tests · sds · Tangled

lance.blue / sds

bayes for days

sds tests

32 files

at claude/expected-crits

@permadeath.com · test(battle-armor): every design in the library through the observation · 1mo ago

feat: match harness with confidence intervals

test(training): a fixture where one machine of two did all the work

test_baseline.py

fix(training): one explore rate drives both phases, and displacement is not a fault

test_battle_armor.py

test(battle-armor): every design in the library through the observation

feat(candidates): the sweep generates every reachable stand, and the wire carries what it needs

test_compressed_logs.py

feat(harness): read decision logs compressed

feat(harness): pin each seat to a physical core of its own

fix(harness): name the runner image after its Dockerfile

test_correlate.py

feat(hierarchy): incoming damage, optionality, scored deployment, and the firing superset

test_crossing.py

feat(scenarios)!: drop River Delta 1 from the map pool

test_explain.py

feat(candidates): the sweep generates every reachable stand, and the wire carries what it needs

test_figures.py

docs(tactics): what the matrix said, and a figure for every tactic

test_firingaudit.py

feat(observability): the decision log, and the tools that read it

feat: match harness with confidence intervals

test_labeldoc.py

docs(training): LABELS.md, every label over one worked match

test_labelrank.py

fix(training)!: read the engagement gap from one side

feat(observability): the decision log, and the tools that read it

test_los_corpus.py

test(los): 2383 MegaMek cases the Rust port must match

test_loscache.py

test(observability): a reading that went backwards is a cleared cache

feat(scenarios)!: drop the four boards that carry a bridge

fix(harness): swap sides per scenario, not per game

test_movestates.py

feat(hierarchy): incoming damage, optionality, scored deployment, and the firing superset

fix(harness): stop a run whose bot answered nothing

test_runreport.py

feat(observability): one report per run, in one place

test_runstats.py

fix(harness): the by-round table averaged totals, not turns

test_scenario.py

feat(scenarios): draw the two sides independently

feat(observability): what each feature actually reported

fix(candidates)!: ask MegaMek whether a prone Mek may stand, and keep no rule

feat(scenario): battle value tiers, rolled crews, and the tools to correlate a result

test_tactics.py

feat(tactics): the vocabulary, and the tool that measures it

feat(training): --drop-displaced, and what it moves

feat(hierarchy): incoming damage, optionality, scored deployment, and the firing superset