simon-v / minipos Goto Github PK

When a payment is requested to a recently used address, and the transaction is still in the mempool, there is a chance that the previous payment will trigger a confirmation of the new payment as soon as the invoice is displayed. The reason for this is that the previous transaction hadn't propagated to some of block explorers yet, which causes them to report the old balance, which is then used as the basis for the payment request. An additional problem, that makes solving this issue harder than necessary is the fact that some block explorers return the current time as the timestamp of any unconfirmed transaction, instead of the time when the it was first seen. This makes identifying "new" transactions unreliable.

New addresses (i.e., previously unused or newly generated via an xpub) are not affected.

Possible ways to solve this issue include:

• Requesting block explorer operators to update their software to return a reliable transaction timestamp;
• Keeping track of address balances ourselves;
• Using a fancy wait-for-transaction API call i once saw mentioned;
• Rewriting the explorer polling code to poll bitcoin nodes directly instead, i.e., making an SPV wallet;
• Explicitly forbidding address reuse, except, maybe, for very old addresses.