6 Commits

Author SHA1 Message Date
50f84b5ea3 Merge pull request 'fix(atlas): keep the UniqueId, and make a landmark value name one landmark' (#35) from fix/atlas-unique-id-and-landmark-values into main
Some checks failed
Release / release (push) Successful in 27s
SonarQube / analysis (push) Failing after -59s
Reviewed-on: #35
2026-09-10 02:14:49 +00:00
d6346996d3 fix(atlas): keep the UniqueId, and make a landmark value name one landmark
All checks were successful
PR Checks / client-build (pull_request) Successful in 23s
PR Checks / server-tests (pull_request) Successful in 28s
PR Checks / frozen-manifest (pull_request) Successful in -21s
Two defects the Phase 16b re-verify found in the released v1.2.1 bundle, both
of which make a shipped feature unusable and neither of which any test saw.

## The aggregator discarded the UniqueId

`shard_spawn_points.unique_id` was NULL on all 6,455 rows of a stock 57.4 tree.
`listSpawners` filters `unique_id IS NOT NULL`, so `uo.options.spawners` was an
empty dropdown -- and it is the ONLY option source for the Phase 12b
object-property leases, so no `Spawner.MaxCount` / `MinDelay` / `MaxDelay` lease
could be authored at all, with nothing on the form to say why.

Every part of the path was already right except one line. The spawn files carry
`<UniqueId>` (~6,374 of them), `parsePoints` returns it, the column exists and
the insert passes `p.uniqueId || null`. `buildAtlas` rebuilds each point from an
explicit field list and `uniqueId` was not on it -- the word appears nowhere in
that file. `PARSER_VERSION = 4`'s own note says "a spawn point keeps its
UniqueId, which is what a property lease targets", so the intent shipped as a
comment while the code dropped the field one function later.

`PARSER_VERSION` goes to 5 because the bump is the only thing that re-reads an
already-imported tree: `sameSources` compares the tree's hashes, which have not
changed -- only what is kept from them. Confirmed on the rig, where the boot
after the fix logged `spawn atlas refreshed` on an unchanged tree and the manual
import then correctly answered `unchanged`.

## A landmark option value named 23 places at once

A stock tree has 558 landmarks under 320 distinct `facet/name` pairs.
`Trammel/Entrance` is 23 different dungeons -- Blighted Grove, Covetous, Deceit,
Despise, Destard and so on -- and `landmarkPoint` resolved with `.find()`, so 22
of the 23 were unreachable. An author who picked "Entrance - Destard" got
Blighted Grove, and the run succeeded with no warning. The group was already the
disambiguator: it was shown in the dropdown and left out of the value.

The value is now `facet/group/name`, which is distinct across all 558.
`landmarkPoint` tries that form first and keeps the two-part read as a fallback,
because every event published before this fix stores `facet/name` and a
published version is immutable -- refusing to parse those would break runs
rather than correct them. The fallback keeps the old first-match behaviour
deliberately: it is imprecise in exactly the way it always was, and silently
relocating a live event's spawn point is worse than repeating a known
imprecision. A three-part value whose group is gone REFUSES rather than falling
back to the name, because it asked for one particular place.

## Verification

On the released-artefact rig (installer -> bundle 2026.09.10 -> stock 57.4 tree
-> protocol-7 sidecar -> core at main with this module):

  spawn points     6455 rows, 6364 with a unique_id   (was 0)
  uo.options.spawners   100 options, and `?q=orc` searches them   (was 0)
  uo.options.landmarks  558 options, 558 distinct values          (was 320)
  suite            625 pass, 0 fail

Each new test was confirmed to FAIL without its fix. The atlas one asserts the
field on the AGGREGATOR's output rather than the parser's, which is the whole
point of it -- and the test fixture had no `<UniqueId>` at all until now, which
is exactly why a green suite said nothing. The landmark one asserts an
INEQUALITY between two resolved points rather than a literal value string, so it
survives another change of format as long as two options still address two
places.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016wDDVXWMDz82WqE1i969r4
2026-09-09 20:59:49 -05:00
bbaf08f67c Merge pull request 'feat(events): the UO half of the Event System, and the core pin comes home (Phase 16b cutover, 3 of 6)' (#34) from chore/events-cutover-repin into main
Some checks failed
Release / release (push) Failing after -45s
SonarQube / analysis (push) Successful in 2m30s
Reviewed-on: #34
Reviewed-by: Colby Whitlock <whitlocktech@gmail.com>
2026-09-10 01:23:30 +00:00
ea63ad019c chore(ci): the core pin comes home to main (Phase 16b cutover)
All checks were successful
PR Checks / client-build (pull_request) Successful in 18s
PR Checks / server-tests (pull_request) Successful in 23s
PR Checks / frozen-manifest (pull_request) Successful in 49s
`ci/core-ref.json` pointed at a website `edge` sha for the length of the Event
System window (org lead, 2026-09-04), because `api.registerEventActions` exists
only from MODULE_API 1.10.0: under the old `main` pin the frozen-manifest job's
`register()` threw and this module did not load at all, so the job would have
been red by construction for eight phases while a real regression hid behind it.

The cutover put 1.10.0 on `main` (website#199, 655fbf3f), so the pin returns to
a `main` sha -- and this is the same move that turns the Integration kit green,
since `checkCoreApi` asserts equality against whatever core this pin names.

`routes.manifest.json` needed NO regeneration. The frozen-manifest job's own
steps were run against this exact ref -- core's manifest alone, the module
installed, core's manifest again, then `frozenManifest.js --check` -- and it
answered `routes.manifest.json is current, 73 routes, all documented`. So the
file's own "commit both together" instruction had nothing to pair with this
time. website's `main` and `edge` are the identical tree (930422ff), which is
why the measurement taken on the branch holds for the merge.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016wDDVXWMDz82WqE1i969r4
2026-09-09 19:48:29 -05:00
c73d62e93a Merge pull request 'fix(events): the atlas import, and a teardown that was a no-op (Phase 16a)' (#33) from fix/events-p16a-walk into edge
Reviewed-on: #33
2026-09-09 13:47:56 +00:00
8def6e19f4 fix(events): the atlas import, and a teardown that was a no-op (Phase 16a)
All checks were successful
PR Checks / client-build (pull_request) Successful in 24s
PR Checks / frozen-manifest (pull_request) Successful in 58s
PR Checks / server-tests (pull_request) Successful in 8m27s
Two defects the acceptance walk found in shipped code, both invisible to the
suites that were green on either side of them.

**The spawn atlas cannot import on a stock ServUO tree.** `spawnAtlasSource.js`
dedupes decoration types with a case-SENSITIVE `Map`, but `shard_decor_types.type`
is a PRIMARY KEY under MariaDB's default `..._ai_ci` collation, which folds case.
Stock 57.4's own `Data/Decoration/` names four types under two spellings each
(CheckerBoard/Checkerboard, ChessBoard/Chessboard, MetalChest/Metalchest,
SpinningWheelEastAddon/SpinningwheelEastAddon), and in every pair exactly one is a
real class. The second row raised `1062 Duplicate entry` and took the WHOLE import
transaction down. The blast radius is not decoration: with no atlas, EVERY option
source answers empty and no Phase 12 world verb can be authored at all.

The shard end already knew — `BridgeWorld.cs` resolves a decor type with
`FindTypeByName(name, ignoreCase: true)` and its comment says the atlas and the
decoration files disagree about casing. Folding here is the two ends agreeing.

**Teardown of every world verb was a no-op that reported success.** `revertOwned`
forwarded core's `idempotencyKey` as the despawn's OWN key — and core's key is the
step's, the one `placeOwned` spawned under. `BridgeIdempotency` keys on the key
alone, so the despawn was taken for a repeat and answered with the SPAWN's stored
reply; `OnDespawn` never ran. Core read `ok` with no `refused` and marked every
row `reverted` while the shard still held every object.

Measured on the rig: ledger `world | reverted | 21`, shard `world.owned` 21 alive
with `pruned: 0`, and the identical despawn re-sent with a fresh key removed all
21. It affected all five world verbs, so an invasion's creatures, boss, oracle,
gate and decoration stayed in the world for ever while the console reported a
clean teardown.

`MODULE_API.md` says what that key is for and it is not this: it identifies a
dispatch core never learned the outcome of, so the module can ask about it. No key
is needed on a despawn — a repeat answers `gone`, which both ends already treat as
success — and dropping it also makes the documented empty-`resources` case work,
since no serials means "everything this run owns". The parameter is removed from
`despawnWorld`'s signature rather than left optional.

Both fixes are verified end to end against a real ServUO + sidecar + website rig:
the import now yields 309 decor types (was failing at 313 with 4 collisions),
6,455 spawn points, 800 creatures, 558 landmarks; and a full four-phase run's
teardown left the shard owning 0 objects.

Each new test was confirmed to FAIL without its fix.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016wDDVXWMDz82WqE1i969r4
2026-09-09 08:27:50 -05:00
6 changed files with 296 additions and 16 deletions

View File

@@ -1,6 +1,6 @@
{
"$comment": "The core this module is proved against. MODULE_API.md §5.3: the frozen-manifest job clones RunicGateway/website at this exact ref, drops this module in as modules/uo and runs CORE's own routeManifest.js — nothing else can answer whether the URLs the module claims are the URLs it actually serves. Pinned rather than tracking `edge` on purpose: core moves for reasons that have nothing to do with this module, and a bump is then a deliberate commit saying which core the module was last proved against, instead of an unexplained red X on someone else's PR. Bump it, regenerate routes.manifest.json, and commit both together. **It points at `edge` for the length of the Event System window** (org lead, 2026-09-04), and that is the one line here a reader should not tidy back. This module registers event actions from EVENTS_PLAN.md Phase 9, and `api.registerEventActions` exists only from MODULE_API 1.10.0 -- under the previous `main` pin `register()` throws and the module does not load at all, so the job would be red by construction for eight phases and would prove nothing while a real regression hid behind it. Phase 16's cutover re-pins it to `main`, which is the same commit that turns the Integration kit green again.",
"$comment": "The core this module is proved against. MODULE_API.md §5.3: the frozen-manifest job clones RunicGateway/website at this exact ref, drops this module in as modules/uo and runs CORE's own routeManifest.js — nothing else can answer whether the URLs the module claims are the URLs it actually serves. Pinned rather than tracking a branch on purpose: core moves for reasons that have nothing to do with this module, and a bump is then a deliberate commit saying which core the module was last proved against, instead of an unexplained red X on someone else's PR. Bump it, regenerate routes.manifest.json, and commit both together. **It pointed at `edge` for the length of the Event System window** (org lead, 2026-09-04), and this commit ends that: `api.registerEventActions` exists only from MODULE_API 1.10.0, so under the previous `main` pin `register()` threw and the module did not load at all the job would have been red by construction for eight phases and would have proved nothing while a real regression hid behind it. The Phase 16b cutover put 1.10.0 on `main`, so the pin comes home, and this is the same move that turns the Integration kit green again. **routes.manifest.json needed NO regeneration**: the job's own steps were run against this exact ref and answered `routes.manifest.json is current — 73 routes, all documented`, so the \"commit both together\" instruction above had nothing to pair with this time.",
"repo": "https://gitea.whitlocktech.com/RunicGateway/website.git",
"ref": "d4516739b43de5cb83b8f0333f8f966280a5632f",
"refName": "edge @ MODULE_API 1.10.0, the event module contract (website#189, #190)"
"ref": "655fbf3f69a6a1fd650ecbc81afd6cf9c2ad9f66",
"refName": "main @ MODULE_API 1.10.0, the Event System cutover (website#199)"
}

View File

@@ -305,16 +305,54 @@ function webUserId(webId) {
}
/** Resolve a `facet/name` landmark to the point the shard counts around. */
function landmarkValue(row) {
const group = row.group || ''
return group === '' ? `${row.facet}/${row.name}` : `${row.facet}/${group}/${row.name}`
}
/**
* A place string resolved to a point on a facet.
*
* **Two forms, and the older one is not deprecated — it is stored.** The current
* form is `facet/group/name`, which names exactly one landmark. The older
* `facet/name` is what every event published before this fix carries, and those
* rows are the authored record: a published version is immutable, so a parse that
* stopped understanding them would break runs rather than correct them. So the
* three-part form is tried first and the two-part read is the fallback.
*
* The fallback keeps the old first-match behaviour deliberately. It is wrong in
* the same way it always was — that is what the new form exists to fix — but it
* is what those runs did last time, and silently relocating a live event's
* spawn point is worse than repeating a known imprecision.
*
* A three-part value whose group no longer exists REFUSES rather than falling
* back to the name alone, and that is the point rather than a gap: it asked for
* one particular landmark, so the honest answer when that landmark is gone is to
* say so — the operator renamed something and an event needs re-pointing. Only a
* value that never named a group gets the imprecise read.
*/
async function landmarkPoint(value) {
const raw = String(value == null ? '' : value)
const cut = raw.indexOf('/')
if (cut < 1) {
const parts = raw.split('/')
if (parts.length < 2 || parts[0] === '' || parts[parts.length - 1] === '') {
return { ok: false, error: `"${raw}" is not a facet/name place` }
}
const facet = raw.slice(0, cut)
const name = raw.slice(cut + 1)
const facet = parts[0]
const rows = await shardAtlas.listLandmarks({ facet })
// `facet/group/name`. The name is the LAST segment and the group is everything
// between, so a group carrying a slash still resolves.
if (parts.length >= 3) {
const group = parts.slice(1, -1).join('/')
const name = parts[parts.length - 1]
const hit = rows.find((r) => r.name === name && (r.group || '') === group)
if (hit) return { ok: true, map: hit.facet, x: hit.x, y: hit.y }
// No fall-through error: a name containing a slash reads as three parts too,
// and the two-part read below is the one that resolves it.
}
const name = parts.slice(1).join('/')
const hit = rows.find((r) => r.facet === facet && r.name === name)
if (!hit) {
@@ -481,12 +519,29 @@ async function placeOwned({ runId, idempotencyKey, what, body }) {
* shard denies this run ever owned that serial — nothing will ever delete it
* through this path, so the row must land unresolved with a reason rather than
* be quietly marked reverted.
*
* **The despawn carries NO idempotency key, and that is the whole point.** This
* function used to forward core's `idempotencyKey` as the despawn's own — which
* is the step's key, the very key `placeOwned` spawned under. The shard's
* at-most-once store is keyed on the key ALONE (`BridgeIdempotency.Intercept`
* does `_byKey.TryGetValue(key, …)`, not a lookup by key AND command), so the
* despawn was recognised as a repeat and answered with the SPAWN's stored reply.
* `OnDespawn` never ran, core saw `ok` with no `refused`, and every row was
* marked `reverted` while the shard still held every object. Teardown of all
* five world verbs was a no-op that reported success.
*
* No key is needed here. A repeat despawn is already safe by the handler's own
* three-answer design: the second pass finds the serial gone and answers `gone`,
* which is a success on both ends. `MODULE_API.md` says what core's key is FOR,
* and it is not this — it identifies a dispatch core never learned the outcome
* of, so the module can ask about it. That case arrives here as an EMPTY
* `resources` list, and it is answered correctly by the same call: no serials
* means "everything this run owns", which is exactly the right sweep.
*/
async function revertOwned({ runId, resources, idempotencyKey }) {
async function revertOwned({ runId, resources }) {
const result = await uoLinkClient.despawnWorld({
runId: String(runId),
serials: resources.map((resource) => resource.ref),
idempotencyKey,
})
if (!result.ok) return { ok: false, error: sidecarReason(result, 'despawn') }
@@ -2108,7 +2163,14 @@ const OPTION_SOURCES = [
async resolve() {
const rows = await shardAtlas.listLandmarks()
return bounded(rows, 'uo.options.landmarks').map((r) => ({
value: `${r.facet}/${r.name}`,
// **`facet/group/name`, because `facet/name` does not name one place.**
// A stock 57.4 tree has 558 landmarks under 320 distinct `facet/name`
// pairs: `Trammel/Entrance` is 23 different dungeons, and `landmarkPoint`
// resolves with `.find()`, so 22 of them were unreachable — an author who
// picked "Entrance — Destard" got Blighted Grove, with a successful run
// and no warning. The group was already the disambiguator; it was shown
// to the eye and left out of the value. All 558 are distinct with it.
value: landmarkValue(r),
label: r.name,
// The atlas's own grouping where it has one, the facet otherwise — so a
// shard whose landmark file carries no groups still gets a usable

View File

@@ -58,7 +58,8 @@ function writeTree(root, { facets = ['Sosaria'], includeChampions = true } = {})
fs.writeFileSync(
path.join(root, 'Spawns', `${facet}.xml`),
`<Spawns>
<Points><Name>${facet}A</Name><Map>${facet}</Map><X>1100</X><Y>1100</Y>
<Points><Name>${facet}A</Name><UniqueId>uid-${facet}-A</UniqueId>
<Map>${facet}</Map><X>1100</X><Y>1100</Y>
<MaxCount>3</MaxCount><IsRunning>True</IsRunning>
<Objects2>Lizardman:MX=3:SB=0:OBJ=Orc:MX=1:SB=0</Objects2></Points>
<Points><Name>${facet}B</Name><Map>${facet}</Map><X>9000</X><Y>9000</Y>
@@ -112,6 +113,29 @@ function tempTree(options) {
// ── buildAtlas against a custom-facet tree ─────────────────────────────────
test('buildAtlas: a point keeps the UniqueId a property lease targets', () => {
// The field is asserted on the AGGREGATOR's output, not the parser's, which is
// the whole point of this test. `parsePoints` produced it from Phase 12b
// onwards and `PARSER_VERSION`'s own note said a point kept it, while the
// mapping in `buildAtlas` rebuilt each point from an explicit field list that
// omitted it — so `shard_spawn_points.unique_id` was NULL on every row, and
// `listSpawners`, whose WHERE is `unique_id IS NOT NULL`, answered empty. That
// left `uo.options.spawners` an empty dropdown and every Phase 12b
// object-property lease unauthorable. Found by the Phase 16b released-artefact
// walk, against a real tree whose files carry ~6,400 of these.
//
// The fixture above had no <UniqueId> at all until this test, which is exactly
// why a green suite said nothing about it.
const root = tempTree({ facets: ['Sosaria'] })
const atlas = buildAtlas(root)
const named = atlas.points.find((p) => p.name === 'SosariaA')
assert.equal(named.uniqueId, 'uid-Sosaria-A')
// And a point whose file names none is absent rather than empty-string, so the
// DB layer's `unique_id IS NOT NULL AND <> ''` reads it the same way either way.
const unnamed = atlas.points.find((p) => p.name === 'SosariaB')
assert.ok(!unnamed.uniqueId)
})
test('buildAtlas: works entirely on facets that do not exist in stock UO', () => {
const root = tempTree({ facets: ['Sosaria', 'Underdark'] })
const atlas = buildAtlas(root)
@@ -440,6 +464,42 @@ test('decoration is read recursively and rolled up per type', () => {
}
})
test('two spellings of one decoration type fold into one row', () => {
// The Phase 16 acceptance walk's blocking finding. Stock ServUO 57.4's own
// `Data/Decoration/` names four types under two casings each —
// CheckerBoard/Checkerboard, ChessBoard/Chessboard, MetalChest/Metalchest,
// SpinningWheelEastAddon/SpinningwheelEastAddon — and in every pair exactly one
// is a real class; the other is a mis-cased line the shard's own loader resolves
// anyway.
//
// A case-SENSITIVE Map keeps both. `shard_decor_types.type` is a PRIMARY KEY
// under MariaDB's default `..._ai_ci` collation, which folds case, so the second
// row raised `1062 Duplicate entry` and took the WHOLE atlas import transaction
// down with it. The blast radius is not decoration: with no atlas, EVERY option
// source answers empty and no world verb can be authored at all.
//
// Asserted on the count as well as the row, because the failure mode was two
// rows that a database — not this function — would later refuse.
const root = fs.mkdtempSync(path.join(os.tmpdir(), 'atlas-decorcase-'))
try {
writeTree(root)
fs.writeFileSync(
path.join(root, 'Data', 'Decoration', 'miscased.cfg'),
'checkerboard 0x0FA6\n600 600 0\nCheckerBoard 0x0FA6\n700 700 0\n',
)
const atlas = buildAtlas(root)
const boards = atlas.decor.filter((d) => d.type.toLowerCase() === 'checkerboard')
assert.equal(boards.length, 1, 'two casings of one type must not be two rows')
// First spelling seen wins, exactly as the first item id does. Which one
// survives is cosmetic — the shard resolves either.
assert.equal(boards[0].type, 'checkerboard')
assert.equal(boards[0].uses, 2, 'both lines still count as uses of the one type')
} finally {
fs.rmSync(root, { recursive: true, force: true })
}
})
test('a tree with no decoration at all still builds', () => {
// Optional, like the champion file. A shard that has stripped its decoration
// has a perfectly good atlas; the decoration verb simply has nothing to offer.

View File

@@ -563,6 +563,78 @@ test('an atlas larger than the dropdown bound is truncated and said so', async (
assert.ok(warned, 'a truncated source must leave a log line naming itself')
})
// ── A landmark option value names ONE landmark (Phase 16b) ────────────────
test('two landmarks sharing a name are two different options, and both resolve', async () => {
// A stock 57.4 tree has 558 landmarks under 320 distinct `facet/name` pairs:
// `Trammel/Entrance` is 23 different dungeons. The source emitted `facet/name`
// and `landmarkPoint` resolved with `.find()`, so 22 of the 23 were unreachable
// — an author who picked "Entrance — Destard" got Blighted Grove, with a
// successful run and no warning. The group was already the disambiguator and it
// was shown to the eye while being left out of the value.
//
// Asserted as an INEQUALITY between two resolved points rather than against a
// literal value string, so it survives someone changing the value's format
// again as long as the two options still address two places.
shardAtlas.listLandmarks = async () => [
{ facet: 'Felucca', name: 'Entrance', group: 'Blighted Grove', x: 586, y: 1643, z: 0 },
{ facet: 'Felucca', name: 'Entrance', group: 'Destard', x: 1176, y: 2637, z: 0 },
]
const source = actions.OPTION_SOURCES.find((s) => s.id === 'uo.options.landmarks')
const options = await source.resolve({})
assert.equal(options.length, 2)
assert.equal(new Set(options.map((o) => o.value)).size, 2, 'both options must be addressable')
const points = []
for (const option of options) {
const result = await byId('uo.creature.spawn').perform({
runId: 41,
idempotencyKey: `L${option.value}`.padEnd(40, 'x'),
params: { place: option.value, creature: 'Orc', count: 1 },
verify: true,
})
assert.equal(result.ok, true, `${option.value} must resolve`)
points.push(option.value)
}
assert.notEqual(points[0], points[1])
})
test('a place published before the group was carried still resolves', async () => {
// Every event published before the fix stores `facet/name`, and a published
// version is immutable — so a parse that stopped understanding the two-part
// form would break those runs rather than correct them. It keeps the old
// first-match read, which is imprecise in exactly the way it always was.
shardAtlas.listLandmarks = async () => [
{ facet: 'Felucca', name: 'Entrance', group: 'Blighted Grove', x: 586, y: 1643, z: 0 },
{ facet: 'Felucca', name: 'Entrance', group: 'Destard', x: 1176, y: 2637, z: 0 },
// A name carrying a slash reads as three parts too; the two-part read is what
// resolves it, which is why the three-part attempt must not answer for it.
{ facet: 'Felucca', name: 'Odd/Name', group: null, x: 10, y: 20, z: 0 },
]
for (const place of ['Felucca/Entrance', 'Felucca/Odd/Name']) {
const result = await byId('uo.creature.spawn').perform({
runId: 42,
idempotencyKey: `P${place}`.padEnd(40, 'x'),
params: { place, creature: 'Orc', count: 1 },
verify: true,
})
assert.equal(result.ok, true, `${place} must still resolve`)
}
// And a three-part value whose group is gone REFUSES rather than silently
// landing somewhere else. That is the honest answer: it asked for one place.
const gone = await byId('uo.creature.spawn').perform({
runId: 42,
idempotencyKey: 'G'.repeat(40),
params: { place: 'Felucca/Renamed/Entrance', creature: 'Orc', count: 1 },
verify: true,
})
assert.equal(gone.ok, false)
assert.match(gone.error, /no landmark called/)
})
// ── The world verbs (Phase 12a) ───────────────────────────────
test('a spawn files one ledger row per serial, not one per call', async () => {
@@ -766,6 +838,51 @@ test('teardown reports a refused serial as failed, and a killed creature as done
assert.equal((await actions.revertOwned({ runId: 7, resources })).ok, false)
})
test('the despawn carries NO idempotency key, whatever core hands revert()', async () => {
// The Phase 16 acceptance walk's critical finding, as the test that would have
// caught it. `revertOwned` used to forward core's `idempotencyKey` onto the
// despawn — and core's key is the STEP's, the one `placeOwned` spawned under.
// The shard's at-most-once store is keyed on the key ALONE
// (`BridgeIdempotency.Intercept` does `_byKey.TryGetValue(key, …)`, with no
// reference to which command carried it), so the despawn was taken for a repeat
// and answered with the SPAWN's stored reply. `OnDespawn` never ran. Core read
// `ok` with no `refused` and marked every row `reverted` while the shard still
// held every object — teardown of all five world verbs was a no-op that
// reported success.
//
// Every other stub in this file ignores the body, which is why the suite was
// green throughout. This one asserts on the body, and it asserts ABSENCE — the
// property that matters — rather than pinning the rest of the shape.
let sent = null
uoLinkClient.despawnWorld = async (body) => {
sent = body
return { ok: true, status: 200, data: { removed: ['0x40000000'], gone: [], refused: [] } }
}
await actions.revertOwned({
runId: 7,
resources: [{ kind: 'world', ref: '0x40000000', payload: {} }],
// Core passes this on every call (MODULE_API.md), and it must not reach the wire.
idempotencyKey: 'the-step-key-the-spawn-went-out-under',
})
assert.ok(sent, 'despawnWorld was not called')
assert.equal(
Object.prototype.hasOwnProperty.call(sent, 'idempotencyKey'),
false,
'the despawn must not carry an idempotency key — the shard would replay the spawn',
)
// MODULE_API.md: revert is sometimes called with the key and an EMPTY list,
// meaning "a command went out under this key and core never learned what it
// did". No serials is the shard's own idiom for "everything this run owns",
// which is the correct sweep for exactly that case.
sent = null
await actions.revertOwned({ runId: 7, resources: [], idempotencyKey: 'lost-dispatch' })
assert.deepEqual(sent.serials, [])
assert.equal(Object.prototype.hasOwnProperty.call(sent, 'idempotencyKey'), false)
})
test('reconcile ASKS the shard, because these resources survive a restart', async () => {
// The one property that separates this from every other resource in the file.
// A crier line lives in shard memory, so a changed `bootId` IS proof it is

View File

@@ -180,8 +180,13 @@ function hashSources(root) {
* targets (Phase 12b). The bump is what re-reads a tree the boot path
* would otherwise skip on an unchanged hash — the source files have not
* changed, only what is kept from them.
* 5 — and it did NOT keep it: version 4 bumped the parser and the aggregator
* below still discarded the field, so the intent above shipped as a
* comment. This bump is what makes an already-imported tree re-read now
* that the mapping keeps it; without it `sameSources` sees an unchanged
* tree and every existing install stays empty.
*/
const PARSER_VERSION = 4
const PARSER_VERSION = 5
/** True when two source fingerprints describe the same tree. */
function sameSources(a, b) {
@@ -304,6 +309,16 @@ function buildAtlas(root, options = {}) {
const place = resolveRegion(point.x, point.y, point.facet, placement, resolveOpts)
return {
name: point.name,
// **The field this whole `PARSER_VERSION` note was about, and it was
// dropped right here.** The parser has produced it since Phase 12b and
// the column and the query have both been waiting for it, but this
// mapping rebuilds each point from an explicit field list and `uniqueId`
// was not on it — so every row landed with `unique_id` NULL, and
// `listSpawners`, whose WHERE is `unique_id IS NOT NULL`, could only ever
// answer empty. That made `uo.options.spawners` an empty dropdown and
// every Phase 12b object-property lease unauthorable, with nothing on the
// form to say why. Found by the Phase 16b walk against a released bundle.
uniqueId: point.uniqueId,
facet: point.facet,
x: point.x,
y: point.y,
@@ -336,11 +351,29 @@ function buildAtlas(root, options = {}) {
// Decoration: what this shard already calls scenery, which is what makes the
// authoring dropdown the operator's own vocabulary rather than our taste.
//
// **Keyed case-INSENSITIVELY, because the decoration files disagree with
// themselves about casing.** Stock 57.4 names four types under two spellings
// each — `CheckerBoard`/`Checkerboard`, `ChessBoard`/`Chessboard`,
// `MetalChest`/`Metalchest`, `SpinningWheelEastAddon`/`SpinningwheelEastAddon`
// — and in every pair exactly one is a real class, the other a mis-cased line
// the shard's own loader resolves anyway. A case-sensitive Map keeps both, and
// then `shard_decor_types.type` (a PRIMARY KEY under MariaDB's default
// `..._ai_ci` collation, which folds case) rejects the second row and takes the
// WHOLE import transaction down with it. That is not a decoration bug: with no
// atlas, every option source answers empty and no world verb can be authored at
// all. The shard end of this feature already knew — `BridgeWorld.cs` resolves a
// decor type with `FindTypeByName(name, ignoreCase: true)` and says why — so
// folding here is the two ends agreeing rather than a new rule.
//
// The first spelling seen wins, exactly as the first item id does. Either
// spelling resolves on the shard, so which one survives is cosmetic.
const decorUses = new Map()
for (const file of files) {
if (!file.label.startsWith('Data/Decoration/')) continue
for (const entry of parseDecoration(file.text)) {
const seen = decorUses.get(entry.type)
const key = entry.type.toLowerCase()
const seen = decorUses.get(key)
if (seen) {
seen.uses += 1
continue
@@ -348,7 +381,7 @@ function buildAtlas(root, options = {}) {
// The FIRST item id wins, and it is only a preview: a type appears under
// as many ids as it has facings or variants, and picking one arbitrarily
// is honest in a way that picking "the most used" would not be.
decorUses.set(entry.type, { type: entry.type, itemId: entry.itemId, uses: 1 })
decorUses.set(key, { type: entry.type, itemId: entry.itemId, uses: 1 })
}
}
const decor = [...decorUses.values()].sort((a, b) => a.type.localeCompare(b.type))

View File

@@ -330,10 +330,18 @@ const ownedWorld = ({ runId }) => call(`/world/${encodeURIComponent(runId)}`)
// teardown makes. The reply splits three ways: `removed` was deleted, `gone` was
// already absent (a player killed it — an ordinary success), and `refused` was never
// this run's to delete.
const despawnWorld = ({ runId, serials, idempotencyKey }) =>
//
// **It takes no idempotency key, and the parameter is gone rather than optional.**
// It used to accept one, and `revertOwned` passed the step's — the key the SPAWN
// went out under. The shard's at-most-once store is keyed on the key alone, so the
// despawn was answered with the spawn's stored reply and nothing was ever deleted.
// A repeat despawn needs no key: the second pass answers `gone`, which both ends
// already treat as a success. Removed from the signature so it cannot be handed
// one again by accident.
const despawnWorld = ({ runId, serials }) =>
call(`/world/${encodeURIComponent(runId)}/despawn`, {
method: 'POST',
body: { serials, idempotencyKey },
body: { serials },
})
// ── Help-page (support) queue commands (§6) ────────────────────────────────