The Atlas server no longer keeps the admin-1 regions of every country in memory. Each country's regions are now a pre-gzipped file of their own, sent to the map as they are.
**Why:** `GET /addons/atlas/regions/geo?countries=…` built `admin1Store` (every country's region text, kept until restart), then joined the requested countries, `JSON.parse`d them, and Nest stringified them again. On the 512 MB – 1 GB boxes TREK runs on, the first z≥5 view could OOM the server (same shape as #1576).
**What changes:**
- `server/assets/atlas/admin1/<ISO_A2>.geojson.gz`: 198 files, 14.85 MB, the same 3,447 features as `admin1.geojson.gz`.
- New `GET /addons/atlas/regions/geo/:code` with a Zod contract (`regionGeoParamsSchema` in `@trek/shared`: two letters, any case). It sends the country's file as it is, with `Content-Encoding: gzip` and the current 1-day `Cache-Control`, as `countries/geo` does. A country without regions gets an empty FeatureCollection. Anything that is not two letters gets a 400.
- `useAtlas.ts` fetches one URL per uncached country and still rebuilds the layer once. Each country has a stable URL, so a country evicted by `REGION_CACHE_MAX` comes back from the browser cache.
- `getRegionFeatures` reads only the country it needs. `admin1Store` is removed.
- `?countries=` keeps working for cached PWA clients. It joins the per-country files as bytes, without a parse.
- The migrations are untouched. `admin1.geojson.gz` still ships, only for `Migration20200101021500` and `Migration20200101031500`. This costs ~15 MB of disk in the image and no RAM. `build-atlas-geo.mjs` writes both, and ATLAS-BUNDLE-008 keeps them equal.
**Memory** (real `atlas-geo` module, first z≥5 view over Europe = 49 country codes, JS heap + Buffers):
| | peak | kept after |
|---|---|---|
| before, first view (builds the store) | +372 MB | +61 MB |
| before, later views | +115 MB | 0 |
| after, `/regions/geo/:code` | +4.7 MB | 0 |
| after, legacy `?countries=` | +59 MB | 0 |
**Tests:**
- New: `atlas-region-files.test.ts` (moved from `atlas.service.test.ts`, plus join and dedupe cases), the `regionGeoParamsSchema` spec, ATLAS-BUNDLE-007 (the file format the byte join depends on), ATLAS-BUNDLE-008 (bundle and per-country files are equal), ATLAS-016/017/018 (the new route over HTTP, including the 400).
- The `getVisitedRegions` tests (ATLAS-UNIT-019..024) used to pass only when `getRegionGeo`'s `beforeAll` had built the store first. They now also pass on their own.
- The shared, server and client gates pass. Baselines go down only where this change removed code (`no-console`, `no-explicit-any`, `no-swallowed-catch`, file sizes).
**Trade-offs:**
- A first zoom over Europe sends ~49 small parallel requests instead of 1. Over HTTP/1.1 the browser queues them 6 at a time. The total bytes are the same (4.4 MB gz).
- The per-country URLs are unversioned and cached for a day, like `countries/geo`. After an upgrade that changes a country's regions, a browser can show the old shapes for up to 24 h.
Approved on Discord in #github-pr: <paste the thread link here>
Related: #1576, #2559
- [x] Bug fix
- [ ] New feature
- [ ] Breaking change
- [ ] Documentation update
- [ ] I have read the [Contributing Guidelines](https://github.com/liketrek/TREK/wiki/Contributing)
- [ ] My branch is [up to date with `dev`](https://github.com/liketrek/TREK/wiki/Development-environment#3-keep-your-fork-up-to-date)
- [x] This PR targets the `dev` branch, not `main` *(wiki-only PRs are exempt)*
- [x] I have tested my changes locally
- [x] I have added/updated tests that prove my fix is effective or that my feature works
- [x] I have updated documentation if needed
Translate the 55 keys dev added since the branch was cut (TREK services
allowance, Valhalla routing service, PDF routing progress, overview
distances, Unsplash key hint, tour day limit) and drop the two reservation
toasts dev removed, so he matches en key for key again. Refresh the two
admin strings and the Valhalla hint whose English changed.
Check he in the placeholder spec and the RTL test, and count 28 languages
with Hebrew in the README and the Languages, Environment Variables,
Contributing and Help Center wiki pages.
Fill in every domain added since the first Hebrew pass (help, tours,
assistant*, studio*, wanderer*, journey, demo, ferratas and the rest),
retranslate strings whose English changed, and give every count group
its Hebrew one/two forms. Register he as a non-Latin locale for the
untranslated ratchet, add the Hebrew 429 message, and count 28 locales
in the tours locale spec.
Full Hebrew translation of every UI domain plus email/notification
templates. Registered in SUPPORTED_LANGUAGES, marked right-to-left,
wired into the client locale loader and the rate-limit message table.
transit is not one of the fixed Costs categories, so typeToCostCategory
would put an uncategorised fare under other. Default to transport, the
category every rail and bus booking maps to.
- create_transit_journey takes price / budget_category, as create_transport
does: a fare > 0 files the linked expense in the booking's own transaction
and rides along in metadata.price. The tool now writes through
ReservationWritesService.createBooking, so its events and booking notice are
the ones every other booking write sends
- create_budget_item and create_budget_item_with_members take reservation_id,
the field POST /budget already accepts; create_budget_item now runs
BudgetWritesService.createItem like its sibling, so a linked booking's
mirrored price is worked out again before budget:created goes out
- bookingPriceSchema / bookingBudgetCategorySchema in @trek/shared back the
price fields of all three booking tools (mcp-inline-zod baseline lowered)
- wiki: MCP-Tools-and-Resources describes the new params
The trimming if swallowed the blank line, so the HTTPRoute note ran
straight on from the Persistence section. Move the blank line inside
the conditional so it only appears with the section.
Add disabled-by-default HTTPRoute support with configurable parent references,
hostnames, paths, annotations and optional timeouts. Route to the existing TREK
Service and keep Ingress support unchanged, including simultaneous use.
Document configuration and extend the existing Helm Chart & Compose job with
HTTPRoute render variants and strict validation using a pinned CRDs catalog.
Approved on Discord:
https://discord.com/channels/1488298068427411591/1553537637455761500
Validation: strict Helm lint, 37 manifests validated without errors or skipped
resources, and both Compose configuration checks passed locally.
Live Gateway WebSocket and streamed assistant response checks were not performed
because no Kubernetes test cluster was configured.
The install guide lists extraEnv and extraEnvFrom but had no word on the
new volume values. Add a Further volumes section with the KItinerary
extractor example and point the Reservations-and-Bookings page at it.
The chart mounts only the data and uploads volumes, and no value adds
another one. Add two lists that go into the Deployment as written:
extraVolumes for the pod, and extraVolumeMounts for the trek container,
next to the data and uploads mounts. Both lists are empty by default,
so the default render does not change.
The values file shows one use: a ConfigMap of KItinerary extractor
scripts, mounted at /usr/local/share/kitinerary/extractors. The chart
README and the Reservations-and-Bookings wiki page tell Helm and Compose
users how to mount extractors there.
The variants render in the Helm Chart & Compose job sets both values,
so kubeconform checks them.
The empty-result fallback can call Google after Transitous, so lint:tx now sees two googleQuota.record() calls in TransitService.plan. Only one of them runs per request, and each is a call counter written after its own upstream request, so it cannot share a transaction with it.
Admin settings, transport, environment and MCP pages still said Transitous
is the only fallback and that Google is billed only when picked. Document
that an empty plan asks the other backend, that the Google fallback uses
only an instance key within the daily limit, and mention it in the
search_transit_routes tool description.
Logroño-Agoncillo (RJL) and fifteen other airports gained their IATA code in
OurAirports after the bundled list was built, so the airport search and the
code lookup of create_transport could not find them. The list is rebuilt from
the current data. The build now keeps every code the previous list carried
that upstream no longer lists (a scheduled-service flag flipped back, PBI moved
to a new record), since bookings and MCP calls still name them.
Refs #2897
A car or other transport booking carried its route only in the description, so
a calendar app had no place to navigate to. Without a location of its own, a
transport event now takes the endpoint the traveller has to get to: the
destination of a drive of their own (car, bicycle) and the departure of a ride
they board, the side the per-leg events already use. A car's pickup and
drop-off events are located at their own endpoints. A location the booking
states keeps winning.
Closes#2909
Claude Sonnet 5.5, Opus 5.5 and Fable 5.1 reject tool_choice type "tool" and
"any" with a 400, so every booking import through the Anthropic provider failed
on them. The request now sends tool_choice auto with parallel tool use off,
which every Anthropic model and Messages-compatible gateway accepts, and the
system prompt tells the model to answer through the emit tool exactly once. A
model that still answers in text is reported as before.
Closes#2902
Opening an entry on the share page showed the hero at no height until the
original arrived, then pushed the story down, and the lightbox stayed black
until the full photo had loaded. The hero now reserves the photo's proportions
(4:3 when none were recorded) with its thumbnail behind it, and the lightbox
shows the thumbnail and a spinner until the original is in.
The phone entry sheet probed the photo providers again on every open and added
the External photos button once the answer came, squeezing the icons of the
other two buttons. The journey screen already knows the connected providers, so
it hands them to the sheet, and the icons keep their size.
The picked neutral mood was a grey tint on a grey sheet; a picked mood is now
framed in its full colour and announced as pressed.
Closes#2907Closes#2908Closes#2905
The live trail arrived in Dawarich's own day palette, counted from the first
recorded day, while a journey colours its days by the entries. A day's line and
its pins therefore wore different colours. The trail of a day that has entries
now takes that day's colour, the way the kept recordings already did.
Closes#2906
The journey map's photo thumbnails set position:relative inline on the marker
element. That overrides the engine's .mapboxgl-marker/.maplibregl-marker
position:absolute, so every thumbnail entered the normal flow below the ones
drawn before it and the engine's translate was added on top: the further down
the list, the further the photo drifted from its spot. The location dot had the
same inline position. Both now leave position to the engine's class.
Closes#2910
The search turned an airport code into its airport by reaching into the
airports domain's data module. airports.contract.ts publishes that lookup
instead, as the import boundaries ask.
maps.helpers.ts and osm.client.ts had grown past the 1000-line limit. The
merge behind the search and suggestion lists, with its lead rules, moves to
maps-merge.helpers.ts; the search records and Photon's half of the
OpenStreetMap geocoding (search, suggestions, airport codes, identity,
reverse, element details) move to osm-geocoding.ts. OsmClient keeps the
same methods and hands them on, so no caller changes.
- The lead is Photon's own row first; a row of its name inside the lead
only stands in where Photon's row went as the same place. Rows without a
line below the name are never folded as reading the same.
- A miss of Photon where it is the geocoder no longer asks it a second time
on the way to an empty answer.
- The transport words of the identity match use a Unicode word boundary
(straße, estación) and match 駅 anywhere.
- Opening an OSM place reads the coordinate it was placed at explicitly.
- The ferry spans of a Valhalla leg move to valhallaFerries.ts, and a day
cut into pieces is asked with one options object.
- "Limit" on the allowance tiles is "Daily limit", which no locale left
as an English copy.
- Tests follow the new behaviour: no Nominatim behind Photon, zoom 13,
and the expectations of the merge and edit-distance cases.
The card names place search, address search and routing with what each
answers and what is left of today's allowance. Each service is a tile with
what is left as its figure, the day's use as a bar and used and limit as
badges; the state, the kind of installation, a key from the environment
and the time the count starts again sit as badges in the header. The
sources add the terrain tiles the routes take their heights from.
A request to the TREK Valhalla that runs out its deadline (a long bike
route) is answered 504 and no longer rests the relay for every user or a
tab for ten minutes; that one day goes to OSRM. Only an unreachable host
or a server error rests it.
A day goes to the relay with up to twenty stops per request and no pause
between pieces; a request that ends up at the public FOSSGIS Valhalla is
cut to ten and paced, as before. The 1.1 s spacing of the server applies
to the public hosts only, not to an admin's own OSRM or Valhalla. The
relay keeps its connections open and asks for maneuvers only, which the
relay contract now accepts.
The suggestions and the full search now share one merge. Photon's hits are
ranked with places that have a Wikidata link before their unlinked
namesakes, and its context bias is narrower (zoom 13). The list opens on a
lead where Photon's answer has one: a country or region always; a town,
island or sight when it is called what was typed (or a near typo of it),
lies within 1000 km of where the traveller is looking and no place of that
name ranks above it. Rows of the lead's name inside it, the same element
under two ids and rows that read the same are shown once. A three-letter
code in capitals is asked for as the airport of the bundled list.
Where Photon is the geocoder it answers alone: an empty answer and a miss
are the answer for search and imports, the reverse lookup asks again
within 50 km, a Google place is matched inside a box through Photon, and
an opened OpenStreetMap place is placed by Overpass and named by Photon.
Nominatim stays the search of installations without the TREK services or
with an operator's own Nominatim.
Measured on 5000 places from Wikidata: the right place first while typing
48 % before, 65 % now; full search 53 % to 68 %; no repeated rows.
A car booking without an end day is a one-day booking everywhere else in
the plan, but the road trip read it as a car never handed back: only its
pick-up desk stood on the road. Two one-way car bookings on one day
(Bologna to Villach, then Villach to Klagenfurt) therefore drove from the
first pick-up to the second and on to whatever came before, counting the
road between the two cities twice and never reaching Klagenfurt. This
applies to the planner, the phone and calculate_roadtrip alike, since all
three seat the desks through the shared carrierSeam.
Such a car now has its return desk on the pick-up day. A return without a
time goes back before the next car of that day is collected rather than
closing the day behind it, so each one-way car drives its own road once.
Bookings with a return day and cars without a located return desk are
unchanged.
Closes#2894
A rail, bus or ferry ticket with changes arrived as one booking per
segment, each with a linked cost for the full fare, so the trip total
was the fare times the number of segments.
Import: the KItinerary mapper now groups consecutive train segments
under one booking reference that connect (same station by name or
within 500 m, no return, within a day) into one train booking with
stops and metadata.legs, the way connecting flights already were. The
connection rule is shared between flights and ground transport, and a
stop carries the onward departure for flights as for trains, as the
planner form and the MCP tools write it. The three copied ground
mappers are folded into one.
Within one such booking a fare repeated on every segment is booked
once, and fares that differ per segment are added in whole cents. That
summing now applies to connecting flights too, which used to keep the
first leg's price only.
Segments of one order that do not join into one booking (an outward
and a return trip, a gap, a change of station named two ways without
coordinates, bus and boat tickets, Rail&Fly) are compared across the
document: when every priced transport booking of a reference carries
the same amount in the same currency, the first to depart keeps it and
the others lose it. Amounts that differ stay. This runs on what
KItinerary extracted only.
AI import: the repeated-total safety net from #2651 now covers train,
bus and ferry tickets and any mix of them with flights, still only
when the document prints that amount on a total line.
MCP and assistant: create_transport says that a ticket with several
segments is one booking with endpoints[] and legs[], and the price
fields say they hold the whole booking's cost. When a new booking
repeats the price an earlier booking with the same confirmation number
already carries in the trip's currency (its own currency, or the
trip's when it names none), create_transport and create_reservation
still save it and return a warning naming the earlier booking. REST is
unchanged.
Existing duplicated costs are left alone: they cannot be told apart
from two tickets that really cost the same, and deleting expenses
automatically would change settled balances. They can be removed in
the Costs tab.
Closes#2895
A stable release without a dialect of its own now ships the one on main only while that one has 90 days left; dialect 0 or less stops the run before anything is committed, unless allow_stale_dialect is ticked. Fetching and checking moved into registration-dialect.yml, which never fails its caller and also type-checks the module under the strict config and lints it with --max-warnings 0, since the bump commit skips CI. The bump and everything after it run whatever became of the dialect job. Prereleases ask the issuer for a dialect of their -pre.N version and build it into the image without committing it. The release check reads the module token by token against an allowlist of keywords, globals and member names, refuses any assignment to a property, computed access and regular expressions, and names Object, JSON and defineProperty among the forbidden. The issuer's kitchen-sink sample is gone; the tests use a hand-written dialect instead.
The stable release asks the issuer for the registration dialect of the
version it cuts and commits it in the bump commit. Fetching and checking run
in a job of their own that holds no write token and keeps no git
credentials: scripts/ci/check-registration-dialect.mjs refuses a file that
imports or names anything beyond hashing, then runs it on its self-check
inputs. Only a file that passes reaches the bump as an artifact. When the
issuer is unreachable or its answer fails the check, the release goes ahead
with the dialect already on main and says so in the log and the summary.
Prereleases get no dialect of their own; dev takes the stable one through
the merge of main into dev after a release, and docker-dev.yml warns while
dev still speaks the original registration or a dialect older than 150 days.
Registration now goes through a registration dialect module: the paths of
the challenge and the registration, the bytes scrypt runs over and the body
and headers of the request. Each stable release will carry the dialect the
TREK services issuer writes for it; until the first such release the module
speaks the original registration, so nothing changes on the wire.
An issuer that no longer takes the release's dialect answers 410. TREK then
says so once per process, waits a day (or what Retry-After asks) and keeps
searching through the shared pool. The issuer client only lets a dialect
register at the issuer's own paths, with X-TREK-* headers that are not the
gate's and a body of registration size.
Renewal needs no dialect, and an expired key may now be renewed for a year
instead of 30 days, the same period the issuer allows, so an installation
that was offline for long comes back without an update. The key format
ignores the new optional dialect claim; the shared vectors gain one key that
carries it.
The marker effect reads the card handler through the component's own ref
again, so exhaustive-deps sees a ref it knows, and the card hook takes it
alongside the map's other refs. The pin event assertion types the mock
calls it maps under strict mode.
The phone card logic moves out of JourneyMap into useEntryCard, which brings
the component back under the line limit. The new card tests reach the
card element through a guard instead of non-null assertions, the public
hook test reads the hook through renderHook, the GL card tests use
mock.lastCall where Array#at is outside the client's lib, and the
complexity baseline drops to the count the branch now holds.
listPublicEntries changed for #2886, and the repository rules move a
touched QueryBuilder call to the entity methods or Kysely. Kysely keeps
the method's answer what it was, the stored row as SELECT * reads it,
which the entity methods would hand back as entities with a journey
reference in place of journey_id. The query-api baseline drops to 8.
ConnectionStop, EntryConnections and JourneyMapStop are only used in the
signatures of their own modules, and knip counts an export nothing
imports. The connection tests take the FE- prefix and header the other
journey tests carry.
The card a phone host opens its entry from was still a tooltip bound to
the pin, so Leaflet opened and closed it on the pin's hover and focus. A
tap on the card reaches a phone browser as the pointer leaving the pin
first: the pin's mouseout closed the card between the tap and its click,
and on iOS a content change in that phase can drop the click entirely.
Where a host wires onMarkerCardClick, the pins now bind no tooltip. One
card per map stands over the active pin, the way the GL popup does: it
follows highlightMarker, leaves when no entry is active or the entry
pins are switched off, and only a change of entry moves it. A press on
it never reaches the map. Desktop hosts keep the hover label on each pin.
The card tests run against the handlers the pins really get and the
card the map really adds, and the file parses again: two test names held
an unescaped apostrophe.
On the phone journey screen and the public share page on a phone, a tap on
a numbered pin brings up the entry's card above it, but the card could not
be tapped: the GL popup and the Leaflet tooltip both let the pointer fall
through to the map, and no phone host gave the card anything to do.
The map now takes onMarkerCardClick, and both renderers make the card open
that entry once a host wires it. The GL popup handles it in its delegated
listener next to the photo strip, which keeps opening the photos where the
desktop asks for that. Leaflet handles it on the tooltip element itself:
a press that reached the map container let Leaflet focus the map, which
blurred the pin and closed the card before its click landed. The phone
screen and the timeline open the entry through the carousel hook, the
same way a tap on the active card does. The desktop asides keep the card
as a hover label.
The unused hideMarkerTooltip prop goes, together with the docs that still
said the phone hides the card.
Closes#2885
The share link has counted its map stops through the whole journey since
#1962, while the owner's journal still restarted the count every day, so
the same stop read 66 on the link and 6 in the journal.
Both views now take their stops from one helper, journeyMapStops: located,
non-skeleton entries in the server's reading order (date, sort_order, id),
numbered across the journey. A draft keeps its pin in the journal but gets
no number, so the published stops after it carry exactly the numbers the
share link shows. The public entry query gains the id tiebreak the owner
query already had, so entries sharing a sort_order come back in the same
order on both sides.
Closes#2886
The journey map joins every located entry with a faint dashed line in the
order of its pins. It drew that line through everything, also across days
whose real route is on the map as a kept recording or the live Dawarich
trail, so straight guesses fanned out over the recorded track. The phone
has no layer switches, so the line could not even be hidden there.
A recorded track now carries the day it belongs to (an optional `date` on
the track contract, set by every builder of recorded lines: kept days, the
shared page's recordings, the live Dawarich trail and the day import
preview). A pure model splits the connection into the stretches still open
and those whose days were all recorded, the night between two recorded days
included. Both renderers draw only the open stretches with the entries; the
covered ones sit on a layer of their own that shows while the reader hides
what was recorded, so the guess comes back exactly where nothing else
explains the way.
The live trail is now matched to a kept day by that date instead of reading
the track name as one.
Closes#2888
The journey map grouped its photo thumbnails and its plan and stay marks
on a grid laid over container pixels. Every pan shifted that grid against
the map, so photos near a cell edge changed groups after each move: the
count badges jumped and the thumbnail switched to another photo, although
the zoom never changed.
Both renderers now bucket in world pixels at the current zoom, Leaflet
through its own CRS (so GCJ-02 basemaps stay right) and GL through a
plain Web Mercator projection that ignores centre, bearing and pitch.
Panning keeps every group as it is; only a zoom regroups.
Closes#2889
Vini