This comes up a lot around here :) The idea behind the paragraph you are quoting is that for in-app billing to be secure, you need to verify transaction signatures. Those are signed with a private key, associated with your developer account. The key resides on Google's servers, so it's fairly safe to assume that no one else can sign data with it. To verify it you need your public key, which you can copy from the developer console. If someone replaced it in your app, they could fool it to accept in-app billing transactions from unauthorized sources, because if they plant the public key, they probably also control the corresponding private key. In practice however, it is far easier to simply modify your code in the right places to always return true for isLicensed()
, hasItem()
or similar methods you might have and no one does this.
The best way to protect the key is, of course, not to have the key in your app at all. Move all of the transaction validation logic to your server, and use HTTPS to connect to it. Properly validate the certificate chain to ensure you are talking to your own server(s). Otherwise, someone might mess around with DNS and fool your app to connect to their own servers. A similar attack against iOS purchasing was announced a couple of weeks ago.
The next best thing is to somehow obfuscate the key, and have it included in your app. This has the advantage that you don't need a server, but the disadvantage is that if someone is determined enough they will figure it out, since they can always reverse the byte code of your app. So your best bet is to come up with your own original way to do it that doesn't show up on public forums :) To make it a bit harder, you can implement the validation part in native code, which is harder (but not impossible) to analyze. Still, as mentioned above, patching byte code in the right places is far easier than trying to replace the public key, so that is what most crackers will do.