[EXPERIMENT] Modify taker to match maker change outputs.

[AdamISZ]

Here "Experiment" means DO NOT USE on mainnet as it is just a simple demonstration of concept and is certainly not guaranteed to be safe, and I have not even checked its safety myself, let alone anyone else.

This is an experimental demonstration. Intended to
show that it is pretty easy to have a Joinmarket
taker choose a different transaction structure,
specifically in the output set, without makers
knowing about it (thus no maker code change).

[AdamISZ]

See here for an example running this code:

https://0bin.net/paste/aPXntZiF#I1TAg7BsDDFq4xL2uDOZ1elUirkZrauQo0z1kjoA7wC

[AdamISZ]

Obviously the intention here is not only to show that it is possible, but that nicely, it only affects one source file.

Whether we actually want to do stuff like this is very debatable. If a person with very good understanding of how the code operates chose to do it on their own, I'm sure makers wouldn't mind the extra income ...

[kristapsk]

I now think this needs more reviews and is potentially good idea. Needs rebase.

[kristapsk]

BTW, just remembered, I saw actual mainnet transaction in a wild some time ago where it looked that somebody is running this code.

[AdamISZ]

BTW, just remembered, I saw actual mainnet transaction in a wild some time ago where it looked that somebody is running this code.

That's cool. I guess we just can't know if it's in Joinmarket or not, though, right.

Not particularly motivated to do the hard work of considering all the edge cases of this idea, and also creating more (very hard to understand!) options for users. I guess some kind of "pro mode" where users could make more flexible transaction structures could make sense, it just seems like a really hard thing to build and release, and feel confident that it's safe.

[kristapsk]

I guess we just can't know if it's in Joinmarket or not, though, right.

I can neither confirm nor deny that I might know or not know that.

guess some kind of "pro mode" where users could make more flexible transaction structures could make sense, it just seems like a really hard thing to build and release, and feel confident that it's safe.

Some fork (alternative packaging) of JM in style of Knots vs Core would be solution, there are sometimes other experimental or controversial things some people might want to use (or risk using), like #1481, so different unmerged PRs could be added. But somebody needs to do those releases after official JM releases.