A stalled request now names what it was actually waiting for

Parts of the platform hold incoming work briefly while they start up — a workspace loading its data, a node type being compiled. A request that arrives in that window is parked, and released the moment start-up finishes. If it is never released, the sender is told so rather than left waiting.

That explanation is the subject here, and it could be empty or wrong.

What was going wrong

The list of things a component is still waiting for is a live list: the moment one of them is ready, it is taken off. The explanation was written at the end — when the request gave up — and it read that live list then. So it did not describe what had held the request; it described what the component happened to still be waiting for at the moment of the complaint.

In the worst case everything had finished by then and the list was empty, which does not read as "no information". It reads as a statement: nothing was holding it. The exact opposite of what happened. That answer went out 364 times over six days before it was noticed, and in one form the sentence beside it — "without ever finishing start-up" — was plainly untrue: start-up had finished, and the request simply never got its turn.

What you get now

Two facts, labelled as two facts:

The same pair now goes to whoever was waiting, not only into the log — a caller is often in a different process and can see none of the component's own diagnostics.

What did not change

No wait got longer, nothing retries, and nothing is now hidden. Every report keeps the level and the detail it had; the part that was missing is added beside it.

Reconnecting…
The connection to the server was interrupted. Trying to restore it…
Trying again…
The connection could not be restored. Reloading the page…
The server was updated. Reloading the page to pick up the latest version.