configuration/reason/data/surface.data.json
configuration/reason/data/surface.data.json is a file in GovLab Context. 594 lines of code and 0 definitions.
{
"check": {
"by": [
"the techniques each surface lists, whose results the surface's predicate and evidence decide",
"the covered-surfaces check, which validates every surface a module declares covered"
],
"population": "every module that declares the surface covered, and the code under that module",
"freshness": "a verdict stands until the module's code, its declaration or the surface changes",
"refusal": "the covered-surfaces check fails a module that declares an unknown surface; a surface whose evidence is empty stays unknown, never pass",
"observation": "the runtime techniques among the surface's list, which locate failures without certifying their absence",
"evidence": "fires-and-accepts: a suite plants a surface with no techniques, an empty verdict domain, a blank predicate and a colliding cell, and the bundled surfaces validate clean",
"authority": "the evidence the surface's techniques produce, which the verdict is decided from, so an empty evidence set leaves it unknown"
},
"records": [
{
"id": "semantic-correctness",
"dimension": "meaning",
"lens": "meaning",
"fit": "An output is correct when it signifies what the input asked for, so the surface compares the meaning of the result with the meaning expected.",
"failureModes": [
"wrong value",
"incorrect algorithm",
"wrong computation"
],
"techniques": [
"unit-testing",
"property-based-testing",
"differential-testing",
"assertion-checking"
],
"predicate": {
"type": "equivalence",
"expression": "output(input) = expected(input)",
"grounds": ["ver-ground-truth"]
},
"invariant": "correct-outputs",
"evidence": {"required": true,
"source": "test-result",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "functional-correctness",
"dimension": "function",
"lens": "behavior",
"fit": "A workflow fulfils its role through the actions it takes, so the surface reads the role from the transitions the system actually performs.",
"failureModes": [
"invalid workflow",
"missing transition",
"incorrect business rule"
],
"techniques": [
"integration-testing",
"model-checking"
],
"predicate": {
"type": "invariant",
"expression": "transition in allowedTransitions",
"grounds": ["ver-ground-truth"]
},
"invariant": "correct-state-evolution",
"evidence": {"required": true,
"source": "test-result",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "state-correctness",
"dimension": "state",
"lens": "sequential",
"fit": "A state is valid only relative to the one before it, so the surface checks conditions in the order they occur.",
"failureModes": [
"invalid state",
"broken state transitions",
"violated lifecycle rules"
],
"techniques": [
"assertion-checking",
"property-based-testing"
],
"predicate": {
"type": "temporal-order",
"expression": "state[t+1] in next(state[t])",
"grounds": ["ver-ground-truth"]
},
"invariant": "valid-state-transitions",
"evidence": {"required": true,
"source": "test-result",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "interface-correctness",
"dimension": "structure",
"lens": "structure",
"fit": "A contract is an arrangement of fields and types, so the surface compares the shape of a payload with the shape the contract declares.",
"failureModes": [
"invalid input/output shape",
"schema mismatch",
"contract violation"
],
"techniques": [
"runtime-validation",
"contract-testing"
],
"predicate": {
"type": "schema",
"expression": "payload models declaredSchema",
"grounds": ["ver-ground-truth"]
},
"invariant": "valid-interfaces-and-contracts",
"evidence": {"required": true,
"source": "test-result",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "interaction-correctness",
"dimension": "relation",
"lens": "relation",
"fit": "An interaction is the connection between two components, so the surface checks each request against the response its counterpart owes.",
"failureModes": [
"wrong API response",
"incorrect UI behavior",
"invalid component communication"
],
"techniques": [
"end-to-end-testing",
"contract-testing",
"integration-testing"
],
"predicate": {
"type": "equivalence",
"expression": "response = contract(request)",
"grounds": ["ver-ground-truth"]
},
"invariant": "correct-interactions",
"evidence": {"required": true,
"source": "test-result",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "temporal-correctness",
"dimension": "time",
"lens": "time",
"fit": "Execution time is a duration measured across a run, so the surface compares elapsed time with the deadline.",
"failureModes": [
"timeout",
"deadline miss",
"stale data",
"starvation"
],
"techniques": [
"load-testing",
"monitoring",
"tracing"
],
"predicate": {"type": "bound",
"expression": "elapsed <= deadline",
"grounds": ["ver-ground-truth"]},
"invariant": "acceptable-execution-time",
"evidence": {"required": true,
"source": "measurement",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "concurrency-correctness",
"dimension": "behavior",
"lens": "time",
"fit": "Concurrent code misbehaves only in particular interleavings, so the surface examines what the system does across the orderings time allows.",
"failureModes": [
"race conditions",
"deadlocks",
"livelocks",
"ordering failures"
],
"techniques": [
"stress-testing",
"deterministic-replay"
],
"predicate": {
"type": "invariant",
"expression": "for-all interleavings: linearizable(history)",
"grounds": ["ver-ground-truth"]
},
"invariant": "safe-concurrent-behavior",
"evidence": {"required": true,
"source": "runtime-observation",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "memory-correctness",
"dimension": "composition",
"lens": "change",
"fit": "A leak shows as the live set growing in content over a long run, so the surface tracks what memory holds as the run develops.",
"failureModes": [
"memory leaks",
"excessive allocation",
"fragmentation",
"retention bugs"
],
"techniques": [
"heap-analysis",
"profiling"
],
"predicate": {
"type": "bound",
"expression": "liveSet bounded as t grows and retained delta approaches 0",
"grounds": ["ver-ground-truth"]
},
"invariant": "controlled-memory-usage",
"evidence": {"required": true,
"source": "measurement",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "resource-correctness",
"dimension": "composition",
"lens": "behavior",
"fit": "A handle leak is a resource the system acquires and never gives back, so the surface pairs each acquisition with its release.",
"failureModes": ["file/socket/connection/handle leaks"],
"techniques": [
"monitoring",
"assertion-checking"
],
"predicate": {
"type": "absence",
"expression": "acquired = released (no leaked handles)",
"grounds": ["ver-ground-truth"]
},
"invariant": "acceptable-resource-consumption",
"evidence": {"required": true,
"source": "runtime-observation",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "performance-correctness",
"dimension": "scale",
"lens": "statistical",
"fit": "Performance depends on load, and one request says little about it, so the surface reads latency and throughput as distributions over many requests.",
"failureModes": [
"slow algorithms",
"excessive CPU",
"high latency",
"throughput degradation"
],
"techniques": [
"profiling",
"load-testing"
],
"predicate": {
"type": "bound",
"expression": "latency_p99 <= budget and throughput >= floor",
"grounds": ["ver-ground-truth"]
},
"invariant": "acceptable-execution-time",
"evidence": {"required": true,
"source": "measurement",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "reliability-correctness",
"dimension": "probability",
"lens": "anomaly",
"fit": "A crash is a rare departure from normal running, so the surface looks for the inputs and faults that make it likely.",
"failureModes": [
"crashes",
"unhandled exceptions",
"process termination"
],
"techniques": [
"chaos-testing",
"fault-injection",
"monitoring"
],
"predicate": {
"type": "absence",
"expression": "no input leads to an unhandled fault",
"grounds": ["ver-ground-truth"]
},
"invariant": "reliability-under-faults",
"evidence": {"required": true,
"source": "runtime-observation",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "availability-correctness",
"dimension": "probability",
"lens": "time",
"fit": "Availability is the share of time the service answers, so the surface measures uptime over a period under a declared fault set.",
"failureModes": [
"service outage",
"cascading failure",
"degraded service"
],
"techniques": [
"chaos-testing",
"fault-injection"
],
"predicate": {
"type": "bound",
"expression": "uptime >= SLO under the declared fault set",
"grounds": ["ver-ground-truth"]
},
"invariant": "availability-under-stress",
"evidence": {"required": true,
"source": "runtime-observation",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "consistency-correctness",
"dimension": "relation",
"lens": "statistical",
"fit": "Replicas agree or drift as a population, so the surface compares the state of every copy with the others.",
"failureModes": [
"stale cache",
"divergent replicas",
"invalid synchronization"
],
"techniques": [
"differential-testing",
"assertion-checking"
],
"predicate": {
"type": "invariant",
"expression": "for-all replicas: converge(state)",
"grounds": ["ver-ground-truth"]
},
"invariant": "consistency-across-components",
"evidence": {"required": true,
"source": "test-result",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "data-correctness",
"dimension": "identity",
"lens": "structure",
"fit": "Persisted data is correct when each record is still the same record in the same shape, so the surface checks identity and layout across writes and migrations.",
"failureModes": [
"corrupted persistence",
"invalid migrations",
"duplicate records"
],
"techniques": [
"assertion-checking",
"integration-testing"
],
"predicate": {
"type": "invariant",
"expression": "persisted = written and migrate then inverse = identity",
"grounds": ["ver-ground-truth"]
},
"invariant": "correct-outputs",
"evidence": {"required": true,
"source": "test-result",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "numerical-correctness",
"dimension": "scale",
"lens": "anomaly",
"fit": "Numbers fail at the edges of their range, so the surface looks for the magnitudes where a result overflows, loses precision or turns into NaN.",
"failureModes": [
"overflow",
"precision loss",
"NaN propagation"
],
"techniques": [
"property-based-testing",
"static-analysis"
],
"predicate": {
"type": "invariant",
"expression": "result is finite and not overflow and not NaN",
"grounds": ["ver-ground-truth"]
},
"invariant": "numerical-validity",
"evidence": {"required": true,
"source": "analysis-report",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "security-correctness",
"dimension": "cause",
"lens": "cause",
"fit": "An exploit is a chain from an input to an effect the boundary should prevent, so the surface traces what each input can cause.",
"failureModes": [
"injection",
"privilege escalation",
"unsafe deserialization"
],
"techniques": [
"static-analysis",
"fuzz-testing"
],
"predicate": {
"type": "absence",
"expression": "no input yields privilege escalation or injection",
"grounds": ["ver-ground-truth"]
},
"invariant": "security-boundaries",
"evidence": {"required": true,
"source": "analysis-report",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "determinism-correctness",
"dimension": "novelty",
"lens": "anomaly",
"fit": "Nondeterminism appears as an output that differs from the one expected for the same input, so the surface repeats runs and looks for the deviation.",
"failureModes": ["the same input producing different outputs"],
"techniques": [
"property-based-testing",
"deterministic-replay"
],
"predicate": {
"type": "equivalence",
"expression": "f(x) = f(x) across runs",
"grounds": ["ver-ground-truth"]
},
"invariant": "deterministic-behavior-where-required",
"evidence": {"required": true,
"source": "test-result",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "protocol-correctness",
"dimension": "relation",
"lens": "sequential",
"fit": "A protocol is a relation between parties that holds only in a given order of messages, so the surface checks the sequence against the protocol's grammar.",
"failureModes": [
"invalid message ordering",
"malformed communication sequence"
],
"techniques": [
"contract-testing",
"model-checking"
],
"predicate": {
"type": "temporal-order",
"expression": "messageSeq in protocolGrammar",
"grounds": ["ver-ground-truth"]
},
"invariant": "protocol-compliance",
"evidence": {"required": true,
"source": "test-result",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "configuration-correctness",
"dimension": "state",
"lens": "structure",
"fit": "Configuration is the condition a deployment starts in, and it is valid when its keys and values take the declared shape, so the surface checks it against its schema.",
"failureModes": [
"invalid environment variables",
"feature flag errors",
"deployment mismatch"
],
"techniques": [
"runtime-validation",
"static-analysis"
],
"predicate": {
"type": "schema",
"expression": "config models configSchema",
"grounds": ["ver-ground-truth"]
},
"invariant": "configuration-validity",
"evidence": {"required": true,
"source": "analysis-report",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
},
{
"id": "observability-correctness",
"dimension": "meaning",
"lens": "frequency",
"fit": "A signal is accurate when every event it stands for is emitted each time the event occurs, so the surface compares how often events happen with how often they are reported.",
"failureModes": [
"missing logs",
"incorrect metrics",
"broken traces"
],
"techniques": [
"monitoring",
"tracing"
],
"predicate": {
"type": "invariant",
"expression": "for-all event: emitted and measurable",
"grounds": ["ver-ground-truth"]
},
"invariant": "accurate-observability",
"evidence": {"required": true,
"source": "runtime-observation",
"grounds": ["ver-evidence"]},
"verdictDomain": [
"pass",
"fail",
"unknown"
]
}
]
}