Stop treating a queued write as a failed one

[mathieuletyrant]

One-character fix: an if that should be an elif.

In set_capability_value, the execution poller tests state 1 in an if of its own, then opens a fresh if/elif chain on state 2:

if execution_state == 1:
    _LOGGER.info("Execution_state waiting execution")

if execution_state == 2:        # new chain
    ...
elif execution_state == 3:
    completed = True
    break
else:                            # state 1 lands here
    _LOGGER.info("Execution_state error")
    break

So state 1 — waiting execution, the normal first answer to a write — logs waiting execution, falls into the else of the second chain, logs Execution_state error and breaks out.

completed stays False, so the optimistic capability["value"] = value after the loop never runs, and the entity keeps its old value until the next scheduled poll, up to a minute later. Nothing had failed: the write went through.

Closing the chain makes a queued execution fall through to the retry sleep, like an in-progress one:

queued then ok    waiting -> waiting -> in progress -> completed
immediate ok      completed
real error        waiting -> ERROR

The giveaway in the logs is an Execution_state waiting execution followed immediately by Execution_state error on a write that visibly succeeded.

Running on my installation: entities now reflect a change as soon as the write completes, instead of waiting for the next poll.

Independent of #160, #161, #162 and #163.

🤖 Generated with Claude Code