Repository navigation
spec(t27f): interactive petal viz MVP -- contract, 27-petal plan, reference petal S SPECS (Closes #12174) - #12193
Merged
github-actions[bot] merged 4 commits intoOct 25, 2026
Conversation
Contributor
This was referenced Oct 25, 2026
gHashTag
marked this pull request as ready for review
October 25, 2026 03:51
Contributor
PR DashboardGenerated at: 2026-10-25 03:54:48 UTC
Summary
Seal Status
|
…27-petal plan (see #12173) viz_contract.t27 is the contract every petal visualization is drawn through: lifecycle (idle, loading, ready, empty, error), truth labels with a claim rule (only LIVE and VERIFIED may be shown as a result), packed entity refs, VizState with one pure reducer (hover, select, step, param, keyboard focus, back, reduced motion), representation kinds with a plan check, and a nested widget host with hit test, validation and embedding. 32 tests, 2 invariants, 0 vacuous passes. petal_viz_plan.t27 lists all 27 petals: the user question, the existing source-of-truth spec (every path checked), the representation kinds and the status (1 reference, 22 planned, 4 deferred with a reason). The host-page column is held equal to agent_parts page_kind. 14 tests, 1 invariant, 0 vacuous passes. Refs #12174; the reference petal (petal_specs_viz.t27) follows on this branch. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…views over the real spec graph (Closes #12174) petal_specs_viz.t27 is the reference petal of the interactive visualization epic, built on viz_contract over tutorial_08_graph() (15 nodes, 16 edges), the graph spec_graph_derive proves equal to the exporter's output on the real tutorial file. - One PsvState (the contract's VizState plus an edge mask), one reducer: contract events pass through viz_apply, PSV_EV_TOGGLE_EDGE flips one edge and drops truth to DEMO until every edge is back on. - Three linked views from that state: node-link (spec_graph_view boxes), adjacency matrix and edge list. Every view's hit test inverts its layout, and selection and hover mark the same entities in all three. - Message passing with a self loop over 4 layers, sum, mean or max, with every intermediate value in the trace and the inspector. A test checks, for every seed, aggregation, layer and node, that a value is nonzero exactly when the breadth-first distance is at most the layer (3375 asserts), and again under 7 other edge sets. - Host: the views are placed through VizHost and validated; psv_reps passes the contract's plan check. Two whole-graph checks are tests, not invariants: invariants are evaluated at compile time, and those loops exceed the evaluator's branch budget. petal_viz_plan.t27: row S now reads its reps from psv_reps() and the test asserts the equality, so the plan cannot drift from the petal. The old row claimed a source view the petal does not draw. t27c test-report: petal 67/67, 0 vacuous, 2 invariants; plan 14/14, 0 vacuous, 1 invariant. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
gHashTag
force-pushed
the
claude/t27os-interactive-viz-mvp-57fe52
branch
from
October 25, 2026 03:56
3e5341d to
45ded41
Compare
Contributor
PR DashboardGenerated at: 2026-10-25 03:56:38 UTC
Summary
Seal Status
|
…ctive-viz-mvp-57fe52
This was referenced Oct 25, 2026
…longer deferred (Refs #12174) Master renamed petal G from ARENA to AGENTS and gave it a bespoke page (agent_parts page_kind(7) == PAGE_AGENTS, t27os#3): glass_window draws the swarm feed and the Queen's claims ledger that specs/t27os/devices.t27 parses. The plan holds its host-page column equal to page_kind, so Spec Guards failed one test on the merge with master. Row G now reads: question "which agent is doing what, and which issue did the Queen assign to whom", source specs/t27os/devices.t27, reps list + timeline + badge, PLANNED, truth LIVE, host page 1. The counts follow: 23 planned, 3 deferred (E, R, Y), 6 live pages, 22 rows that may be drawn as a result. t27c test-report specs/t27f/petal_viz_plan.t27: 14/14, 0 vacuous. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Contributor
PR DashboardGenerated at: 2026-10-25 04:07:13 UTC
Summary
Seal Status
|
This was referenced Oct 25, 2026
Merged
gHashTag
added a commit
that referenced
this pull request
Oct 25, 2026
…12173) The bee review on #12282 read "the contract, the plan and the reference did" as a claim that those three specs carry seals. They do not: the only viz seal under .trinity/seals/ is viz_VizDistillGraph.json. The sentence now says that none of the three is sealed. The two "(checked 2026-10-25)" notes about gaps in the reference petal now name what they were checked against, the reference as merged in #12193, so a reader can verify them later. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
gHashTag
added a commit
that referenced
this pull request
Oct 25, 2026
…ore petal (Refs #12173) (#12282) * docs(t27f): VIZ-PETAL-GUIDE -- how to add the visualization for one more petal (Refs #12173) The epic's success criterion SC-007 asks for tests and a contributor guide. The tests landed with the contract, the plan and the reference petal S; this is the guide. docs/nona-03-manifest/VIZ-PETAL-GUIDE.md walks a contributor through one petal: find the row in petal_viz_plan.t27, name the file and the prefix, write the parts in the reference's order (state through viz_apply, entities from the source spec via use, one view per rep kind, highlights from the one state, inspector, host), the test categories the success criteria need, the light and heavy checks, and how it lands (4-heading child issue, Refs or Closes on every commit, merge master and re-run test-report before each push, no plan edit in a petal PR). It is a map, not a copy: every rule is named by the spec and function that holds it, and no value, table or count is restated. It also records two gaps in the reference that new petals should not copy: no test of its own for the EMPTY and ERROR lifecycles, reduced motion, vh_focus_next or vh_embed. docs/README.md gets the index row next to VIZ-DISTILL-GRAPH. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(viz): mark tri viz graph as planned so the dead-tri ceiling holds (Refs #12173) VIZ-DISTILL-GRAPH.md line 107 named `tri viz graph`, a command no surface has yet. That raised the dead `tri` mention count to 87, one over MAX_DEAD_TRI = 86, and the "Documented t27c subcommands exist" check has been red on master since #12249 (see #12218). Every PR inherits that red through its merge with master. The paragraph now says the command is planned and not built. That is the checker's own DECLARED wording, and it is what #12218's follow-up says too. The gate run locally (TRI_T27C, TRI_BIN set) prints the 86 recorded ceiling and exits 0. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(viz): use the same tri viz graph wording as #12280 (Refs #12173) #12280 (Refs #12218) fixes the same line of VIZ-DISTILL-GRAPH.md. Its wording is "`tri viz graph`, not built yet". This branch now uses exactly that text, so the two PRs merge cleanly in either order and nothing conflicts. The gate run locally still reads 86 against the ceiling of 86. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(t27f): guide says plainly that no viz spec is sealed yet (Refs #12173) The bee review on #12282 read "the contract, the plan and the reference did" as a claim that those three specs carry seals. They do not: the only viz seal under .trinity/seals/ is viz_VizDistillGraph.json. The sentence now says that none of the three is sealed. The two "(checked 2026-10-25)" notes about gaps in the reference petal now name what they were checked against, the reference as merged in #12193, so a reader can verify them later. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #12174, see #12173.
First milestone of the interactive-visualizations epic: the reusable contract, the plan for all 27 petals, and one complete reference petal. Everything is a pure
.t27spec tested byt27c test-report; drawing it in the glass window is the next child issue.What lands
specs/t27f/viz_contract.t27: the contract every petal visualization is drawn through.VizStatewith one pure reducer covering hover, select, step, param, keyboard focus, back and reduced motion.vh_embed, so a widget can be reused inside another petal without copying its logic.specs/t27f/petal_viz_plan.t27: 27 rows, one per petal, each with the user question, the source-of-truth spec (every path checked), the representation kinds and a status: 1 reference, 23 planned, 3 deferred with a reason (E, R, Y). Petal G follows master's new AGENTS page (t27os#3) and is LIVE. The host-page column is held equal toagent_parts.page_kind.specs/t27f/petal_specs_viz.t27: the reference petal S SPECS overtutorial_08_graph(), whichspec_graph_deriveproves equal to the exporter's output on the real tutorial file. It has three linked views (node-link, adjacency matrix, edge list), message passing with sum, mean or max over 4 layers, an inspector, and an edge toggle that drops truth to DEMO. Every view's hit test inverts its layout, and one state marks the same entities in all three views.psv_reps(), and a test asserts the two are equal, so the plan cannot drift from the petal.Measured
Command:
t27c test-report specs/t27f/<spec>.t27(the reference path, t27c + zig). The native backend agrees on the petal:t27b test specs/t27f/petal_specs_viz.t27gives 67 passed, 0 failed, 2 invariants held.Not claimed
No host draws a
VizStateyet. The glass window page for petal S is the next child issue of #12173.🤖 Generated with Claude Code