multi: Keep trying init with the same secret. #3269
Conversation
Data from tx history is not available always e.g. electrum. Am seeing some refund issues that happily seem to resolve. |
|
For a quick fix, it looks like just failing and waiting for refund may be cleanest, but a waste and will hurt the rep of the user. I guess as a backup for nonce's getting in a twist it's preferable to the ghost init that cant easily be tracked down and requires numerous steps to refund. |
JoeGruffins
force-pushed
the
keepsecretwhenfail
branch
2 times, most recently
from
6a01b20 to
f635d80
Compare
|
This is alot of extra things to do, but note that it only triggers if the first attempt at init failed. |
|
Also contains marton fixes from #3082 Maybe one of them makes the nonce issue better anyway. |
srry, I guess the check also triggers first try if taker. |
JoeGruffins
force-pushed
the
keepsecretwhenfail
branch
from
f635d80 to
3e57287
Compare
| // InitChecker is an account wallet that checks the status of a swap init. | ||
| type InitChecker interface { | ||
| // AlreadyInitialized returns if the swap was already initialized and | ||
| // some data needed to interact with the contract later. | ||
| AlreadyInitialized(assetVersion uint32, contract *Contract) (initialized bool, zeroHash, contractData []byte, err error) | ||
| } |
There was a problem hiding this comment.
Could Swap just check whether or not the swap is already initialized, the same way how we check isRedeemable and isRefundable, and if it's already initialized then return without error as if the transaction was just sent. Then this interface and the checkInitialized function in trade.go would not be needed. The server just needs to be updated to not require the transaction hash.
There was a problem hiding this comment.
I didn't want to update the server as thats a bigger change. But we can do that. May not have time to work on this until week after next.
There was a problem hiding this comment.
It is quite a big change.. I think let's just go with revoking for now.
There was a problem reported with an init was sent twice. This was due to a nonce issue. This pr ignores that and attempts to prevent a second init being sent regardless.
It currently doesn't work because I'm unsure how to get the txid of the swap and server wants that. I think @martonp said something about this being possible before, by examining the data from tx history, but unsure. Or we could just fail, which seems like a waste. Or change the server logic.