Checking the encoding before the version byte keeps a wallet from naming the wrong chain

Any decoder that checks integrity and then reads fields out of the same bytes has to pick an order. The post picks integrity first. A testnet address with one character mangled fails its checksum while its version byte still says testnet, and both statements are true ^[11]. Naming the network from a payload that failed its own check means reporting a field you have no reason to trust, so a corrupted testnet address comes back as a bad checksum ^[12]. "Confidently wrong output is worse than a vague rejection, and precedence is the only place you get to prevent it," the author wrote ^[13]. The other ordering ends in the ticket the post imagines: "it told me this was a Bitcoin address, it is not." ^[14]

The guard is what makes this cheap to adopt. The result is falsy when the address is not acceptable, so `if not result` reads exactly like the boolean check it replaces, and the reason, format and network are attached when a caller wants them ^[10]. The published example tests `Reason.BAD_CHECKSUM`, then `Reason.WRONG_NETWORK`, then falls through to a message saying the format is not supported here ^[9]. Retry advice goes only to the first branch, the one failure where copying again has a good chance of working ^[5]. The wrong-network branch names the network the address belongs to and stops, since re-copying sends the user back to the same clipboard for the same string ^[6].

Without a reason code, every call site invents its own message out of the absence of information, and the post's description of the outcome is three screens telling the same user three different half-truths ^[8].

The split that keeps callers stable is between an open set and a closed one. Address formats are open: `bitcoin-p2pkh`, `bitcoin-p2sh`, `bitcoin-p2wpkh`, `bitcoin-p2tr`, their testnet variants, and EIP-55 hex for EVM chains ^[16]. Failure reasons are closed at three: the encoding did not verify, the address belongs to another network, the version is not one you serve ^[17]. Counted from the post's own lists, nine format names sit against three reasons ^[20]. New chains keep arriving in the first set while the second has barely changed ^[18]. Callers branch on the closed set, so adding a chain does not ripple into them ^[19].

For the three-way split to pay for itself, your rejections have to land in all three buckets. The post reports no error rates and no ticket counts for the three classes ^[21]. If nearly all your failures are truncated pastes, one message and a retry prompt recovers most of the value, and the other two branches are insurance you carry for correctness. The third reason has a second job anyway: an address type you do not serve yet is often not the user's fault, and the post counts it separately from typos as a roadmap signal ^[7].