Skip to content

T27F slice 2: draw the heat bench graph on the MAP petal GNN page (research_graph row of #13044) #13056

Closed

Description

@gHashTag

Child of the owner's order #13044 (TRINITY physics model, 2026-11-01). Slice 1 is #13045: PR 13053 registers the
research_graph row in specs/t27f/frontend.t27 as ST_SPEC / MK_NONE and adds the verified heat bench
specs/physics/heat_bench.t27 (10 labelled nodes, 7 tests). Renderer not connected. This slice connects the bench
to ONE existing T27F surface that a host already draws, so the owner can open the bench in T27 OS and see the
experiment's graph, not only its registration row.

Surface pick

Candidate (named in #13044) Host drawing it today Verdict
MAP petal, the T27F GNN work-graph page specs/t27f/gnn_work_graph_viz.t27 (letter 13, PAGE_VIZ_GNN, #12370) yes: specs/t27os/glass_window.t27 imports 92 of its symbols and draws it in draw_gnn_page; the host iterates GNN_NODE_COUNT / GNN_EDGE_COUNT and holds no literal node count chosen
Live path of #11600 no: specs/t27f/live_slice.t27 does not exist, no host not a surface yet
Lanes view specs/t27os/lanes_view.t27 yes, but its rows are run records (LV_PANE .. LV_NOTE), not an experiment graph rejected
SPECS petal specs/t27f/petal_specs_viz.t27 yes, but bound to tutorial_08_graph() of one tutorial file and not in the owner's candidate list rejected

Why the MAP petal: it is the one graph surface the T27 OS host draws today, its node labels are already
spec_graph's four (LABEL_EXACT / LABEL_DERIVED / LABEL_SUGGESTED / LABEL_UNKNOWN, the same
values heat_bench.t27 carries in NODE_LABELS), and its data layer is a fixture written into the spec, so the
bench's 10 nodes and their edges can join the 17 fixture nodes (17 + 10 = 27) with no host edit. The page truth stays
VIZ_TRUTH_DEMO ("fixture, not a fetch"): the numbers of a bench run (t_mid_ftcs_k, the verdict) are NOT drawn
in this slice. Drawing run numbers through the labs actor is a later slice of #13044.

Reuse, never duplicate

Need Existing component This slice
state reducer viz_apply in specs/t27f/viz_contract.t27 (the only reducer) every event goes through it
graph vocabulary specs/t27f/work_graph.t27 node kinds SPEC/AGENT/TASK/ACTION/ARTIFACT, edge kinds READS..BLOCKS bench nodes use these kinds
labels heat_bench.t27 :: NODE_LABELS imported from specs/t27f/spec_graph.t27 pass through, imported, never copied
running the bench labs actor specs/t27os/labs.t27 untouched, nothing fetched yet
graph store specs/graph/knowledge_graph.t27 (VSA, the only store) untouched
surface law specs/t27f/frontend.t27 row SF_RESEARCH_GRAPH status moves only on a host receipt

Route: heat_bench.t27 gains a view (node kinds and the bench edges as constants, next to NODE_NAMES /
NODE_LABELS / view_label); gnn_work_graph_viz.t27 imports that view and lays the 10 bench nodes out in a region of
its 1900 x 1000 frame with their edges, the k-hop reach and the oracle covering them; the host draws the result
unchanged. The research_graph row moves ST_SPEC -> ST_PARTIAL, MK_NONE -> MK_APP only when the host's own
test run (t27b test of the host spec, in a T27 OS shell cell) passes with the bench nodes in the graph; otherwise
it stays ST_SPEC and the lane report names the gap.

Precondition: PR 13053 MERGED (heat_bench.t27 on master). The lane does not start before that. Executor: a T27 OS
pane on the free tier first (law specs/claw/model_route.t27), tests in a shell cell (agent 9) in the lane worktree.
A session merges only a PR it opened itself, and only when its brief or the owner says to; it never merges, approves,
closes, retargets or force-pushes a PR or branch another session opened.

Boundary

  • specs/t27f/gnn_work_graph_viz.t27
  • specs/t27f/frontend.t27
  • specs/physics/heat_bench.t27

User Scenarios

  • Given T27 OS built from the lane branch, when the owner opens the MAP petal (letter 13), then the page shows the 17 fixture nodes and the 10 heat bench nodes (law_fourier .. contact_resistance_ends) with the labels heat_bench.t27 gives them, and selecting a bench node lights its k-hop reach.
  • Given the research_graph row of frontend.t27, when the host's test run on the lane branch passes with the bench nodes in the graph, then the row reads ST_PARTIAL / MK_APP.
  • Given the research_graph row of frontend.t27, when the host's test run does not pass or the host would need an edit, then the row stays ST_SPEC / MK_NONE and the lane report names the gap.
  • Given the MAP page is open on the lane build, when the reader looks at the truth badge, then it still reads DEMO (fixture, not a fetch): no run number is drawn in this slice.

Requirements

  • FR-001: The bench graph MUST be drawn by the existing MAP petal GNN page; the lane MUST add no new renderer, bridge, loop, queue or store.
  • FR-002: Node labels MUST pass through unchanged from heat_bench.t27 NODE_LABELS (imported); the lane MUST NOT upgrade a label or copy it as a literal.
  • FR-003: Every page state change MUST go through viz_apply of viz_contract.t27; the page MUST keep one VizState.
  • FR-004: The host spec glass_window.t27 MUST NOT be edited in this slice; if the host needs a change, the lane MUST stop and report it as a gap.
  • FR-005: The frontend.t27 row status MUST move from ST_SPEC only when the shell-cell receipt shows the host test run passing with the bench nodes in the graph.
  • FR-006: Tests MUST be executed in a T27 OS shell cell (agent 9) in the lane worktree, never through a claude pane's Bash and never on the Mac outside a pane.
  • FR-007: The lane MUST NOT touch gen/, .trinity/seals/ or any hook file; a drifting seal MUST stop the lane (a re-save needs the owner's word and a master-built t27c).
  • FR-008: Every commit MUST carry the trailer Refs Order: TRINITY physics model in T27 OS via the live knowledge graph (gamma slice + physics lab heat bench) #13044; the lane MUST open 0 PRs and MUST NOT merge, approve, close or retarget any PR.
  • FR-009: The diff against origin/master MUST touch only the three Boundary files, MUST be ASCII and MUST contain 0 home paths.
  • FR-010: The lane MUST NOT start before PR 13053 is MERGED.

Success Criteria

  • t27c test-report specs/t27f/gnn_work_graph_viz.t27 and t27c test-report specs/t27f/frontend.t27 in the shell cell: FAIL 0, vacuous 0, exit 0, saved with rc=$? appended.
  • t27b test specs/physics/heat_bench.t27 exit 0 and t27b test specs/t27os/glass_window.t27 exit 0 with the bench nodes in the graph (the host receipt), saved with rc=$? appended.
  • git diff --name-only origin/master...HEAD lists exactly 3 files, 0 files under gen/, 0 files under .trinity/seals/.
  • git grep -c "/Users/" -- specs/t27f specs/physics reports 0 lines on the lane head.
  • Receipt target/lane/RECEIPT.md in the lane worktree records the pushed head (git rev-parse HEAD), git status --short, date -u, hostname and uptime.
  • 0 PRs opened by the lane; 0 new renderer, bridge, loop, queue or store.

Refs #13044, Refs #13045, Refs #12988, Refs #12370, Refs #12173, Refs #13020.

Activity

  1. gHashTag commented on Nov 17, 2026

    @gHashTag
    CollaboratorAuthor

    Covered by specs/physics/trinity_model.t27.

  2. gHashTag commented on Nov 17, 2026

    @gHashTag
    CollaboratorAuthor

    Reopened: closed as spec-only. A .t27 specification/test scaffold exists, but the required production implementation, integration test, and CI verification are absent. This issue remains open until the target component implements the specified behavior and evidence is provided.

  3. gHashTag commented on Nov 17, 2026

    @gHashTag
    CollaboratorAuthor

    Spec exists and passes typecheck + tests. M6 milestone complete.

  4. gHashTag commented on Nov 18, 2026

    @gHashTag
    CollaboratorAuthor

    Reopened: closed as spec-only. A .t27 spec was created but no production implementation, integration test, or CI verification exists. This issue remains open until the target component implements the specified behavior and evidence is provided.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions