It sounds like, by deciding to make a charge based on what amounts to a temporary cache of the balance, there has always been a race condition in the billing co

It sounds like, by deciding to make a charge based on what amounts to a temporary cache of the balance, there has always been a race condition in the billing code? I would think it would have always been necessary to successfully update the balance before proceeding with the charge in order to avoid double charges. I can see how this sort of situation is hard to predict and test for, but that means it needs to be designed and implemented very carefully. $0->charge card just isn't a reasonable approach at all.