taker checks for duplicate utxos now in recv_txio() #455
Conversation
| @@ -112,14 +112,19 @@ def recv_txio(self, nick, utxo_list, cj_pub, change_addr): | |||
| log.debug(('recv_txio => nick={} not in ' | |||
| 'nonrespondants {}').format(nick, self.nonrespondants)) | |||
| return | |||
| self.utxos[nick] = utxo_list | |||
| utxo_data = jm_single().bc_interface.query_utxo_set(self.utxos[nick]) | |||
| duplicate_utxos = set(sum(self.utxos.values(), [])).intersection(set(utxo_list)) | |||
There was a problem hiding this comment.
self.utxos.values()is the sets of coins sent by each nick as inputssum(list_of_lists, [])is an idiom for flatten
There was a problem hiding this comment.
effectively [] + list_of_lists[0] + list_of_lists[1] + ...
|
Although the taker could choose to infer that both copies are from a "naughty" maker and treat both as non-responding, I presume there (might?) be a dos vector that way (although a bit theoretical? auth is under encryption). Do we consider that this needs a test that triggers the error? |
|
Yes you're right, it could blame bother makers rather than the one that happened to respond second. That should've be too hard, I'll add it to that branch soonish. Yes a test would be great. I'll add that too (maybe ask you guys on IRC some details about it) |
|
The reason I've left this so long is it's not trivial, there would need to be some records kept in the code for which maker contributed what in fees and utxos, so that both makers can be evicted if they clash. |
AdamISZ
force-pushed
the
develop
branch
3 times, most recently
from
97d2603
to
47479d5
Compare
Please review