sdr-visualizer 1.2.4__tar.gz → 1.2.6__tar.gz

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/CHANGELOG.md

@@ -4,6 +4,32 @@ All notable changes to `sdr-visualizer` will be documented here. The format foll
 
 ## [Unreleased]
 
+## [1.2.6] - 2026-10-27
+
+### Changed
+
+- Advance the catalog tested-through generator markers to `cja_auto_sdr`
+  3.13.2 and `aa_auto_sdr` 1.24.1. Snapshots that declare those versions no
+  longer print a newer-than-tested warning. Later versions are still accepted
+  with a compatibility warning. The separate lineage allowlist is unchanged.
+
+## [1.2.5] - 2026-10-17
+
+### Fixed
+
+- Kill a Workspace worker directly when it rejects process-group cleanup and
+  stays running, then report the original cleanup error, so collection no
+  longer waits on a worker that ignored the group signal.
+- Fail the Python performance gate when any published tier fixture is missing,
+  warn when a report reaches 90% of its size budget, and treat a Workspace
+  interaction of exactly 100 ms as over budget, matching the published limit.
+
+### Changed
+
+- Document CJA catalog generator-version provenance, correct adapter
+  tested-through marker guidance, and remove obsolete lineage POC/private
+  module descriptions.
+
 ## [1.2.4] - 2026-10-17
 
 ### Changed
@@ -12,6 +38,8 @@ All notable changes to `sdr-visualizer` will be documented here. The format foll
   3.12.12 and `aa_auto_sdr` 1.21.24. Snapshots declaring those versions no
   longer raise a newer-than-tested warning; later versions remain accepted
   with a compatibility warning. The separate lineage allowlist is unchanged.
+- Clarify release retention-setting versus artifact-expiry checks and the
+  marker-only sibling review disposition.
 
 ### Fixed
 
@@ -34,7 +62,6 @@ All notable changes to `sdr-visualizer` will be documented here. The format foll
 - Make live AA Workspace collection and manual assistive-technology testing
   optional release evidence, retain automated accessibility and keyboard checks,
   and scope release license signoff to project and vendored D3 notices.
-
 - Update the optional AA Workspace SDK to `aanalytics2==0.5.5.post2` and align
   its collection version guard with the package pin.
 - Automatically review and merge verified, lockfile-only patch updates to
@@ -42,6 +69,8 @@ All notable changes to `sdr-visualizer` will be documented here. The format foll
   dependency changes on the manual-review path.
 - Replace the README catalog screenshot with a feature slideshow covering search,
   references, anatomy, comparisons, trends, and offline reports across AA and CJA.
+- Add canonical `AGENTS.md` contributor guidance and ship it in the source
+  distribution.
 
 ## [1.2.2] - 2026-09-21
 
@@ -752,7 +781,9 @@ The following are explicitly internal and may change without notice:
 - Workspace project visualization (v0.4)
 - Schema map view (v0.5)
 
-[Unreleased]: https://github.com/brian-a-au/sdr-visualizer/compare/v1.2.4...HEAD
+[Unreleased]: https://github.com/brian-a-au/sdr-visualizer/compare/v1.2.6...HEAD
+[1.2.6]: https://github.com/brian-a-au/sdr-visualizer/releases/tag/v1.2.6
+[1.2.5]: https://github.com/brian-a-au/sdr-visualizer/releases/tag/v1.2.5
 [1.2.4]: https://github.com/brian-a-au/sdr-visualizer/releases/tag/v1.2.4
 [1.2.3]: https://github.com/brian-a-au/sdr-visualizer/releases/tag/v1.2.3
 [1.2.2]: https://github.com/brian-a-au/sdr-visualizer/releases/tag/v1.2.2

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: sdr-visualizer
-Version: 1.2.4
+Version: 1.2.6
 Summary: Static-output visual catalog generator for Adobe Customer Journey Analytics and Adobe Analytics implementations.
 Project-URL: Homepage, https://brian-a-au.github.io/sdr-visualizer/
 Project-URL: Documentation, https://github.com/brian-a-au/sdr-visualizer#documentation
@@ -280,8 +280,9 @@ the trend is the newest usable snapshot in the directory.
 The output is CI-gated against the budgets in
 [`docs/PERFORMANCE.md`](https://github.com/brian-a-au/sdr-visualizer/blob/main/docs/PERFORMANCE.md).
 Build time and HTML size are enforced at every published tier (100 / 500 / 1,000 / 2,000
-components). Browser-measured budgets are enforced at the 1,000-component tier
-(initial render < 1s, filter/search < 150ms) and the 2,000-component tier
+components). Browser-measured budgets for initial render and filter/search are
+enforced at all four tiers, from 100 components (< 200ms, < 50ms) and 500
+components (< 500ms, < 100ms) up to 1,000 (< 1s, < 150ms) and 2,000
 (< 2s, < 300ms), plus a 700ms cap on the graph view's main-thread block.
 These guarantees cover up to 8,000 reference edges. Denser valid reports use an
 explicit graph opt-in and sit outside the published size and latency limits.
@@ -334,7 +335,7 @@ arguments and exit codes are described in the
 Its embedded payload and Python modules are internal and do not extend the
 catalog schema or JSON sidecar contract.
 
-**Performance budgets.** The tier table above is a guarantee, not a goal: loosening a budget is a breaking change; tightening one is minor.
+**Performance budgets.** The tier table in [`docs/PERFORMANCE.md`](https://github.com/brian-a-au/sdr-visualizer/blob/main/docs/PERFORMANCE.md) is a guarantee, not a goal: loosening a budget is a breaking change; tightening one is minor.
 
 Warnings (snapshot generator newer than the tested version; 5,000+ component reports) are informational and never make a valid snapshot fail.
 

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/README.md

@@ -257,8 +257,9 @@ the trend is the newest usable snapshot in the directory.
 The output is CI-gated against the budgets in
 [`docs/PERFORMANCE.md`](https://github.com/brian-a-au/sdr-visualizer/blob/main/docs/PERFORMANCE.md).
 Build time and HTML size are enforced at every published tier (100 / 500 / 1,000 / 2,000
-components). Browser-measured budgets are enforced at the 1,000-component tier
-(initial render < 1s, filter/search < 150ms) and the 2,000-component tier
+components). Browser-measured budgets for initial render and filter/search are
+enforced at all four tiers, from 100 components (< 200ms, < 50ms) and 500
+components (< 500ms, < 100ms) up to 1,000 (< 1s, < 150ms) and 2,000
 (< 2s, < 300ms), plus a 700ms cap on the graph view's main-thread block.
 These guarantees cover up to 8,000 reference edges. Denser valid reports use an
 explicit graph opt-in and sit outside the published size and latency limits.
@@ -311,7 +312,7 @@ arguments and exit codes are described in the
 Its embedded payload and Python modules are internal and do not extend the
 catalog schema or JSON sidecar contract.
 
-**Performance budgets.** The tier table above is a guarantee, not a goal: loosening a budget is a breaking change; tightening one is minor.
+**Performance budgets.** The tier table in [`docs/PERFORMANCE.md`](https://github.com/brian-a-au/sdr-visualizer/blob/main/docs/PERFORMANCE.md) is a guarantee, not a goal: loosening a budget is a breaking change; tightening one is minor.
 
 Warnings (snapshot generator newer than the tested version; 5,000+ component reports) are informational and never make a valid snapshot fail.
 

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/docs/ADAPTER_GUIDE.md

@@ -37,6 +37,36 @@ Reads the JSON output of [`cja_auto_sdr`](https://github.com/brian-a-au/cja_auto
 - Calc metrics ship `metric_references` and `segment_references` arrays; the adapter merges these and de-dupes.
 - Derived fields appear only in CJA — there's no AA equivalent.
 
+### CJA generator version provenance
+
+The adapter reads `metadata["Tool Version"]` or `metadata["tool_version"]`
+when declared; otherwise `adapter_version` is `unknown` and no
+newer-than-tested warning is possible. The example above illustrates accepted
+version metadata, not a field guaranteed by the exporter.
+
+Public source inspection at `cja_auto_sdr` **v3.12.12**, commit
+`73c3f60e531358d339828abbbb94bbb0afb69b0b`, confirms that its
+[catalog metadata builder](https://github.com/brian-a-au/cja_auto_sdr/blob/73c3f60e531358d339828abbbb94bbb0afb69b0b/src/cja_auto_sdr/generator.py#L1160-L1308)
+omits the generator version. The
+[JSON payload builder](https://github.com/brian-a-au/cja_auto_sdr/blob/73c3f60e531358d339828abbbb94bbb0afb69b0b/src/cja_auto_sdr/output/sdr/__init__.py#L507-L575)
+passes that metadata through without adding a version. There is no additional
+catalog field for the adapter to read in this version.
+
+The separate
+[run summary](https://github.com/brian-a-au/cja_auto_sdr/blob/73c3f60e531358d339828abbbb94bbb0afb69b0b/src/cja_auto_sdr/generator.py#L2516-L2530)
+and
+[diff snapshot format](https://github.com/brian-a-au/cja_auto_sdr/blob/73c3f60e531358d339828abbbb94bbb0afb69b0b/src/cja_auto_sdr/diff/models.py#L300-L375)
+record `tool_version`, but those are different artifacts. The catalog adapter
+does not join external provenance or infer a version from the locally installed
+exporter, timestamps, or inventory schema versions. Existing undeclared catalog
+snapshots remain `unknown`; exporter-side version metadata would be needed to
+make their generator version available at export time.
+
+The tested-through marker comes from recorded validation with an exact upstream
+version and the historical compatibility corpus, as required by the
+[compatibility policy](PRODUCT_CONTRACT.md#compatibility-policy). Taking the
+highest version string found in fixtures or old snapshots is insufficient.
+
 ### CJA component mappings
 
 | CJA input | Normalized collection | Embedded type |

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/docs/PERFORMANCE.md

@@ -191,19 +191,44 @@ uv run python scripts/perf_browser_check.py
 
 CI runs `perf_check.py` after `pytest` in the test job. The browser gate runs in
 the separate `browser-perf` job after browser functional tests. A regression
-that exceeds any budget fails the build.
+that exceeds any budget fails the build. `perf_check.py` fails when any tier
+fixture is missing and prints a non-failing warning when an HTML size reaches
+90% of its budget.
 
 ## What we measure
 
 One representative Python-side run looks like this:
 
 ```text
-[CJA] build time: 0.01s   (budget 6.0s)
-[CJA] HTML size : 0.71MB  (budget 4.0MB)
-[AA] build time: 0.01s   (budget 6.0s)
-[AA] HTML size : 0.61MB  (budget 4.0MB)
-[CJA-XL] build time: 0.01s   (budget 12.0s)
-[CJA-XL] HTML size : 0.98MB  (budget 8.0MB)
+[CJA-100] build time: 0.01s   (budget 1.0s, median of 3)
+[CJA-100] HTML size : 0.43MB  (budget 0.5MB)
+[CJA-500] build time: 0.02s   (budget 3.0s, median of 3)
+[CJA-500] HTML size : 0.57MB  (budget 2MB)
+[CJA] build time: 0.06s   (budget 6.0s, median of 3)
+[CJA] HTML size : 0.81MB  (budget 4MB)
+[AA] build time: 0.04s   (budget 6.0s, median of 3)
+[AA] HTML size : 0.70MB  (budget 4MB)
+[CJA-dense] build time: 0.09s   (budget 6.0s, median of 3)
+[CJA-dense] HTML size : 1.42MB  (budget 4MB)
+[CJA-compare] build time: 0.07s   (budget 9.0s, median of 3)
+[CJA-compare] HTML size : 0.82MB  (budget 4.5MB)
+[CJA-trend] build time: 0.19s   (budget 6.0s, median of 3, 6 snapshots)
+[CJA-trend] HTML size : 0.82MB  (budget 4.5MB)
+[CJA-XL] build time: 0.08s   (budget 12.0s, median of 3)
+[CJA-XL] HTML size : 1.09MB  (budget 8MB)
+[cja_snapshot_small-usage] build time: 0.02s   (budget 1.0s, median of 3)
+[cja_snapshot_small-usage] HTML size : 0.48MB  (budget 0.5MB)
+warning: [cja_snapshot_small-usage] HTML size at 95% of budget
+[cja_snapshot_medium-usage] build time: 0.24s   (budget 3.0s, median of 3)
+[cja_snapshot_medium-usage] HTML size : 1.18MB  (budget 2MB)
+[cja_snapshot_large-usage] build time: 0.30s   (budget 6.0s, median of 3)
+[cja_snapshot_large-usage] HTML size : 1.58MB  (budget 4MB)
+[aa_snapshot_large-usage] build time: 0.29s   (budget 6.0s, median of 3)
+[aa_snapshot_large-usage] HTML size : 1.40MB  (budget 4MB)
+[cja_snapshot_xl-usage] build time: 0.36s   (budget 12.0s, median of 3)
+[cja_snapshot_xl-usage] HTML size : 2.06MB  (budget 8MB)
+[CJA-usage-repeated] build time: 0.35s   (budget 6.0s, median of 3)
+[CJA-usage-repeated] HTML size : 1.78MB  (budget 4MB)
 OK: all budgets met
 ```
 

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/docs/PRODUCT_CONTRACT.md

@@ -207,8 +207,8 @@ unusable, and capped-out members do not warn. The current markers are:
 
 | Platform | Warning threshold |
 |---|---:|
-| CJA (`cja_auto_sdr`) | 3.12.12 |
-| AA (`aa_auto_sdr`) | 1.21.24 |
+| CJA (`cja_auto_sdr`) | 3.13.2 |
+| AA (`aa_auto_sdr`) | 1.24.1 |
 
 A marker is evidence of exercised compatibility, not a promise about every
 possible snapshot produced by that version. It may advance only after a

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "sdr-visualizer"
-version = "1.2.4"
+version = "1.2.6"
 description = "Static-output visual catalog generator for Adobe Customer Journey Analytics and Adobe Analytics implementations."
 readme = "README.md"
 license = "MIT"

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/scripts/perf_browser_check.py

@@ -313,8 +313,8 @@ def _workspace_failures(label: str, measured: dict, project_count: int) -> list[
     ):
         failures.append(f"[{label}] page replacement did not advance the project range")
     for field, action in (("openMs", "first open"), ("pageMs", "page replacement")):
-        if measured[field] > WORKSPACE_INTERACTION_BUDGET_MS:
-            failures.append(f"[{label}] Workspace {action} {measured[field]:.1f}ms > 100ms")
+        if measured[field] >= WORKSPACE_INTERACTION_BUDGET_MS:
+            failures.append(f"[{label}] Workspace {action} {measured[field]:.1f}ms >= 100ms")
     return failures
 
 

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/scripts/perf_check.py

@@ -4,14 +4,14 @@ Renders fixtures and asserts the build-time + HTML-size budgets across all
 four documented tiers. CI runs this after pytest (see .github/workflows/test.yml).
 
   - CJA small (100 components, `generate_large_fixture.py --scale 0.083`):
-    100-component budgets (build < 1s, size < 0.5MB). Skipped when absent.
+    100-component budgets (build < 1s, size < 0.5MB). Required.
   - CJA medium (~500 components, `--scale 0.417`): 500-component budgets
-    (build < 3s, size < 2MB). Skipped when absent.
+    (build < 3s, size < 2MB). Required.
   - CJA / AA large (~1,200 / ~900 components): 1,000-component budgets
     (build < 6s, size < 4MB)
   - CJA XL (~2,000 components, generated on demand via
     `generate_large_fixture.py --scale 1.67 --output tests/fixtures/cja_snapshot_xl.json`):
-    2,000-component budgets (build < 12s, size < 8MB). Skipped when absent.
+    2,000-component budgets (build < 12s, size < 8MB). Required.
 
 Build time is the median of 3 runs per fixture.
 
@@ -71,6 +71,25 @@ TREND_SERIES_LEN = 6
 TREND_BUILD_BUDGET_S = TREND_SERIES_LEN * 1.0
 TREND_SIZE_BUDGET_BYTES = SIZE_BUDGET_BYTES + 500_000
 
+# Non-failing early warning for deterministic HTML size growth.
+HEADROOM_WARNING_RATIO = 0.9
+
+GENERATE = "uv run python scripts/generate_large_fixture.py"
+REQUIRED_FIXTURES = [
+    ("CJA_SMALL", f"{GENERATE} --scale 0.083 --output tests/fixtures/cja_snapshot_small.json"),
+    ("CJA_MEDIUM", f"{GENERATE} --scale 0.417 --output tests/fixtures/cja_snapshot_medium.json"),
+    ("CJA_LARGE", GENERATE),
+    ("AA_LARGE", "uv run python scripts/generate_aa_large_fixture.py"),
+    ("CJA_XL", f"{GENERATE} --scale 1.67 --output tests/fixtures/cja_snapshot_xl.json"),
+]
+
+
+def _headroom_note(label: str, size_bytes: int, budget_bytes: int) -> str | None:
+    if HEADROOM_WARNING_RATIO * budget_bytes <= size_bytes < budget_bytes:
+        pct = size_bytes * 100 // budget_bytes
+        return f"warning: [{label}] HTML size at {pct}% of budget"
+    return None
+
 
 def _load_usage_fixture():
     spec = importlib.util.spec_from_file_location(
@@ -118,6 +137,8 @@ def _measure_compare(old_snap: dict, new_snap: dict) -> tuple[list[str], str]:
         f"[CJA-compare] build time: {elapsed:.2f}s   (budget {COMPARE_BUILD_BUDGET_S}s, median of 3)",
         f"[CJA-compare] HTML size : {size_mb:.2f}MB  (budget {COMPARE_SIZE_BUDGET_BYTES / 1_000_000:g}MB)",
     ]
+    if note := _headroom_note("CJA-compare", size_bytes, COMPARE_SIZE_BUDGET_BYTES):
+        msgs.append(note)
     failures = []
     if elapsed >= COMPARE_BUILD_BUDGET_S:
         failures.append(
@@ -152,6 +173,8 @@ def _measure_trend(base_snap: dict, mutate) -> tuple[list[str], str]:
         f"median of 3, {TREND_SERIES_LEN} snapshots)",
         f"[CJA-trend] HTML size : {size_mb:.2f}MB  (budget {TREND_SIZE_BUDGET_BYTES / 1_000_000:g}MB)",
     ]
+    if note := _headroom_note("CJA-trend", size_bytes, TREND_SIZE_BUDGET_BYTES):
+        msgs.append(note)
     failures = []
     if elapsed >= TREND_BUILD_BUDGET_S:
         failures.append(f"CJA-trend build time {elapsed:.2f}s over budget {TREND_BUILD_BUDGET_S}s")
@@ -187,6 +210,8 @@ def _measure(
         f"[{label}] build time: {elapsed:.2f}s   (budget {build_budget_s}s, median of 3)",
         f"[{label}] HTML size : {size_mb:.2f}MB  (budget {size_budget_bytes / 1_000_000:g}MB)",
     ]
+    if note := _headroom_note(label, size_bytes, size_budget_bytes):
+        msgs.append(note)
     failures = []
     if elapsed >= build_budget_s:
         failures.append(f"[{label}] build time {elapsed:.2f}s >= {build_budget_s}s budget")
@@ -198,17 +223,18 @@ def _measure(
 
 
 def main() -> int:
-    for fixture, generator in [
-        (CJA_LARGE, "scripts/generate_large_fixture.py"),
-        (AA_LARGE, "scripts/generate_aa_large_fixture.py"),
-    ]:
-        if not fixture.exists():
-            print(
-                f"sdr-visualizer: fixture {fixture} missing; run "
-                f"`uv run python {generator}` first.",
-                file=sys.stderr,
-            )
-            return 2
+    missing = [
+        (globals()[name], command)
+        for name, command in REQUIRED_FIXTURES
+        if not globals()[name].exists()
+    ]
+    for fixture, command in missing:
+        print(
+            f"sdr-visualizer: fixture {fixture} missing; run `{command}` first.",
+            file=sys.stderr,
+        )
+    if missing:
+        return 2
 
     cja_snap = json.loads(CJA_LARGE.read_text(encoding="utf-8"))
     aa_snap = json.loads(AA_LARGE.read_text(encoding="utf-8"))
@@ -218,13 +244,10 @@ def main() -> int:
         (CJA_SMALL, "CJA-100", SMALL_BUILD_BUDGET_S, SMALL_SIZE_BUDGET_BYTES),
         (CJA_MEDIUM, "CJA-500", MEDIUM_BUILD_BUDGET_S, MEDIUM_SIZE_BUDGET_BYTES),
     ]:
-        if path.exists():
-            snap = json.loads(path.read_text(encoding="utf-8"))
-            tier_failures, tier_report = _measure(label, snap, cja_adapt, build_b, size_b)
-            small_failures += tier_failures
-            print(tier_report)
-        else:
-            print(f"note: {path.name} not generated; skipping {label} gate")
+        snap = json.loads(path.read_text(encoding="utf-8"))
+        tier_failures, tier_report = _measure(label, snap, cja_adapt, build_b, size_b)
+        small_failures += tier_failures
+        print(tier_report)
 
     cja_failures, cja_report = _measure("CJA", cja_snap, cja_adapt)
     print(cja_report)
@@ -245,15 +268,11 @@ def main() -> int:
     trend_failures, trend_report = _measure_trend(cja_snap, mutate)
     print(trend_report)
 
-    xl_failures: list[str] = []
-    if CJA_XL.exists():
-        xl_snap = json.loads(CJA_XL.read_text(encoding="utf-8"))
-        xl_failures, xl_report = _measure(
-            "CJA-XL", xl_snap, cja_adapt, XL_BUILD_BUDGET_S, XL_SIZE_BUDGET_BYTES
-        )
-        print(xl_report)
-    else:
-        print("note: cja_snapshot_xl.json not generated; skipping 2,000-component gate")
+    xl_snap = json.loads(CJA_XL.read_text(encoding="utf-8"))
+    xl_failures, xl_report = _measure(
+        "CJA-XL", xl_snap, cja_adapt, XL_BUILD_BUDGET_S, XL_SIZE_BUDGET_BYTES
+    )
+    print(xl_report)
 
     # Preserve every original gate, then repeat its tier with bounded evidence.
     # Small approaches the HTML cap; XL approaches the normalized 1 MiB cap.
@@ -266,8 +285,6 @@ def main() -> int:
         (AA_LARGE, aa_adapt, BUILD_BUDGET_S, SIZE_BUDGET_BYTES, 10_000),
         (CJA_XL, cja_adapt, XL_BUILD_BUDGET_S, XL_SIZE_BUDGET_BYTES, 10_000),
     ]:
-        if not path.exists():
-            continue  # Same optional fixtures as the original Python matrix.
         snap = json.loads(path.read_text(encoding="utf-8"))
         encoded = usage_fixture.usage_json(adapter(snap), project_count=projects, pad_input=True)
         errors, report = _measure(

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/src/sdr_visualizer/__init__.py

@@ -1,3 +1,3 @@
 """sdr-visualizer: visual catalog generator for CJA / AA implementations."""
 
-__version__ = "1.2.4"
+__version__ = "1.2.6"

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/src/sdr_visualizer/adapters/aa.py

@@ -351,11 +351,12 @@ def _walk_segment_definition(definition: Any) -> tuple[int, list[str]]:
 # ---------------------------------------------------------------------------
 
 
-# The newest aa_auto_sdr version this release was validated against
-# (bundled fixtures + the private real corpus; re-derive at each release:
-# grep -rho '"Tool Version": "[^"]*"|"tool_version": "[^"]*"' over corpus
-# and fixtures, take the max). The compatibility policy warns — never refuses.
-TESTED_THROUGH_GENERATOR_VERSION = "1.21.24"
+# The newest aa_auto_sdr version this release was validated against.
+# Advance from recorded release validation with that exact generator version
+# and the historical compatibility corpus; snapshot declarations alone do not
+# establish this marker. Undeclared versions stay unknown and do not warn.
+# The compatibility policy warns about newer declared versions, never refuses.
+TESTED_THROUGH_GENERATOR_VERSION = "1.24.1"
 
 
 def generator_version_warning(adapter_version: str) -> str | None:

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/src/sdr_visualizer/adapters/cja.py

@@ -363,11 +363,12 @@ def _walk_for_contexts(node: Any, out: list[str]) -> None:
 # ---------------------------------------------------------------------------
 
 
-# The newest cja_auto_sdr version this release was validated against
-# (bundled fixtures + the private real corpus; re-derive at each release:
-# grep -rho '"Tool Version": "[^"]*"|"tool_version": "[^"]*"' over corpus
-# and fixtures, take the max). The compatibility policy warns — never refuses.
-TESTED_THROUGH_GENERATOR_VERSION = "3.12.12"
+# The newest cja_auto_sdr version this release was validated against.
+# Advance from recorded release validation with that exact generator version
+# and the historical compatibility corpus; snapshot declarations alone do not
+# establish this marker. Undeclared versions stay unknown and do not warn.
+# The compatibility policy warns about newer declared versions, never refuses.
+TESTED_THROUGH_GENERATOR_VERSION = "3.13.2"
 
 
 def generator_version_warning(adapter_version: str) -> str | None:

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/src/sdr_visualizer/analysis/lineage_layout.py

@@ -1,4 +1,4 @@
-"""Deterministic, build-side geometry for the CJA lineage POC."""
+"""Deterministic, build-side geometry for the CJA lineage report."""
 
 from __future__ import annotations
 

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/src/sdr_visualizer/cli/lineage_output.py

@@ -1,4 +1,4 @@
-"""Private, content-free output boundary for the CJA lineage POC."""
+"""Content-free output boundary for the CJA lineage report."""
 
 from __future__ import annotations
 

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/src/sdr_visualizer/render/lineage_payload.py

@@ -1,4 +1,4 @@
-"""Build the private, source-faithful payload for the CJA lineage POC."""
+"""Build the source-faithful payload for the CJA lineage report."""
 
 from __future__ import annotations
 

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/src/sdr_visualizer/usage/collector.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import contextlib
 import importlib.metadata
 import os
 import selectors
@@ -416,10 +417,16 @@ def _stop(process):
         except PermissionError as error:
             # On macOS an exiting process group can reject a signal before
             # waitpid reports its leader's exit. Only accept cleanup once the
-            # child has actually exited, within the existing wait bound.
+            # child has actually exited, within the existing wait bound;
+            # otherwise kill the leader directly so the Popen context can finish
+            # once it exits, then report the refused group signal.
             try:
                 process.wait(timeout=2)
             except subprocess.TimeoutExpired:
+                with contextlib.suppress(OSError):
+                    process.kill()
+                with contextlib.suppress(subprocess.TimeoutExpired):
+                    process.wait(timeout=2)
                 raise error from None
             return
     process.wait(timeout=2)

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/src/sdr_visualizer/usage/sdk_worker.py

@@ -125,10 +125,16 @@ def _stop(process):
             pass
         except PermissionError as error:
             # An exiting process group can reject a signal before its leader is
-            # reaped. Accept cleanup only after exit within the existing bound.
+            # reaped. Accept cleanup only after exit within the existing bound;
+            # otherwise kill the leader directly so the Popen context can finish
+            # once it exits, then report the refused group signal.
             try:
                 process.wait(timeout=2)
             except subprocess.TimeoutExpired:
+                with contextlib.suppress(OSError):
+                    process.kill()
+                with contextlib.suppress(subprocess.TimeoutExpired):
+                    process.wait(timeout=2)
                 raise error from None
             return
     process.wait(timeout=2)

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/tests/test_adapters_aa.py

@@ -521,5 +521,5 @@ def test_equal_older_or_unparseable_versions_do_not_warn():
 def test_tuple_length_mismatch_versions_compare_correctly():
     from sdr_visualizer.adapters.aa import generator_version_warning
 
-    assert generator_version_warning("1.22") is not None
-    assert generator_version_warning("1.21") is None
+    assert generator_version_warning("1.25") is not None
+    assert generator_version_warning("1.24") is None

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/tests/test_adapters_cja.py

@@ -858,12 +858,12 @@ def test_equal_older_or_unparseable_versions_do_not_warn():
 
 
 def test_tuple_length_mismatch_versions_compare_correctly():
-    # (3, 13) > (3, 12, 4), while the shorter (3, 12) sorts below the
+    # (3, 14) > (3, 13, 2), while the shorter (3, 13) sorts below the
     # longer tested tuple. This pins Python tuple ordering as the contract.
     from sdr_visualizer.adapters.cja import generator_version_warning
 
-    assert generator_version_warning("3.13") is not None
-    assert generator_version_warning("3.12") is None
+    assert generator_version_warning("3.14") is not None
+    assert generator_version_warning("3.13") is None
 
 
 # ---------------------------------------------------------------------------

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/tests/test_cli.py

@@ -1101,10 +1101,10 @@ def test_tested_generator_version_does_not_warn(tmp_path, capsys):
 @pytest.mark.parametrize(
     ("platform", "version", "warns"),
     [
-        ("cja", "3.12.12", False),
-        ("cja", "3.12.13", True),
-        ("aa", "1.21.24", False),
-        ("aa", "1.21.25", True),
+        ("cja", "3.13.2", False),
+        ("cja", "3.13.3", True),
+        ("aa", "1.24.1", False),
+        ("aa", "1.24.2", True),
     ],
 )
 def test_catalog_generator_warning_boundary(platform, version, warns, tmp_path, capsys):

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/tests/test_perf.py

@@ -160,3 +160,68 @@ def test_fixture_and_corpus_tier_budgets_agree_exactly():
     assert corpus._tier_budget_bytes(2001) is None
     assert perf_check.COMPARE_SIZE_BUDGET_BYTES == 4_500_000
     assert perf_check.TREND_SIZE_BUDGET_BYTES == 4_500_000
+
+
+@pytest.mark.parametrize("delta,expected", [(-1, None), (0, "90%"), (1, "90%")])
+def test_headroom_note_boundary(delta, expected):
+    note = perf_check._headroom_note("CJA-100", 450_000 + delta, 500_000)
+    if expected is None:
+        assert note is None
+    else:
+        assert expected in note
+        assert "CJA-100" in note
+
+
+def test_headroom_note_absent_at_or_over_budget():
+    assert perf_check._headroom_note("x", 500_000, 500_000) is None
+    assert perf_check._headroom_note("x", 500_001, 500_000) is None
+
+
+def test_measure_report_includes_headroom_note_without_failing(monkeypatch):
+    monkeypatch.setattr(perf_check, "render", lambda impl: "x" * 460_000)
+    monkeypatch.setattr(perf_check.time, "perf_counter", lambda: 0)
+    failures, report = perf_check._measure(
+        "CJA-100", {}, lambda snap: None, size_budget_bytes=500_000
+    )
+    assert failures == []
+    assert "warning: [CJA-100] HTML size at 92% of budget" in report
+
+
+@pytest.mark.parametrize("mode,label", [("compare", "CJA-compare"), ("trend", "CJA-trend")])
+def test_comparative_report_includes_headroom_note_without_failing(monkeypatch, mode, label):
+    monkeypatch.setattr(perf_check, "cja_adapt", lambda snap: None)
+    monkeypatch.setattr(perf_check, "build_payload_with_options", lambda impl: {"meta": {}})
+    monkeypatch.setattr(perf_check, "diff_implementations", lambda *args: {"baseline": {}})
+    monkeypatch.setattr(perf_check, "build_trend", lambda *args, **kwargs: {})
+    monkeypatch.setattr(perf_check, "render_payload", lambda payload: "x" * 4_050_000)
+    monkeypatch.setattr(perf_check.time, "perf_counter", lambda: 0)
+    if mode == "compare":
+        failures, report = perf_check._measure_compare({}, {})
+    else:
+        failures, report = perf_check._measure_trend({}, lambda snap: snap)
+    assert failures == []
+    assert f"warning: [{label}] HTML size at 90% of budget" in report
+
+
+def test_main_fails_when_any_tier_fixture_missing(tmp_path, monkeypatch, capsys):
+    for name in ("CJA_SMALL", "CJA_MEDIUM", "CJA_LARGE", "AA_LARGE", "CJA_XL"):
+        monkeypatch.setattr(perf_check, name, tmp_path / f"{name.lower()}.json")
+    assert perf_check.main() == 2
+    out = capsys.readouterr()
+    for flag in ("--scale 0.083", "--scale 0.417", "--scale 1.67", "generate_aa_large_fixture"):
+        assert flag in out.err
+    for name in ("cja_small", "cja_medium", "cja_large", "aa_large", "cja_xl"):
+        assert f"{name}.json missing" in out.err
+    assert "OK: all budgets met" not in out.out
+
+
+def test_main_names_only_the_missing_tier(tmp_path, monkeypatch, capsys):
+    present = tmp_path / "present.json"
+    present.write_text("{}", encoding="utf-8")
+    for name in ("CJA_MEDIUM", "CJA_LARGE", "AA_LARGE", "CJA_XL"):
+        monkeypatch.setattr(perf_check, name, present)
+    monkeypatch.setattr(perf_check, "CJA_SMALL", tmp_path / "cja_snapshot_small.json")
+    assert perf_check.main() == 2
+    err = capsys.readouterr().err
+    assert "cja_snapshot_small.json" in err
+    assert "--scale 0.417" not in err

{sdr_visualizer-1.2.4 → sdr_visualizer-1.2.6}/tests/test_perf_browser_check.py

@@ -157,3 +157,19 @@ def test_comparative_browser_render_boundary(monkeypatch, mode, delta):
         page.evaluate.side_effect = [1, 0, 59, 0, 100]
         failures = perf_browser_check._check_trend(page, Path("/tmp/boundary.html"))
     assert bool(failures) == (delta >= 0)
+
+
+@pytest.mark.parametrize("ms,fails", [(99.9, False), (100.0, True)])
+def test_workspace_gate_rejects_exact_budget(ms, fails):
+    observations = {
+        "eager": 0,
+        "total": 10000,
+        "first": 50,
+        "second": 50,
+        "firstRange": "Showing 1–50 of 10000",
+        "secondRange": "Showing 51–100 of 10000",
+        "openMs": ms,
+        "pageMs": 3,
+    }
+    failures = perf_browser_check._workspace_failures("synthetic", observations, 10000)
+    assert bool(failures) == fails