WPPilot 1.6.3 — Stranded queue batches retry themselves
One fix, the other half of a problem 1.6.2 started on. No new abilities, no permission changes, and existing connections keep working.
A stranded batch retries itself. A tab that picked up a Block Editor Queue batch and then went away — closed, navigated off, or left in the background long enough for the browser to throttle it past the five-minute lease — left that batch marked running, with its item leased to a runtime that no longer existed. No other tab would take it, so it sat there until an administrator cancelled it and queued the work again.
The recovery for exactly that case was already in the plugin and already correct: it fails the batch with a message naming the abandoned tab, which makes it claimable again. Nothing ever reached it. The scan that calls it looked only at ready and failed batches, and a stranded batch is neither — so the one state the recovery exists for was the one state it could not see. Running batches are scanned now.
A batch another tab is actively working is still left alone: its lease is valid, so it is refused as not claimable and skipped, exactly as before.
1.6.2 stopped WPPilot putting a batch into that state from a malformed request. This is getting out of it when something else did.
Install wppilot.zip below, or update from wp-admin. Tested against WordPress 7.0.3 on PHP 8.3.