fdp imagemap (cmapx/imap) parity dashboard
Differential survey of @knowvah/dot-engine fdp imagemap (cmapx/imap) output vs the native dot -K fdp oracle (test/corpus/map-walk.ts), semantic comparison per docs/conformance.md (±0.01 deterministic tolerance, ±0.5 for the iterative engines). Regenerate: npx tsx test/corpus/map-walk.ts fdp && npx tsx test/corpus/parity-report.ts.
Summary
- Surveyed: 762
- pass: 739 (97.0%) · diverged (tracked): 0 · accepted (documented, won't-fix): 18
- errors (oracle/port/timeout, excluded from scoring): 5
Diverged (0)
(none)
Accepted (18) — documented, not chased
| id | #diffs | firstDiff | reason |
|---|---|---|---|
1447 | 138 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. |
1447_1 | 661 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. |
1879 | 2148 | area[0].coords[0] | A1 drift, chaos-proven 2026-07-24: fdp's spring layout on this graph is chaotically input-sensitive — port self-perturbation experiments (2620: node width +0.001in -> median 1120pt/max 3390pt movement of the port's OWN layout; badvoro: ONE edge len +0.0001 -> median 322pt/max 1322pt across all 1235 nodes) move the layout by the same magnitude as the port-vs-native divergence. Sub-ULP V8-vs-libm FP dust separates basins; no port stage is implicated. Same class as the fdp engine track's documented 149-id drift tail. |
graphs-b102 | 1201 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. |
graphs-b103 | 3766 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. T6 2026-07-24: classified — port completes in 63-109s vs oracle 38-47s (1.7-2.3x, NOT the >=10x perf-quarantine class); earlier diverged/timeout flips were 8-way sweep-load artifacts (180s budget). Divergence itself is the fdp chaos class (see 1879/badvoro perturbation proof). Re-verdict big fdp ids with MAP_CONCURRENCY=1 if a load flip recurs; registry row keeps flips harmless. |
graphs-b106 | 206 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. |
graphs-b143 | 280 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. |
graphs-b29 | 436 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. |
graphs-b85 | 4 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. |
graphs-root | 16 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. T6 2026-07-24: classified — port completes in 63-109s vs oracle 38-47s (1.7-2.3x, NOT the >=10x perf-quarantine class); earlier diverged/timeout flips were 8-way sweep-load artifacts (180s budget). Divergence itself is the fdp chaos class (see 1879/badvoro perturbation proof). Re-verdict big fdp ids with MAP_CONCURRENCY=1 if a load flip recurs; registry row keeps flips harmless. |
graphs-xx | 1204 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. |
linux.i386-b102 | 1193 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. |
linux.x86-root_circo | 16 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. |
linux.x86-root_twopi | 16 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. T6 2026-07-24: classified — port completes in 63-109s vs oracle 38-47s (1.7-2.3x, NOT the >=10x perf-quarantine class); earlier diverged/timeout flips were 8-way sweep-load artifacts (180s budget). Divergence itself is the fdp chaos class (see 1879/badvoro perturbation proof). Re-verdict big fdp ids with MAP_CONCURRENCY=1 if a load flip recurs; registry row keeps flips harmless. |
nshare-root_circo | 16 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. |
nshare-root_twopi | 16 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. T6 2026-07-24: classified — port completes in 63-109s vs oracle 38-47s (1.7-2.3x, NOT the >=10x perf-quarantine class); earlier diverged/timeout flips were 8-way sweep-load artifacts (180s budget). Divergence itself is the fdp chaos class (see 1879/badvoro perturbation proof). Re-verdict big fdp ids with MAP_CONCURRENCY=1 if a load flip recurs; registry row keeps flips harmless. |
share-b102 | 1198 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. |
windows-b102 | 1186 | area[0].coords[0] | layout drift inherited from the fdp xdot engine track: id diverged there with the documented accepted mechanism; imagemap re-emits the same layout as rounded hotspot coords. Journal 2026-07-24. |
Errors and timeouts (5)
| id | message |
|---|---|
1652 | oracle exit null |
2222 | oracle exit null |
2239 | Warning: Warning: no hard-coded metrics for 'sans'. Falling back to 'Times' metrics |
2620 | Warning: Warning: no hard-coded metrics for 'Times-Bold'. Falling back to 'Times' metrics |
2717 | oracle exit null |
Passing ids (739) are omitted for brevity — the full roster is in map-parity-fdp.json.