bug(replicator): replicator Handles message, deletedMessages in event out of order.

[TecSong]

What is the bug?
A concise, high level description of the bug and how it affects you
After modifying the profile, the data in the replicator pg table record is not new, but old.

Others also have the same problem, please see: #1603

How can it be reproduced? (optional)
Background: I run a hub and a replicator, and the replicator syncs data from the hub to pg.

When I change the display name in warpcast, the hub will generate an event, the details are as follows:
{ "type": "HUB_EVENT_TYPE_MERGE_MESSAGE", "id": 395450384842752, "mergeMessageBody": { "message": { "data": { "type": "MESSAGE_TYPE_USER_DATA_ADD", "fid": 19062, "timestamp": 96545504, "network": "FARCASTER_NETWORK_MAINNET", "userDataBody": { "type": "USER_DATA_TYPE_DISPLAY", "value": "yikosong" } }, "hash": "0x96240ec645f6de86fd77ce363b4ecb3bb73f7246", "hashScheme": "HASH_SCHEME_BLAKE3", "signature": "4wcqCGUGgD2tHfG9iQsrUqwZJN/86PkXEPOWWFehq7DK9nn6njIHjik+91lKR1+EyaJmfUDFD2DOEiP4vYnuAw==", "signatureScheme": "SIGNATURE_SCHEME_ED25519", "signer": "0x11637e5ee0cc28ec00d58a8bdade374e22a796a73334a69b1f42dc37b3b9b518" }, "deletedMessages": [ { "data": { "type": "MESSAGE_TYPE_USER_DATA_ADD", "fid": 19062, "timestamp": 96540291, "network": "FARCASTER_NETWORK_MAINNET", "userDataBody": { "type": "USER_DATA_TYPE_DISPLAY", "value": "yikosong.eth" } }, "hash": "0x30ba0804cb09a353a83b6cc4d23bcfa7400e709e", "hashScheme": "HASH_SCHEME_BLAKE3", "signature": "+6Ccg+vh6QbgCWQMIxQLEfGRQLPAfwigp7h3K2A5qfD+9uwi8WmYTjxT/vHeJ8mri/SwWG3ZYjEU/iKWqx++Bw==", "signatureScheme": "SIGNATURE_SCHEME_ED25519", "signer": "0x11637e5ee0cc28ec00d58a8bdade374e22a796a73334a69b1f42dc37b3b9b518" } ] } }

• It contains message and deletedMessages, and the replicator handles the event, but in the log, it processes the message first, and then the deletedMessages, and the update operation of the user_data table causes the data that should be deleted to overwrite the latest data. I found the location in the replicator repo where the resolution handles the event, as follows:

  const { message, deletedMessages } = hubEvent.mergeMessageBody;
  await executeTx(db, async (trx) => {
    await mergeMessage(message, trx, log, redis);

    for (const deletedMessage of deletedMessages) {
      await deleteMessage(deletedMessage, trx, log, redis);
    }
  });
}```
file path: hub-monorepo/apps/replicator/src/processors/index.ts

**Additional context (optional)**
So I'd like to ask why (message->deletedMessages) is handled in this order, whether it's a bug, and whether it can be fixed in a timely manner, this problem has existed for a long time

[sds]

Thanks for the report, sorry for the delay in response.

The old replicator at this point has a number of issues. We recommend using the Shuttle project instead.