There is no assurance that what you actually want to happen will indeed happen. But the numbers you asked for will steadily improve.

So. Let’s say you’re a manager and you’re running this publishing function. Your bosses get together and decide that responsiveness is an important metric, so they’ll be measuring how long it takes for your group to ship a product once it’s been ordered. They give you some target to meet. Start missing that target, and you’re looking at lost bonuses and maybe a lost job.

Then something bad happens. It’s out of your control, but no one really cares about that. There’s a problem and there’s no way you can meet your targets. What do you do?

One thing you could do is stop accepting orders entirely. No product ships and no targets are missed. You may have just saved your job.

Of course, no one is happy with that solution. Not you, not your bosses, not the clients. But it’s an obvious response to the reward system that has been implemented. And the behaviour won’t change unless the system changes.

The first is to add a new revision with trivially modified data to the existing Diaspora project. This retains the old project ID which is important because that’s encoding the URL to buy the product and exists in several places, few under my control.

The second is as you describe, though again, unless I anticipate the nature of the print error and correct it with a change in data, there’s no reason to expect a new result.

Have you tried simply setting it up a second instance of the product, to bridge the gap? Or is that simply unacceptable to you?