tx
rbndg
All 16 comments
https://gist.github.com/dabura667/1bb77d63d38bfd99a0ce453db74e0115~~
(Edit: Gist is gone, here's a copy https://gist.github.com/junderw/b43af3253ea5865ed52cb51c200ac19c)
// 2-of-3 multisig with 3 inputs sent to 2 P2PKH outputs.
getByteCount({'MULTISIG-P2SH-P2WSH:2-3': 3}, {'P2PKH': 2})
// 373 virtual Bytes (1/4 weight)
dabura667
on 8 Jan 2018
remove P2SH- if you want to measure for native segwit
dabura667
on 8 Jan 2018
This assumes compressed pubkeys in all cases.
It also assumes the S value is always 32 bytes length and the R value of each signature is always 33 bytes (with the 0x00 padding) so that we estimate the maximum possible size. This will usually create a 1 weight (4 weight for non-segwit) variance per signature contained.
dabura667
on 8 Jan 2018
Example:
https://www.smartbit.com.au/tx/c8e32240ed79baa07564b30216b9d148f41ef9141b96ff6ebb0e6dd08f49ac96
getByteCount({'P2PKH': 77}, {'P2PKH': 2})
// 11474
but if you look at the API response by clicking the </> button, it says 11433... if we search the page for "asm": "3045022100 (Which means the r value was 33 bytes for the input) we get 36 hits... 77 inputs - 36 == 41 inputs with a 32 byte r value...
11474 - 11433 == 41 exactly. So we overestimated by 41 bytes because there were 77 signatures and 53% of them were 32 byte r values. (s values are always 32 or less, because s values must be low-s due to recent standardness rules)
dabura667
on 8 Jan 2018
@dcousens I wonder if there's a way we can include this function as a prototype for Transaction or TransactionBuilder..... hmmm....
dabura667
on 8 Jan 2018
@dabura667 I've been working on a size estimation helper I was eventually going to push here https://github.com/blocktrail/blocktrail-sdk-nodejs/blob/master/lib/size_estimation.js It's a little heavier, but think it's pretty clear and avoids the magic numbers that many size estimators happen to have.
This assumes compressed pubkeys in all cases.
This is a curious limitation :P technically you need to know the private key and all other secret information to do size estimation. Hashlocks take whose size can be variable.. Interesting challenges in terms of exposing all these degrees of freedom in a nice way. (arguments? options? notice the compressed arg that gets passed from estimateUtxo to estimateP2PKHStackSize)
afk11
on 8 Jan 2018
my colleague made this a while back https://github.com/you21979/node-bitcoin-util-fee/blob/master/index.js
Doesn't support segwit... but it's very... verbose.
dabura667
on 8 Jan 2018
... very verbose.
sneurlax
on 14 Jan 2018
@afk11 does size_estimation.js allow estimation of P2SH-P2WSH ?
rbndg
on 20 Jan 2018
Yes it does.
junderw
on 20 Jan 2018
@dabura667 you should incorporate https://gist.github.com/dabura667/1bb77d63d38bfd99a0ce453db74e0115~~ into https://github.com/bitcoinjs/coinselect (maybe? if relevant)
(Edit: new link https://gist.github.com/junderw/b43af3253ea5865ed52cb51c200ac19c)
Or at least, help us make https://github.com/bitcoinjs/coinselect better to avoid needing your gist :+1:
dcousens
on 20 Jan 2018
@junderw I don't believe it does,
In the estimateUtxo function, the array of allowed script contains only:
var allowedTypes = [types.MULTISIG, types.P2PKH, types.P2PK]
However correct me if i'm wrong :)
rbndg
on 24 Jan 2018
@dabura667 i'm going to use your gist, It's quiet handy and easy to understand
rbndg
on 25 Jan 2018
the gist is no longer available -- anyone know where one can access it or a similar resource ?
tipu
on 19 Dec 2018
@rbndg it does support P2SH-P2WSH - the allowedTypes vars refer to the types of scripts which can be nested _inside_ a P2SH-P2WSH script
afk11
on 19 Dec 2018
@tipu I have a copy on my gist: I'll replace the links.
https://gist.github.com/junderw/b43af3253ea5865ed52cb51c200ac19c
junderw
on 19 Dec 2018
Related issues
dcousens
路
3Comments
LeonYanghaha
路
3Comments
Beardcoding
路
3Comments
ghost
路
3Comments
tuyennvtb
路
3Comments
Most helpful comment
https://gist.github.com/dabura667/1bb77d63d38bfd99a0ce453db74e0115~~(Edit: Gist is gone, here's a copy https://gist.github.com/junderw/b43af3253ea5865ed52cb51c200ac19c)