Reactions Interoperability Guide

How to implement emoji reactions compatible with MeshCore One.

Message Hashing

Reactions target messages by a hash of their content. Every node that received the original message can compute the same hash independently.

1. Concatenate the message's UTF-8 text bytes with the sender's original timestamp as a little-endian UInt32
2. SHA-256 the result
3. Take the first 5 bytes (40 bits)
4. Encode as 8 characters of Crockford Base32

SHA-256( UTF-8(text) + LE(UInt32(senderTimestamp)) )
  → first 5 bytes → Crockford Base32 → "b45pc4ek"

Use the sender's original timestamp, not your local receive time in order for all nodes to agree on the hash.

Why Crockford Base32?

Crockford Base32 maps 5 bytes to exactly 8 characters using only alphanumerics — no special characters that could conflict with the wire format delimiters (@, [, ], \n). Hex would need 10 characters for the same data. Base64 fits in 8 but uses +, /, =. The parser is case-insensitive and normalizes common substitutions (O → 0, I/L → 1).

Wire Format

Reactions are sent as regular mesh messages with a specific text format.

Channel (includes target sender to disambiguate identical messages from different users):

{emoji}@[{targetSenderName}]\n{hash}

Example: 👍@[AlphaNode]\nb45pc4ek

DM (two-party, sender is unambiguous):

{emoji}\n{hash}

Example: 👍\nb45pc4ek

Receiving Reactions

When you receive a message, check if it matches the reaction format before treating it as a regular message.

If it's a reaction: look up the target message by hash. If the target hasn't arrived yet (out-of-order delivery is common on mesh), queue the reaction and match it when the target appears.

If it's a regular message: compute its hash and index it so future reactions can find it. Also check your pending queue for reactions already waiting on this hash.

Deduplicate by (targetHash, senderName, emoji) — a node may relay the same reaction more than once.