Oh I see you just keep getting your coin design more complicated as you get peer review. No wonder you haven't implemented a single line of code yet because it already has prohibitive difficulty to implement correctly. I wish you luck, but i wouldn't bet a penny that your coin would ever be implemented. No to say working.
That's a good idea actually, but I was talking about if the receiver comes online after the block of interest has been pruned from the mini-blockchain. They will be able to verify the money went to them because of the address database (which I now prefer to call the account database), but if they come online too late they wont be able to identify the sender or get the message. They could as you mentioned just get the block again from the sender if the sender saved it and sender node was still online. Or they could ask around in the hope another node was still holding onto the block they were after. But they would really only need to bother with this to retrieve detailed information about the transaction, it's always possible to confirm received funds without needing the block in which the transaction happened.
I don't think miniblock lifetime would be so short that you will get this kind of problems often. Anyway there is probably going be some person who will store entire blockchain for preservation purposes and would send you needed blocks for small fee. And even where the isn't one there should be multiple copies of all blocks on network because all transaction senders from a block would likely keep it.
