tegwick
bf72a1863a
Some checks failed
ci / check (push) Failing after 4s
Provenance (tier S, one paragraph in lieu of survey and ADR): the human check that kept INTENT stage 1 open was run and the drag was broken. Root cause, worth more than the instance: drop targets were ids, and an id must be unique, so exactly one element could ever be seat-0. The relationship-graph circle took it and the seat card that every action card's own text points at -- 'drag Attack onto a seat' -- silently had none. A seat is drawn twice and both drawings are the seat; the document model could not express that. Drop keys are now data-drop. Any number of elements may carry the same key, so a seat is droppable on its card and on its graph node. Measured on a live server: seat-0/1/2 each appear twice, id survives only on cb-status which is the one element the script looks up, and down=action-attack&up=seat-1 returns ok. Second defect: a drop on nothing returned without posting and without touching the status line, so a broken target was indistinguishable from a working page. resolve already refuses rather than defaulting, which is right; refusing SILENTLY is not. The page now reports the raw fact -- 'took action-attack, let go over nothing droppable' -- which names elements, not moves, so ADR-0007 control 5 holds. And the honest part: the general check added here -- every offered affordance names a key that exists, driven through Policy::choose over four real bot games -- does NOT catch the reported defect. seat-0 did exist, on the graph circle. It is kept because a wholly absent target is a real class, and paired with a targeted regression test that does catch it. Three mutations, each red for its stated reason, including the reported defect reintroduced; only the targeted test fires on that one. A cb-play assertion matched id="action-ground" as a substring while describing itself as checking the page; rewritten through drop_keys. make all exits 0. Stage 1 stays open: verified by tests, mutation and a live server, not by a human dragging. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
181 lines
7.3 KiB
Markdown
181 lines
7.3 KiB
Markdown
---
|
||
id: CB-WP-0016
|
||
kind: product
|
||
title: "The drop target that was never there"
|
||
status: done
|
||
---
|
||
|
||
# Purpose
|
||
|
||
```
|
||
structural tier S (a defect fix inside an existing capability — no new
|
||
port, no canonical interface, no new dependency, and
|
||
the check it adds is a product test, not a control
|
||
gate)
|
||
chaos d4 = 3 → no override
|
||
declared tier S
|
||
```
|
||
|
||
Declaration 11 of 12. Tier S: one provenance paragraph in the commit
|
||
message, no survey and no ADR. Steps 4–5 are not skipped.
|
||
|
||
## Found by the human check, and by nothing else
|
||
|
||
CB-EV-0012 §4 kept INTENT stage 1 open on one action the loop could not
|
||
perform: run `cb-play --serve 0`, open the URL, confirm the table reads and
|
||
a drag works. The maintainer ran it on 2026-08-02. **The table reads. The
|
||
drag does not.**
|
||
|
||
Diagnosed against the running server before any code was changed:
|
||
|
||
```
|
||
POST down=action-attack&up=seat-1 → ok (the game advanced)
|
||
POST down=action-attack&up=action-attack → "not a legal move here"
|
||
```
|
||
|
||
So the socket, the token guard, `resolve`, and command dispatch are all
|
||
correct. The defect is in the emitted page.
|
||
|
||
**`seat-{n}` ids exist only on the ~26 px circles inside the relationship
|
||
graph** (`<g id="seat-0">`, `doc.rs:185`). The visible *seats* section
|
||
cards are emitted by `player_card` as `<div class="card" data-viewer=…>`
|
||
with **no id**. Every action card says *"drag Attack onto a seat, a
|
||
problem, or the table"*, and the seat it points at is not a target.
|
||
|
||
`#table`, `problem-{n}` and `freedom-{n}` are real targets, so the page is
|
||
not wholly inert — the most natural target is.
|
||
|
||
## Why every test in the repo passed
|
||
|
||
`jsrun::gesture(html, down, up)` calls `__down(id)`, which synthesizes
|
||
`{ target: { id: id } }`. **It feeds element ids straight in and never
|
||
hit-tests.** So it establishes *"the script posts the ids it was given"*
|
||
and never *"the page has an identified element where the instruction says
|
||
to drop."*
|
||
|
||
The 42-path coverage gate is blind for a related reason: it asserts each
|
||
view field is **present in the parsed document**, which a `<div>` with no
|
||
id satisfies perfectly.
|
||
|
||
CB-WP-0015 closed a clause whose test was named for a property it did not
|
||
check. This is the same shape one layer out — a harness answering a
|
||
narrower question than its name implies — and it is the argument for
|
||
having kept the human check as a blocker rather than closing stage 1 on
|
||
green tests two passes ago.
|
||
|
||
## Task: make the seat a drop target, and keep ids unique
|
||
|
||
```task
|
||
id: CB-WP-0016-T01
|
||
status: done
|
||
priority: high
|
||
```
|
||
|
||
The seat card must carry the id `affordance()` already names for it.
|
||
`seat-{n}` is currently taken by the SVG node, and duplicate ids are
|
||
invalid HTML — so one of the two has to change, and **both must remain
|
||
droppable.** The graph circle works today and someone will have learned it.
|
||
|
||
Decide where the aliasing lives. Rust decides what a pointer fact means
|
||
(ADR-0007 D5), so widening what `resolve` accepts is in bounds and
|
||
inventing a new id vocabulary in the page is not.
|
||
|
||
**Control:** dropping on the seat *card* and dropping on the graph *node*
|
||
must produce the same command, and a test must go red if either stops
|
||
resolving.
|
||
|
||
**Done 2026-08-02.** The aliasing was not needed, because the constraint
|
||
that caused the defect was removed instead.
|
||
|
||
**Drop keys are now `data-drop`, not `id`.** An `id` must be unique, so
|
||
exactly one element could ever be `seat-0` — the graph circle took it and
|
||
the card went without. Any number of elements may carry the same
|
||
`data-drop`, so a seat is droppable on both of its drawings. Measured on
|
||
the live page: `seat-0/1/2` each appear **twice**, and `id` survives on
|
||
exactly one element, `cb-status`, which is the only one the script looks
|
||
up. `down=action-attack&up=seat-1` against a real server returns `ok`.
|
||
|
||
The mutation that reintroduces the reported defect — the seat card losing
|
||
its key — goes red, **and only the targeted test fires**; the general
|
||
existence check stays green. That is the point of T03's finding.
|
||
|
||
## Task: a gesture that lands nowhere must say so
|
||
|
||
```task
|
||
id: CB-WP-0016-T02
|
||
status: done
|
||
priority: high
|
||
```
|
||
|
||
`SCRIPT` currently does `if (!down || !up) { down = null; return; }` — no
|
||
POST, no status line, nothing. That silence is why this defect survived a
|
||
human sitting in front of it.
|
||
|
||
The Rust side already holds the right principle: `resolve` refuses rather
|
||
than substituting a default, because *"a drag that means nothing must mean
|
||
nothing, not the first legal move."* Refusing is correct. Refusing
|
||
**silently** is not.
|
||
|
||
Make the outcome visible. Keep ADR-0007 control 5 intact — whatever the
|
||
page reports must still be raw pointer facts with no game vocabulary, and
|
||
the existing body-shape assertion must still hold.
|
||
|
||
**Done 2026-08-02.** The page now writes *"took action-attack, let go over
|
||
nothing droppable"* to the status line. It names elements, not moves, so
|
||
control 5 holds and the body-shape assertion is untouched. Restoring the
|
||
bare `return` turns
|
||
`jsrun::tests::a_drop_on_nothing_reports_instead_of_going_quiet` red with
|
||
*"a drop on nothing said `\"\"`"*.
|
||
|
||
The JS DOM stub had to grow a real `getAttribute` and a null-parent node
|
||
to model a pointer landing on nothing — it previously could not express
|
||
the case at all, which is part of why the silence was invisible.
|
||
|
||
## Task: the check that closes the class
|
||
|
||
```task
|
||
id: CB-WP-0016-T03
|
||
status: done
|
||
priority: high
|
||
```
|
||
|
||
**For every legal command, both ids `affordance()` produces must occur in
|
||
the emitted document.** Pure Rust, no JS engine, no browser, deterministic.
|
||
It fails on today's tree and it would have caught this the day the
|
||
renderer was written.
|
||
|
||
This is the deliverable that matters — T01 fixes one instance, this closes
|
||
the class. An affordance naming an element that does not exist is exactly
|
||
the harness-does-nothing shape, and it currently has no detector at all.
|
||
|
||
Then `evidence/CB-EV-0014-*.md`:
|
||
|
||
- **What the human check bought**, priced honestly against what it cost to
|
||
keep stage 1 open for two passes.
|
||
- **Whether stage 1 closes now.** It needs the maintainer to drag again;
|
||
say so rather than closing it on a passing test, which is the failure
|
||
mode this whole pass is evidence for.
|
||
- **Quote CB-WP-0015's final cost, not this pass's** — and note that
|
||
CB-EV-0013 §5 found the self-quoting rule fixes the wrong boundary, so
|
||
say plainly whether the figure has settled.
|
||
- **Chaos: declaration 11 of 12.** The calibration window closes next
|
||
declaration.
|
||
|
||
**Done 2026-08-02.**
|
||
[CB-EV-0014](../evidence/CB-EV-0014-the-drop-target.md). `make all`
|
||
exits 0.
|
||
|
||
- **The general check does not catch the reported defect, and the evidence
|
||
says so plainly.** `seat-0` existed — on the graph circle — so an
|
||
existence check over the whole document cannot tell that the element the
|
||
user is *pointed at* is not the one that works. It is kept because a
|
||
wholly absent target is a real class, and paired with a targeted
|
||
regression test that does catch it.
|
||
- **Three mutations, each red for its stated reason**, including the
|
||
reported defect reintroduced.
|
||
- **Stage 1 does not close here either.** The fix is verified by tests, by
|
||
mutation, and against a live server — not by a human dragging, which is
|
||
the standard that found it.
|
||
- **A cb-play assertion drifted**: it matched `id=\"action-ground\"` as a
|
||
substring and silently described itself as checking the page. Rewritten
|
||
through `doc::drop_keys`.
|