dangernoodle-io / TaipanMiner / 28241005660

90%

push

github

refactor: unify settling under a single readiness predicate (TA-452) (#579)

* refactor: extract is_ready predicate from wait_until_ready (TA-452)

extract ReadinessSnapshot + is_ready(snapshot, criteria) as a pure,
reusable predicate from the inline logic in wait_until_ready.

- ReadinessSnapshot: normalised view (heap_free, is_mining, hashrate_ghs,
  pool_connected) built by wait_until_ready from live API responses
- is_ready returns (bool, list[reasons]); reasons empty when ready
- logic unchanged: heap floor; mining-board pool+hashrate checks; non-mining
  boards ready on heap+reachability; pool_connected=None not penalised
- wait_until_ready builds snapshot, delegates to is_ready, then applies
  vcore check (requires sensors API, not carried in snapshot)
- soak.py passes criteria to poll so the predicate can be evaluated

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* refactor: monitor warmup uses shared readiness predicate (TA-452)

replace blind warmup timer in monitor.poll with the shared is_ready
predicate from readiness.py.

- poll accepts an optional criteria parameter; when supplied and
  settle_delay > 0, warmup ends when BOTH conditions hold:
  (a) settle_delay floor has elapsed, AND
  (b) is_ready(snapshot_from_sample, criteria) is True
- _device_is_ready() builds ReadinessSnapshot from the Sample (heap from
  heap.internal.free / free_heap fallback; is_mining + hashrate from stats
  via the existing hashrate_ghs() normaliser; pool_connected=None since
  monitor does not fetch /api/pool — hashrate signal covers the gap)
- when criteria is None (legacy callers), warmup is purely time-based
- when settle_delay=0 (default), behaviour is completely unchanged:
  no warmup, detectors active from t=0
- stats field is added to the fetch set when settle is active with criteria
- reboot re-arm logic is unchanged

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* test: is_ready predicate and monitor warmup predicate covera... (continued)

2886 of 3627 branches covered (79.57%)

Branch coverage included in aggregate %.

5385 of 5540 relevant lines covered (97.2%)

149095.43 hits per line