test(scale): bounded billion-edge scale ladder and first-fail contract (#736)

Makefile

@@ -1,4 +1,4 @@
-.PHONY: help lint format type-check security workflow-lint license-check third-party-notices third-party-notices-check cargo-deny-licenses test pre-push pre-push-clean pre-push-preflight pre-push-fast bazel-test clean test-tck docstring-coverage test-network benchmark test-perf test-perf-xs test-perf-slow test-perf-large coverage coverage-rust coverage-python coverage-node coverage-quick coverage-report coverage-diff coverage-strict check-coverage check-coverage-rust check-coverage-python check-coverage-node check-patch-coverage test-durations test-analytics docs-serve docs-build docs-clean cargo-build codspeed-build codspeed-run bench-traversal bench-fixed-hop-limit bench-fixed-hop-livejournal bench-m4-entry bench-g500-scale20 bench-adjacency-200m bench-file-backed-128m m4-entry-matrix-check durability-isolation-check native-consumers release-load-matrix-check release-load-matrix bulk-construction-conformance-check bulk-construction-conformance cargo-test cargo-check cargo-clippy cargo-fmt cargo-fmt-check clean-builds clean-builds-all pnpm-install pnpm-build pnpm-test-bdd install build release-version-check package-license-verify publish-dry-run publish-dry-run-npm publish-dry-run-docs publish-dry-run-python publish-dry-run-cargo record-release-artifacts clean-env-verify-check clean-env-verify-preflight clean-env-verify
+.PHONY: help lint format type-check security workflow-lint license-check third-party-notices third-party-notices-check cargo-deny-licenses test pre-push pre-push-clean pre-push-preflight pre-push-fast bazel-test clean test-tck docstring-coverage test-network benchmark test-perf test-perf-xs test-perf-slow test-perf-large coverage coverage-rust coverage-python coverage-node coverage-quick coverage-report coverage-diff coverage-strict check-coverage check-coverage-rust check-coverage-python check-coverage-node check-patch-coverage test-durations test-analytics docs-serve docs-build docs-clean cargo-build codspeed-build codspeed-run bench-traversal bench-fixed-hop-limit bench-fixed-hop-livejournal bench-m4-entry bench-g500-scale20 bench-g500-ladder bench-adjacency-200m bench-file-backed-128m m4-entry-matrix-check durability-isolation-check native-consumers release-load-matrix-check release-load-matrix bulk-construction-conformance-check bulk-construction-conformance cargo-test cargo-check cargo-clippy cargo-fmt cargo-fmt-check clean-builds clean-builds-all pnpm-install pnpm-build pnpm-test-bdd install build release-version-check package-license-verify publish-dry-run publish-dry-run-npm publish-dry-run-docs publish-dry-run-python publish-dry-run-cargo record-release-artifacts clean-env-verify-check clean-env-verify-preflight clean-env-verify
 
 help:  ## Show this help message
 	@grep -E '^[a-zA-Z_-]+:.*?## .*$$' $(MAKEFILE_LIST) | sort | awk 'BEGIN {FS = ":.*?## "}; {printf "\033[36m%-20s\033[0m %s\n", $$1, $$2}'
@@ -309,6 +309,10 @@ bench-m4-entry:  ## Emit the M4 entry large/manual evidence envelope (#334; hard
 bench-g500-scale20:  ## Official-parameter SCALE-20 public-facade engineering green (#710; ignored, not Official-track)
 	cargo test -p graphforge-api --release --test scale_g500_scale20 scale20_public_facade_engineering_green -- --ignored --nocapture --test-threads=1
 
+bench-g500-ladder:  ## Bounded billion-edge scale ladder S20-S26 first-fail evidence (#736; ignored, provisioned scale-host)
+	GF_G500_LADDER_EVIDENCE_OUT="$(CURDIR)/docs/development/g500-ladder-evidence.json" \
+	cargo test -p graphforge-api --release --test scale_g500_ladder ladder_public_facade_first_fail_evidence -- --ignored --nocapture --test-threads=1
+
 bench-adjacency-200m:  ## >200M-edge public adjacency build evidence (#336; ignored, scale-host)
 	GF_ADJACENCY_SCALE_EVIDENCE_OUT="$(CURDIR)/docs/development/adjacency-200m-evidence.json" \
 	GF_ADJACENCY_SCALE_WORK="$(CURDIR)/build/adjacency-200m-work" \

crates/graphforge-api/BUILD.bazel

@@ -186,6 +186,17 @@ gf_rust_integration_test(
     deps = _API_DEPS,
 )
 
+gf_rust_integration_test(
+    name = "scale_g500_ladder",
+    srcs = ["tests/scale_g500_ladder.rs"],
+    compile_data = ["tests/fixtures/scale_g500_ladder.v1.json"],
+    crate = ":graphforge_api",
+    data = _API_TEST_DATA,
+    size = "large",
+    timeout = "long",
+    deps = _API_DEPS,
+)
+
 gf_rust_integration_test(
     name = "adjacency_scale_evidence",
     srcs = ["tests/adjacency_scale_evidence.rs"],
@@ -504,6 +515,7 @@ test_suite(
         ":public_facade_remaining_conformance",
         ":public_lifecycle_conformance",
         ":release_load_construction",
+        ":scale_g500_ladder",
         ":scale_g500_scale20",
         ":strict_runtime_properties",
         ":value_access_semantics",

crates/graphforge-api/tests/fixtures/scale_g500_ladder.v1.json

@@ -0,0 +1,58 @@
+{
+  "schema": "graphforge-billion-edge-ladder/1",
+  "schema_version": "1",
+  "objective": "Bounded-memory Graph500-parameter (ef=16 undirected Kronecker) scale ladder that measures the first envelope bottleneck on the public GraphForge facade. Engineering-green, NOT Official-track and NOT TEPS. Certification of one billion live edges is #745, not this profile.",
+  "seed": 1,
+  "edgefactor": 16,
+  "initiator": { "A": 0.57, "B": 0.19, "C": 0.19, "D": 0.05 },
+  "policy": {
+    "directionality": "undirected",
+    "self_loops": "drop",
+    "duplicates": "drop",
+    "canonicalization": "sorted_lo_hi_pair",
+    "identity": "uuidv7(vertex_index + 1)",
+    "reconciliation": "raw_attempts == live_unique_edges + self_loops_rejected + duplicates_rejected"
+  },
+  "envelope": {
+    "rss_bytes": 137438953472,
+    "disk_bytes": 1099511627776,
+    "timeout_s": 86400,
+    "note": "128 GiB peak RSS, 1 TiB local NVMe, 24 hours (M5 #735 host envelope)."
+  },
+  "metrics": [
+    "raw_attempts",
+    "self_loops_rejected",
+    "duplicates_rejected",
+    "live_unique_edges",
+    "wall_time_s",
+    "rss_peak_bytes",
+    "disk_used_bytes"
+  ],
+  "invocation": "CARGO_TARGET_DIR=/tmp/cargo-g500-ladder make bench-g500-ladder",
+  "rungs": [
+    {
+      "id": "S10",
+      "scale": 10,
+      "buffer_edges": 4096,
+      "tier": "ci",
+      "note": "Small always-on validation rung. buffer_edges is intentionally below the rung's edge count so the external sort/spill/merge path is exercised in normal CI."
+    },
+    {
+      "id": "S20",
+      "scale": 20,
+      "buffer_edges": 8388608,
+      "tier": "provisioned",
+      "note": "Historical baseline (parity with the #710 SCALE-20 reference client)."
+    },
+    { "id": "S22", "scale": 22, "buffer_edges": 16777216, "tier": "provisioned" },
+    { "id": "S24", "scale": 24, "buffer_edges": 33554432, "tier": "provisioned" },
+    { "id": "S25", "scale": 25, "buffer_edges": 67108864, "tier": "provisioned" },
+    {
+      "id": "S26",
+      "scale": 26,
+      "buffer_edges": 134217728,
+      "tier": "provisioned",
+      "note": "SCALE-26/ef=16 = 1,073,741,824 raw attempts. Live persisted edges after self-loop and duplicate policy may fall below one billion; raw attempts MUST NOT be reported as live."
+    }
+  ]
+}