I am adding the below to this thread or it will be lost in a wall of text from the chat. I thought this was relevant to keep in mind and store here.
@lehnberg
Jul 31 16:16
@yeastplume so what’s discussed here is correct or not? Reverse transaction building
@hashmap one round-trip, i.e. two single trips afaiu.
jaspervdm
@jaspervdm
Jul 31 16:21
3 steps. but it can be done (and currently is done) with 1 HTTP request (and response)
Yeastplume
@yeastplume
Jul 31 16:24
any transaction can be set up in a number of ways with an arbitrary number of participants, and the transaction has to be passed around to all of them somehow. The current model is the shortest path with a sender sending, which I’d think is the most natural way of approaching it… but all sorts of other permutations are possible
Gary Yu
@garyyu
Jul 31 16:26
@yeastplume mimblewimble/grin#1305
lehnberg
@lehnberg
Jul 31 16:27
But to be perfectly clear: the number of steps (3) and times of interaction (2) are identical regardless of who initiates. Right?
Gary Yu
@garyyu
Jul 31 16:28
And mimblewimble/grin#1292 also ready for merge. but I don’t have Windows PC for test, only test on Linux and Mac. @yeastplume
jaspervdm
@jaspervdm
Jul 31 16:29
@lehnberg yes, basically what needs to happen is that all parties need to know the sum of public nonces to calculate the partial signature. and of course they have to let the other parties know what inputs/outputs they want to add to the tx
Yeastplume
@yeastplume
Jul 31 16:31
right, so everyone needs to pass around their nonces to everyone else, then everyone has to sign with their partial sigs, then someone needs to add them all together to create the final sig. So there are 2 rounds+finalisation necessary no matter what happens