check:composite-freshness found stale on main: none. Rebuilt, tree + SRI
regenerated; the landing precondition (fresh, one folder, SRI) passed in main-capture scan.
The lockfile's workspace section had "7" for @capacitor/android, cli and core, where package.json says "^7.6.9". --frozen-lockfile accepts that (7.6.9 satisfies both), but a plain `bun install`, as the HOWTO shows, rewrote the three lines, so a new contributor's checkout was dirty before their first edit and their first push would carry the lockfile. Regenerated with bun 1.3.14: exactly those three lines change, installs report no changes, and --frozen-lockfile still passes. Found by the fourth fresh reader.
check:poster-browser-equiv is chained in gates:pre and exited 75 NOT MEASURED
every run. Its input, out/_records.json, is gitignored and written only by
generate.mjs against a live API, so on CI and on any box without credentials the
gate never looked at anything. The runner reported that honestly rather than as
a pass, but an identical NOT MEASURED every run is indistinguishable from an
unwired gate -- and this is the gate that proves a sheet generated in the browser
is byte-identical to the one Node writes.
An equivalence check needs the SAME input on both sides, not fresh input, so a
committed fixture loses nothing it was measuring. The fixture is the 124 public
records the two arms actually consume: Nepal's 34 for the country arm, and the
90 slugs the Gujarat zone selects for the zone arm. Same fields the served maps
show; no other data.
Verified equivalent to the full 2,106-record cache: identical zone pins in the
same order, identical Nepal set, count 90, pop 3,164,770, series {2209, 28, 1}.
Record ORDER is load-bearing -- it drives draw and label order -- so the fixture
keeps the cache's order. A first cut sorted it: every aggregate matched and the
pin SET matched, while the order differed. Sorting a fixture would have quietly
changed the sheet it produces.
generate.mjs gains RECORDS_FILE so a caller can name the offline input instead of
only the API-written cache. It still fails closed without an API when no offline
file is given. The gate passes it the very file the browser arm read, so both
sides are fed the same records by construction.
The gate now says which input it measured, because the full cache widens coverage
to all 175 countries where it exists and the fixture is the floor everywhere else.
Two-sided, on this tree: with the fixture it PASSES byte-identical on both arms
(834,174 and 921,522 bytes). Dropping the province boundaries at
posterSheets.js:206 -- the defect this gate's header records, where a country
beside India or China printed without them -- makes the country arm 251,576
against Node's 921,522 and the gate exits 1. Restored, it passes again.
research-map-plus-grid and regions-mfe draw a map through a child element, so their isMap is false,
and their published snippets had no engine preload and no scriptMapbox. A pasted composite still
draws on Mapbox given provider "mapbox" and a token (measured), so this is the snippet's
convenience, not a broken map.
generate-tree now decides after every profile's isMap is known: a composite (nested-elements-mfe)
whose OWN composes reach a map profile gets the map form of the default snippet and the Mapbox
variant. It reads the profile's own composes, never the bundle's hoisted list, which the bundle's
plain leaves (charts, data-tables, the poster wall) share while drawing nothing. The gallery shell
is left out until its passing of provider to its tabs is measured. Regenerated: exactly those two
profiles' embed changes, in both manifests; nothing internal leaks into the output.
check:snippet-engine asserts the same rule on the built manifest (21 map snippets, 19 maps + 2
composites). Its self-test adds a composite stripped of its variant, which fails. Against the
manifest as it stood before this change, the widened gate fails on exactly these two composites;
on the regenerated one it passes.
Every owner's sweep has landed. The two remaining lines were public-hygiene-check.mjs's pattern list, which must spell the role word in order to catch it on the way out, as this gate's own file must. It is exempt by name in a one-file list with its reason, not by any wider rule. The baseline is now empty: from here, every line that names the role is a new reference and fails.
The pin click handler was registered on two layers that always overlap: the
visible pins and the invisible hitbox drawn over them. A layer-scoped click
fires once per registered layer with a feature under the point, and every
visible pin has a hitbox twin at its own point -- same source, larger radius,
added alongside it so taps are easier to land. So every pin click ran the
handler twice, 3ms apart. Measured on the leaf at 1440, on two different pins,
both reporting both layers at the click point; exactly twice, never more.
That already shipped as a visible bug once: the second call overwrote the
legend state the first had saved, so closing a people-group detail reopened a
legend the reader had deliberately collapsed. The store is now idempotent, so
that particular symptom cannot return, but the duplicate call was the cause and
is still here -- anything that counts, toggles, appends or reports from a pin
click is wrong by a factor of two, and nothing in the code warns the next
author.
Bind the click to the hitbox only. It covers every click the visible layer
would catch: both layers are created together by addLanguageFamilyLayer, from
the same source, and the two places that filter the hitbox mirror the pin
filter rather than narrowing it, so the two layers always carry the same
features. The handler reads only properties and geometry, which are identical
between them. Cursor handlers stay on both layers -- mouseenter and mouseleave
are idempotent, so they were never part of this.
Checked before relying on it: exactly one addLayer creates each of the two
layers, both inside addLanguageFamilyLayer as consecutive statements with no
branch between them; no flag gates the hitbox; no removeLayer anywhere touches
either; and the only layout writes on the pin layer set circle-sort-key. The
invariant is now stated at the binding, because it is not visible from the call
site: a pins layer added without a hitbox would get no clicks.
Also fixes a dead trap in the same file. The people-groups-pins branch called
the handler with no argument, so it fell back to the language-family pins and
bound the wrong layer. Nothing creates that layer today -- the id occurs only
in this file -- so the branch never runs, but whoever first adds it would have
got a duplicate binding, or no click handler at all if the language-family
layer were absent.
DataTable's sheet heading fell back to a hardcoded English 'Detail', one word and so outside check:hardcoded-english's reach. It now reads aria.rowDetail ('Row detail'), which this component already uses for the dialog's own label: translated in all 11 locales, and it brings no new catalogue area into any bundle (a dashboard.* key would have added the whole dashboard area to every bundle that shows a table).
A poison run answered 404 to posters/<id>.html for the browser only, and the gate still PASSED.
It counted the .html REQUEST, and its failure pattern (not available|failed|blank|error) never
matched what the wall then shows, "poster not found (404)"; the HUD reader did not even collect
that text. So it could not fail on the one path it exists for.
It now requires the .html to ANSWER 200 and a print frame to actually OPEN (counted at creation
with the appendChild hook the degrade gate already uses, since the frame removes itself after
printing). "not found" is read as the failure it is, in the HUD reader, the poll and the verdict.
Its output reports both outcomes. The browser runs are rega's: the positive arm must still pass,
and the 404'd source must now fail.
rega's L12 photo probe read host.__wallLod().cardPx as 0 after 46 wheel ticks: the field was set once when the cards were first laid out and never again, since the canvas renderer replaced the LOD component that used to update it. The canvas now records one card's on-screen width at every frame it draws (stats.cardPx, both the grid and the poster), and the hook returns that. The container's header comment no longer claims Font Awesome is loaded: the canvas draws its own icons, and d3 is pinned and kept off window.d3 (src/wallD3.js).
The source change needs its bundle: main-capture scan refused the landing with
COMPOSITE-FRESHNESS, committed 26db9ec7f664 against a rebuild of 6a08c67b57e4.
BUNDLE=cities-internet bun x vite build, then bun run tree and fix:sri. One
integrity hash moves, nijTwn… to I39zqe…, in the cities-internet rows of both
manifests. The four MapLibre and Mapbox CDN hashes are untouched, and no other
bundle's path appears in any changed line.
rega saw "Other" in ar and zh. The reported cause — `c.region || 'Other'` at
:65 — is not one: of 175 rows in connectivity.json NONE has a falsy region, so
that fallback never fires. The "Other" section is one row (United States) whose
region IS the string "Other". Patching :65 would have changed nothing on screen.
The render site is `{{ sec.name }}`, and it printed the raw data value for ALL
NINE regions, so this was nine English headings rather than one.
All nine already had translated keys — dashboard.wagfRegion.* for the seven WAGF
regions, dashboard.other, wall.na — in all 11 locales. None needed adding.
THE LABEL IS PROJECTED AT RENDER, NEVER AT GROUPING. sec.name keys the Map, the
:key, and the collapse state, so translating it where the rows are grouped would
reset every open section on a locale switch and filter on a value no row carries.
That is the value/label trap useFacetAggregation.js documents at OTHER_KEY; this
follows the same shape.
The name->key map is built by asking the ENGLISH catalogue what each key says,
rather than writing the English phrases here: the data's region strings are the
catalogue's own English values, so English is the join and never has to be
duplicated. Written the other way it added two check:hardcoded-english findings
to a file that had none — an English phrase used as a lookup key reads exactly
like a user-visible string.
tStatic(path, 'en'), not t(). t()'s signature is (key, params, plural), so
t(k, 'en') passes 'en' as params and returns the RUNTIME locale's value: the map
would be built from Arabic values and match no region at all.
The keys are written out in full because internal/i18n-areas.mjs bundles an area
only when its name appears literally in a reached file. cities-internet shipped
14 areas and neither dashboard nor wall; assembling these at runtime would have
left all nine unresolved in the shipped bundle while working in the dev tree.
Measured both ways — written out, the bundle takes 16.
`|| 'Other'` stays as a data guard so a future row with no region cannot key the
Map on undefined, and 'Other' is the sentinel dashboard.other projects.
9 of 9 headings translate in ar and zh, verified against the real catalogues.
rega's L15 (e7f343d4c, ar + zh): every PPLR map view's fullscreen button read "Toggle fullscreen" in English, title and aria-label. It is PPLR's own control (FullscreenToggleControl in src/utils/pplrMap.js, not MapLibre's), and it wrote the English itself. It now takes the shared buttons.toggleFullscreen, already translated in all 11 locales, when the map is built.
cities-list.vue grouped region-less rows under a hardcoded 'Other' section heading, a single English word, which is the hardcoded-english gate's blind spot. It now reuses keyCities.posters.otherRegion (all 11 locales, landed in R25). The same literal in cities-internet's connectivity-list, where rega's R25 look found it in ar and zh, is gospel's and goes to gospel. The SSR check (check-keycities-i18n.mjs) passes; key-cities is gated, so nothing in public/ changes.
check:composite-freshness found stale on main: template-bundle. Rebuilt, tree + SRI
regenerated; the landing precondition (fresh, one folder, SRI) passed in main-capture scan.
check:coverage-consistent:chained sets the link-age threshold out of reach. The proof that this narrows its scope and does not disarm it is the poison arm: its clock is computed RELATIVE to the threshold, so all 15 clauses (A-O, L included) still fail when poisoned at any setting (regions measured it at 99999). That proof was an orphan; it now runs on every landing, like the other gates' self-tests. 1.6 s.
check:dxlife walked the whole root-dx-life umbrella, so git worktrees of doxa-life (wt-rg-*) and
a dangling one (research-map-clean, whose .git points at a path that no longer exists) read as
+4,524 "added" files and held it red while the clones were untouched. It now skips any directory
whose .git is a FILE beginning `gitdir:` (keyed on that file, not on `git worktree list`, which
cannot see the dangling one) and .cache with the other generated dirs, and compares the baseline
only within that scope. Result against the 2026-09-20 baseline: 2,844 clone files, +0 -0 ~0. So the
law has held since that date, and the baseline stays as it is: re-taking it would reset the evidence.
Nothing is deleted; research-map-clean may hold the only copy of its work. With no clone on the
machine it reports NOT MEASURED (75) instead of exit 2, so chaining it cannot fail other machines.
check:coverage-consistent fails when the oldest verified link is over 45 days old: its verdict
depends on the calendar, and chained it would turn gates:pre red on 2026-11-08 with no change
(regions measured the margin). Every other clause depends on code and data and is worth running
on every landing, so gates:pre runs check:coverage-consistent:chained (the age knob set out of
reach). The plain script is declared on demand as the freshness run, with that reason. Its poison
self-test still proves every clause can fail.
gates:pre 94/0/2 of 96. The orphan baseline shrinks to 105; 7 checks are declared on demand.
Collapse the legend, tap a pin, then close the detail: the legend comes back
open, though nothing asked it to. Reported against the leaf research map;
reproduced at 1440 in en, collapsed-first.
selectPeopleGroup saved the pre-detail legend state on every call that carried
a group, not only on the call that entered detail. The second call therefore
saved the 'open' the first call had just forced, on top of the 'collapsed' it
correctly saved on entry, and closing the detail restored that 'open'.
The measured part is why a second call happens at all. The pin click handler is
registered per map layer, and every visible pin has an invisible hitbox twin at
the same point -- same source, larger radius, added alongside it so taps are
easier to land. A layer-scoped click fires once per layer with a feature under
the point, so a single pin click calls selectPeopleGroup exactly twice, 3ms
apart. Measured on two different pins, both reporting both layers at the click
point; never more than twice. So ONE pick was already enough to lose the
collapsed state -- it never needed two, which is why this reproduced every time
rather than intermittently.
The fix is to make the method idempotent: save only when entering detail (no
group was selected before), restore only what was actually saved, then clear it
so a later detail cannot inherit a stale value. The duplicate binding is a real
defect in its own right and is tracked separately; this method should not have
depended on being called exactly once.
Verified two-sided against the shipped source, driving the real method body
rather than a copy: seven scenarios pass on the fix and four fail with the
previous body restored, while the scenarios that must be insensitive stay green
both ways. The poisoned run reproduces the browser measurement value for value
-- saved collapsed, overwritten to open by the second call, restored open.
Trace in _qa-cycle so the next reader can re-run it.
From regions' triage of the headless orphans, every candidate re-run on this round's tip before
it was chained (all exit 0; the slowest, no-conflict-markers, 8.2 s):
gates:pre click-sheet coverage-consistent dashboard-filter-a11y filter-multivalue hover-channel
kit-options-reachable layout-key-namespaced legend-mirrors map-unmount-race
no-conflict-markers series-isolation snapshot-guard stale-sweep-claim
surfaces-read-the-tier translator-contacts (source readers)
gates:post embed-base link-coverage no-public-binds shared-reached tree-weight (built tree)
stale-sweep-claim reports SKIP while 1 of 1184 gap codes is undecided: it measures once the pool is
fully read, and the wave shows it as not measured until then, never as a pass.
Declared on demand instead: catalog-currency and published (they read the published host),
composite (minutes of rebuilds; landing's fresh step does this per landing), hardcoded-english
(report-only while its findings are triaged; its self-test is chained), and shot-fonts (it checks
the SCREENSHOT MACHINE's fonts, so chained it would fail a contributor's gates for their fonts).
Two proposals were declined: check:settled stays on demand (it fails inside any build by design,
and passing standalone is how it was wrongly chained before), and shot-fonts as above.
The orphan baseline shrinks 131 -> 106. gates:pre 93/0/2 of 95, gates:post 91/0/16 of 107.
A fresh reader given only the README and the HOWTO reached the Studio and a plugin's /__leaf page with
nothing else needed, but one step. After "port 8090 is already in use ... set DX_PB_PORT", neither
doc said the setting is read when bun run dev starts, so the new port only applies after a restart.
The README copies .env.example before starting the server, and the port message appears only once it
runs, so the steps looped with no stated way out. All three places that mention DX_PB_PORT now say to
pick a free port and restart.
The two scripts arrived without //scripts entries. The text is regions', written to match the neighbouring entries; its '8 self-tests' was re-read against the file this round carries (sha256 a886af2dca66e012, 124 lines).
check:no-orphan-gates only REPORTED scripts outside the chains, and the list grew to 87 main checks.
That day two of them hid real defects: check:i18n-never-translated, while a dead key identical in all
11 catalogues sat on development for hours, and check:auto-engine, which asserted a retired engine
rule that nothing ran. The report also matched by substring, so check:hardcoded-english counted as
run because its :self-test is in gates:pre.
The arm now decides. Every check:/verify:/smoke: script must be:
reached by exact name from gates:pre, gates:post, the browser suite's GATES or the build,
following what those scripts `bun run`;
on demand in internal/fixtures/on-demand-checks.json WITH its reason (check:settled, whose
reason this gate already documented: it fails inside any build by design);
baselined in internal/fixtures/orphan-checks-baseline.json, today's 131 orphans, for their
owners to chain, declare or delete. That list only shrinks (--shrink).
A new unchained script fails by name. A self-test runs first on every invocation. Proven two-sided on
the real tree: a planted unchained script exits 1 naming it; swapping exact matching for the old
substring test makes the self-test refuse; restored, it passes.
A browser given a 390px viewport still lays the page out at the 980px desktop
fallback unless the document declares <meta name="viewport">. Measured in
Chromium at 390x844 with isMobile:true:
without the meta: window.innerWidth = 980 (body 964px)
with the meta: window.innerWidth = 390 (body 374px)
Nothing warns. The page renders, the probe reads its numbers, and the check
goes green having measured a viewport no phone has. A full set of mobile runs
was taken this way before the layout width was ever compared to the requested
one, so the conclusions drawn from them described the wrong viewport.
Three gates already read window.innerWidth and none compares it to the width it
asked for, so all three would pass at 980 while reporting 390. This asserts the
property that makes their numbers mean what they say.
Checked on the built artifact rather than the generator template: a template
containing the right string proves nothing about the HTML that shipped. No
server needed.
width=device-width is required specifically, not just the tag: a meta carrying
only initial-scale still lays out at 980, so matching the tag name alone would
leave the gate unable to fail on its own founding bug.
The 28 sheets under assets/posters/ are exempt by their stated shape - a fixed
paper canvas (.stage{width:1500px;height:2000px}) has no mobile layout to get
wrong. Keyed on that property, not the directory, so a poster keeps its
exemption if it moves and an app page in that folder does not inherit one.
Mutation-tested against the real tree: stripping the meta from a shipped page
fails it, and replacing it with an initial-scale-only tag fails it. 8 self-tests
cover both, plus the exemption and its inverse.
check:gospel-whiteboard does not cover the adoption in the previous commit. I
poisoned GiMap into whiteboard:true — the exact regression that blanks the
production geographic map — and it still exited 0. It asserts chart geometry
(projection, radii, the bounded gap) and imports no component, so a call-site
option is invisible to it. Its pass was never evidence for this change.
check:board-map-options asserts what actually regresses: every
createWhiteboardMap call states `whiteboard` explicitly, and a geographic caller
(whiteboard:false) also states `boundaries`. A caller that inherits defaults
gets the whiteboard's, which are the inverse of a real map's.
Two-sided, because a check that cannot fail is not evidence: poisoned, it exits
1 naming GiMap.vue; restored, 0. Both negative controls run in the gate itself —
a call omitting the option must be judged non-compliant, and a compliant call
must not trip it, or the gate would just ban the helper.
Registered in all three places: script, gates:pre, //scripts. gates:pre is 77.
boardEngine.js was createWhiteboardMap under another name: the bodies are
identical line for line, and every import it made already came from library.
It is deleted and the four components import @map/maps/whiteboard-map.js.
The copies were NOT interchangeable, which is the part worth recording. Their
DEFAULTS are opposites, because the library file was extracted for the
whiteboard case and boardEngine served both map kinds:
boardEngine whiteboard:false boundaries:true centre [60,20] zoom 1.4
library whiteboard:true boundaries:false centre [0,0] zoom 1
Three of the four callers pass all four options explicitly, so they were never
at risk. GiMap — the geographic map — passed only a token and took every
default. Swapping its import alone would have set whiteboard:true, hiding every
basemap layer, dropping boundary attribution and moving the camera: a blank map
in production, from a change that reads like pure de-duplication.
So GiMap now states its four options. Every effective option is unchanged from
today's values, verified field by field, and the call reads the same against
either module.
Twenty-six references in capitals, which the earlier case-sensitive pass
could not see. Every requirement, date and measurement is kept: the 2026-09-23
instruction that removed the engagement and status filters, the MapLibre ask
with its Mapbox-compatibility condition, the blue-water-black-galaxy style,
the red ramp and the bin that collided with the heartbeat zone, the white
rings around the blue dots, the zoom that did not work inside the bounded map.
Two that are not quotes and were handled differently. The reduced-motion gate
opens with a medical reason — a reader of this map is photosensitive, and a
camera that sweeps the globe on every filter is a strobe — which is the whole
justification for the gate, so it stays as a fact about a reader rather than
about a person. And BpResearchHeader's "CAT-1 DRIVER CONTRACT" was a code
term; it is now the CAT-1 HEADER CONTRACT, checked for other references first.
Eleven lines broke and were found by re-reading every changed line, not by the
count. Removing an attribution left seven continuation lines holding the
closing half of a quotation that no longer opened, and four sentences carried
a pronoun for the speaker who had just been removed. The reference count was
correct at every point while the prose was wrong.
grep -wi driver across the three plugins: 0. The widened gate reports 40
swept, 14 still baselined. All six SFCs parse.
check:composite-freshness found stale on main: none. Rebuilt, tree + SRI
regenerated; the landing precondition (fresh, one folder, SRI) passed in main-capture scan.
Lines swept on development since the capitals form joined the gate: research-dashboard, people-groups-grid, my-upg-100-list, regions-dashboard and others. --shrink drops only what is gone; nothing is added.
The check was defined and in no chain. So gospel.json's sheet.isoLine, identical in all 11
catalogues and referenced by no code, stayed on development through every landing since it
arrived; it surfaced only when someone ran the check by hand. This round deletes the key and
chains the check beside check:i18n-keys-exist, so it lands green and stays guarded.
Run before chaining, on this round's tip: PASS (18.8 s in the chain), 2 keys deferred pending a
ruling. On development without the deletion it fails on that one key, as gospel measured.
The suite lists a gate only after it has been run, so these waited for one. On 5b05f019e, in the
browser slot:
check:auto-engine PASS. no token -> MapLibre, 0 Mapbox requests; a host token alone ->
MapLibre, 0 Mapbox requests; ?provider=mapbox + the token -> Mapbox,
carrying it
check:engine-contract PASS. 2 providers x 3 load states: the engine asked for, or a visible
failure, never a substitute
Both entries record those results. check:engine-contract cannot demonstrate a failure at runtime
without the defect rebuilt into a bundle; its falsifiable arm is the source gate's self-test, so
check:no-silent-engine:self-test now runs in gates:post beside the gate it proves. It was chained
nowhere.
Its case 2 expected a host token alone to draw Mapbox, the rule until 2026-09-28. It was in no
chain, so it never ran against the change that made that false, and nothing else measured the case
the change is about. The cases now:
1 no token: MapLibre, zero Mapbox requests, painted, no prompt (unchanged)
2 a host token alone: still MapLibre, painted, zero Mapbox requests
3 ?provider=mapbox with that token: Mapbox, carrying that token
It joins gates:post beside check:leaf-engine-per-provider, together with check:engine-contract
(gospel's resources-map uses the engine it was asked for, or fails visibly), which was also in no
chain. Both are browser gates: NOT MEASURED outside the browser slot, measured in it.
Since 2026-09-28 resolveMapProvider returns MapLibre for 'auto' whether or not a token is present;
only an explicit provider selects Mapbox. Four comments still said "Mapbox when a token is
available": useMapProvider.js's own header (and its note on the `token` argument),
useMapInstance.js at the provider option, check-provider-switch.mjs's contract note, and the
PROVIDER note that generate-tree writes into every leaf page.
The leaf page's note now says what the page does: it finds a token first only when one is on offer,
because place search uses one on either engine, and the element decides the engine. The 64 leaf
pages are regenerated and differ only in that note. The three JS files change in comments only
(compared minified, without renaming).
The README says a plugin folder holds "an index.html for previewing it", and a fresh reader
followed that to /plugin-designs/<plugin>/index.html under `bun run dev`. leafRewrite reads any
<a>/<b>/index.html as <bundle>/<profile>, so the request reached the leaf route as bundle
"plugin-designs" and answered only {"error":"unknown bundle"}, which pointed nowhere.
The 404 now carries a hint: for a folder that is a plugin, its own page
(/__leaf/all-people-groups/all-people-groups, taken from the same registry the gallery uses);
otherwise the /__leaf/<plugin>/<profile> shape and the manifest that lists every plugin. The
status and the error stay as they were. The README now names the page as well.
A fresh reader given only the README and HOWTO-join-the-table got through install, bun run dev and
the Studio. One gap: no address for a single plugin's page. The Capability 1 fold says each plugin's
folder holds "an index.html for previewing it", which reads as /plugin-designs/<plugin>/index.html,
and that 404s under bun run dev. The page is /__leaf/<plugin>/<profile>, and the Studio's list is
/__plugin-gallery/manifest.json. Both are now named.
Three smaller notes from the same reader:
- the README copied .env.example to .env while the HOWTO uses .env.local. Both work and .env.local
wins; the README now uses .env.local too, which also keeps .env free for compose.yaml;
- the Mapbox token read as required; it is one of two ways to get a background;
- the 8090 port hint (DX_PB_PORT) was only in the HOWTO; it is now named where the README first
mentions PocketBase.
Also, the engine rule (was a follow-up commit, f0e965e4d):
Since 2026-09-28 useMapProvider.js resolves `auto` to MapLibre, and Mapbox is an explicit choice:
"provider": "mapbox" plus a token. The docs still said "Mapbox when a token is available":
- README's map-engines row;
- how-to.md's provider section;
- the plugin-designer skill's props table, which even says to confirm in useMapProvider.js.
All three now state the current rule. So do four lines that told readers to set provider maplibre to
avoid needing a key: README's Get it running and its "No token" fold, .env.example's token comment,
and reference.md's no-key row.
Case-insensitive git grep "Mapbox when a token" over README.md, documentation/ and skills/: 3 before, 0 after.
sheet.isoLine turned check:i18n-never-translated red: "· ISO 639-3" is identical
in all eleven locales, because a standard's designation correctly is.
The suggested fix was an ALLOWED entry, but the key has ZERO call sites — I
authored it during the label batch and never wired it. An exemption would have
made the gate quiet about a string nothing renders, which is worse than the red:
it spends the exemption list's credibility on a key that should not exist.
The line it was meant for (charts.vue:577) keeps "· ISO 639-3" as plain text, and
check:hardcoded-english does not flag it — correctly, since a standard's name is
not prose. So there was never anything to translate here.
Removed from all eleven. never-translated PASSES, parity and keys-exist unchanged.
studio's judgement on the nineteen REGION_OF findings: the English values are
identities, not reader text. Node ids derive from them, and every surface
that shows one translates it at render — upg-100-list.vue:440 and
poster-wall.vue:129 both read REGION_KEY[english] and call t(`poster.${…}`).
The exemption is conditional, and the condition is the whole rule. It holds
only for a name that IS a key of REGION_KEY. A region added to REGION_OF
without a matching key falls through the `|| english` arm and renders English
in all eleven locales, which is the defect upg100 already shipped once with
Americas and Europe. So the gate keeps working as the table grows rather than
blessing the file.
Deliberately not "object values" and not "legend-tree.js", as studio asked:
either would exempt the next table nobody has checked.
The parser had no newline requirement in the end, and finding out why is the
part worth keeping. The first version anchored on a closing brace at the
start of a line, so it never matched a REGION_KEY written on ONE line —
regionKeyNames returned null, the rule exited early, and BOTH two-sided cases
passed while the exemption never ran once. Poisoning the membership test
produced no failure, which is what exposed it. With the anchor gone the
poison fails, so the conditional is now proven load-bearing.
.js findings 41 -> 34.
territory-counts.js is `{ 'Andhra Pradesh': 81, 'Assam': 15, … }`, and those
keys are matched against props.shapeName off the map data. Translating one
breaks the join silently: the count goes missing and nothing errors. Same
inversion as a lookup table whose values are i18n keys, except the values are
numbers, so there is no key to resolve and that rule cannot vouch for it.
Judged on the whole object, not one property. A single numeric value proves
nothing — a map of sentences with one count in it would exempt the sentences.
Every value must be a plain non-key literal, there must be at least six of
them, and the key itself must read as a name: anything with sentence
punctuation stays loud whatever its value is.
Two refinements found by running it rather than reasoning about it. The word
limit is eight, not six, because "Dādra and Nagar Haveli and Damān and Diu"
is one real union territory and the longest key in a 46-entry table; the
punctuation test is what separates a name from a sentence, not the length.
And a bare key is now tried against every area, because legend-tree's
REGION_KEY maps a region to 'region_middle_east' and the call site reads
t(`poster.${…}`) — a suffix table is exempt on the same terms as a dotted
one: the key it names must resolve.
Four two-sided cases, and one of them exists because poisoning found the
first three did not exercise the whole rule. Removing the all-values-are-data
check left every test passing, because the rule only ever exempts KEYS and
those fixtures caught their prose as VALUES. The case that bites is a table
whose keys are names and whose values are prose: a label table, not a join
table, where the keys must still count.
With the exemption in, .js files are scanned. 220 .js files were invisible
against 142 .vue. The .js findings fall from 67 to 41.
Chrome only, as scoped: the controls, status lines and navigation, not the data labels or the research notes. 192 English strings → 155; the 155 left are data and notes by that decision:
- the Semantic Trees reference (85: sample tree rows, API field names, taxonomy levels);
- the data-source panel entries (layer names, FeatureServer paths);
- the language family names and the tree level definitions;
- the index tiles' descriptions and source line;
- the cross-system ID names;
- five CSS/SVG values the scanner reads as prose.
What changed:
- The ten view names were written twice, English beside their tKey. The dead English goes, and the bar's brand is pplr.title.
- The theme and projection buttons on the three map views said "Theme: Dark (click to toggle)" / "Switch to Globe". They now say what they do, from the shared keys buttons.switchTo{Light,Dark}Mode and mapui.{enable,disable}Globe, as both title and aria-label.
- The sunburst's status is a key in state (pplr.status.d3Failed / readyD3), translated where it shows, so it follows a language change. Its count uses formatCount like the other views.
- The name-locales view's pill (pplr.pill.translations) and its data-sources button and header (pplr.ui.dataSources) are keys.
- The coverage scale's "0 locales" stop reads 0, like its other stops; the legend title names the unit.
- The index's title and help line are keys, and its tiles are named by their views' own names. Its old help line pointed at an Overview button that no longer exists.
22 English strings outside the catalogue. 10 were dead duplicates: the filter, tab and band labels were already translated from keyCities.* in the templates, so the English copies go. The rest now come from keys:
- the list's caveat is three translated lines, and its long sentence is ONE key rendered with useSentence, with the named-centre count and "Nominatim/OpenStreetMap" passed in as values (the name as its own left-to-right island, never translated);
- a row's group count reuses keyCities.list.peopleGroups, and the unknown internet pill reuses keyCities.map.unknown;
- a load failure reads "The data could not be loaded: {detail}" in the reader's language, with the server's own words as the detail (list and tables);
- the poster wall's intro and its "Other" region come from keys.
Headless, SSR with the real catalogue (_qa-cycle/exp-shepherd/check-keycities-i18n.mjs): the caveats render from keyCities in en, ar and zh, the gazetteer's name stays a <bdi dir="ltr"> island, no raw key or placeholder. The experimental English scan: key-cities 22 → 0. key-cities stays gated (dev only); it builds.
The starter every new plugin copies had 7 English strings outside the catalogue: profile A's "map engine not loaded" and all of profile B. Now profile A keeps its error's KEY in state (errorKey) and translates it in the template, with the engine's own message passed in as a value; profile B's heading, its note line and its sentence are keys, and "Served from {file} in the {bundle} bundle." is ONE key rendered with useSentence, the file and bundle names passed in as their own <code dir="ltr"> islands so each language orders the sentence itself. t() is called in the template throughout, never frozen into a constant at setup. The README's pieces table and "Your own text" section say the same three habits.
Headless, SSR with the real catalogue (_qa-cycle/exp-shepherd/check-template-i18n.mjs): profile B renders in en, ar and zh from template.json, with the sentence in each language's own order (zh puts the bundle first), both islands intact, and no raw key or placeholder. The experimental English scan: template-bundle 7 → 0.
check:no-new-driver-references matched "Driver" case-sensitively, so lines in
capitals passed unseen -- and my own greps were blind the same way, which is why
I reported this plugin clean twice. A case-insensitive sweep finds four, and one
of them carries a verbatim quote.
All four keep their engineering reason and lose the speaker:
· the reuse check states the negotiating position rather than whose it is;
· the substituted button font is blocked on a licensing decision, not a person;
· the code, not just the screenshot, is required to match;
· the table's ring and gridlines live in a global stylesheet, which is why
reading RegionDetail.vue alone could not show them -- the misspelt verbatim
quote is gone, the finding it pointed at is not.
Verified with grep -wi rather than the grep that missed them: 0 remain across
regions-dashboard and country-map. The four reasons are each still present by
their own text. sfc-parse, mjs-parse and no-new-driver-references all pass.
The gate matched "Driver" only, so "THE DRIVER'S ASK" and "HELD FOR THE DRIVER"
passed unseen — and so did my own verification, which used the same
case-sensitive pattern and reported zero. I checked the thing the gate checked
rather than the thing that mattered.
Seven lines across six files: five banner headings, one quoted requirement, and
the row-130 marker. Same rule as the first sweep — keep the reason, drop the
speaker — with the verbatim quote in build-doxa-gospel-access.mjs restated as
the question the map answers, since its content is the requirement and its
framing was attribution.
Comment-only: 0 non-comment lines changed, tree regenerates identically.
Case-insensitive grep across all three plugins now returns nothing.
check:no-new-driver-references matched 'Driver' only, so lines in capitals
passed unseen (and so did my own grep -w Driver). In my-upg-100-list:
'BATCH-UNION DRIVER' and 'gate drivers' are test handles, so they say so;
'in its driver' is generate.mjs's main routine; the zoning declarations'
'PENDING DRIVER' and 'In DRIVER-DECISIONS-DIGEST' become a pending product
decision and the product-decisions digest. grep -rni driver across both
plugins: 0.
Since the regional grouping, buildLegendNodes assembles the tree from the
LEAVES under the poster page's nine region sections. The old India, China,
"Asia (admin-1)" and "Frontier Asia (uncovered)" parents were still built
and never returned: dead code whose English labels the call-site triage
read as reader text. They are deleted, with sumCount, which only they used.
The legend's output is byte-identical for all four configurations
(with/without records, with/without the localizing options; 2,107 records).
The L15 shape assertion now composes its text inside the throw: the
offending leaves are collected as data and formatted in new Error(...), so
the message is visibly developer text. Same message, checked on bad input.
pop-proof.mjs had been failing on every run, because section 3 looked up
those same dead parents, and section 1 expected their count (32). It now
rolls up the nine region parents the tree actually has (each pop = the sum
of its children, and there must be at least one) and expects 28 leaves + 9
regions = 37. ALL PASS, identically against the previous legend-tree.js.
Fourteen references, each rewritten to state the requirement directly. Every
date, measurement and pixel coordinate is kept — the y=1827/y=1865 redundancy
that prompted the count line, the 2026-09-22 interoperability report behind
the Language and Status props, the 220x130 letterbox the square image
replaced, the dev-server path that showed "view more information" reaching
the wrong page, and the India click that left the grid unchanged.
Thirteen were found by a case-sensitive search and a fourteenth was not: a
line in capitals, carrying a verbatim quote, which `grep -w Driver` cannot
see. The check that catches both is `grep -wi driver`, and it now reports 0.
Two sentences broke in the first pass and were found by re-reading every
changed line rather than by the reference count: one left "the control the"
with its object removed, and one left a quotation with an unclosed quote when
the surrounding attribution went. The count was correct while the prose was
wrong, which is why the count is not the check.
check:no-new-driver-references passes. research-map-plus-grid and
dashboard-plus-grid both rebuild byte-identical.
