Teknium
9af6e9fc8b
Based on the earliest batching proposal by BrunoBza (#104686) and JoaoMarcos44 structured correction (#104703). Slim redo into topical siblings and shared TUI poller/post-turn routing. Reported by Xipong (#104671). Local shell-to-loopback probes demonstrate 12 to 1 turn dispatches; suites await the campaign test lock. Co-authored-by: joaomarcos <joaomarcosdias444@gmail.com>
22 lines
1010 B
Python
22 lines
1010 B
Python
"""Completion backlogs preserve results without multiplying autonomous turns."""
|
|
from evals.completion_backlog_probe import probe
|
|
|
|
|
|
def test_ready_completions_share_one_turn_across_interactive_routes(tmp_path):
|
|
for surface in ("cli", "poller", "post-turn"):
|
|
for scenario in ("backlog", "single"):
|
|
result = probe(surface, scenario, tmp_path / surface / scenario)
|
|
assert result["wire_turns"] == 1, result
|
|
assert result["all_payloads_preserved"], result
|
|
if scenario == "single":
|
|
assert result["single_exact"], result
|
|
|
|
|
|
def test_consumed_or_foreign_completions_never_start_a_turn(tmp_path):
|
|
for surface in ("cli", "poller", "post-turn"):
|
|
for scenario in ("consumed", "foreign"):
|
|
result = probe(surface, scenario, tmp_path / surface / scenario)
|
|
assert result["wire_turns"] == 0, result
|
|
if scenario == "foreign":
|
|
assert result["queue_remaining"] == result["children"], result
|