CoinSelectionSimulator

CoinSelectionSimulator copied to clipboard

Would you be willing to test against https://www.coinsayer.com 's api and publish how it performs against the other algos? Happy to throw a $200 bounty (payable in bitcoin) to do...

RHavar

Letting wallets to define failure

2

Instead of guarding insufficient value in simulator, lets wallets themselves define insufficient funds in selectCoins by returning None I needed to tear apart selectCoins and spend for this. TODO: measuring...

karelbilek

Min change as a factor of input cost

1

Instead of just a number, it makes more sense for it to be multiplier of input cost so with larger fees, the min change is larger, with smaller input cost...

karelbilek

Ignore detrimental values

1

This should be added to all "accumulative" wallets, but at this point, I don't care really; since the code shoule be refactored anyway, to avoid all this copy-pasting And in...

karelbilek

As discussed elsewhere. Reduces fees. :)

karelbilek

One last pull request :)

karelbilek

Implementing more "failure modes"

1

"Skip" mode is a bit nonsensical right now; if we "skip", we should also log skipped ones somewhere. Example: the bitcoinjs/coinselect library sort results first by the percentage of skipped...

karelbilek

Helps when there are lots of simulations, or simulations are slow. If there are few/fast, it actually slows down a bit. I am using `MayPar` because if it runs with...

karelbilek

Making BnB deterministic

1

Those little changes make BnB algorithm deterministic. Note that they are still using random shuffles on failure, but it is shuffled every time the same way.

karelbilek

Lower target gets better results

2

(continuation of discussion here https://github.com/bitcoinjs/coinselect/pull/13#discussion_r125195929 ) When I was playing with my JS port, I found out that using "lower" target - only sum of the outputs, without the added...

karelbilek