feat: implement multi transport types for the same dataset by diogodanielsoaresferreira · Pull Request #2579 · TimefoldAI/timefold-solver

diogodanielsoaresferreira · 2026-08-10T17:32:47Z

Allow to solve routing problems with more than one transportType.

Created additional properties in Location so that we keep the hot path without the Map access.

Related with https://github.com/TimefoldAI/timefold-platform/issues/2690

sonarqubecloud · 2026-08-10T17:54:58Z

Quality Gate failed

Failed conditions
31.1% Coverage on New Code (required ≥ 70%)

Add a TransportType value type and per-mode travel-time/distance matrices on Location, so the same routing problem can use more than one transport profile (e.g. car + bike). The default CAR mode keeps the existing scalar/ timeframe fields and index-cache fast path; extra modes are stored in opt-in maps that stay null for single-mode problems. The TravelTimeMatrixEnricher fetches one matrix set per transport type (each resolving to its own OSRM instance), and map-service.transport-type now accepts a comma-separated list of modes (first entry is primary). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

…type only in solver

timefold-automations · 2026-10-01T11:39:14Z

Warning

Threat Detection Engine Failure — The analysis engine could not complete. This is a tooling failure, not a security finding.

What happened

The threat detection engine failed to produce results.

Review the workflow run logs for details.

gh-aw-workflow-id: solver-docs-drift-review

The documentation drift reported earlier is gone; this pull request now covers the transport-type API and multi-transport enrichment behavior.

Generated by Solver Docs Drift Review · copilot · gpt56 · 8.94 AIC · ⊞ 25.2K · ◷

rsynek · 2026-10-01T13:02:15Z

+@Schema(description = "The type of transport used (car, bicycle, ... ).")
+public enum TransportType {
+
+    CAR("car"),


Do we need the lower-case value? If we drop it, we don't need to override the enum methods nor creating the constructor.

rsynek · 2026-10-01T13:08:12Z

+
+    @JsonCreator
+    public static TransportType of(String value) {
+        Objects.requireNonNull(value, "TransportType value must not be null.");


Compared to the constructor, we don't check if the value is empty.

But see my proposal about simplifying the enum, which would leave only this method and eliminate the constructor.

rsynek · 2026-10-01T13:21:06Z

        String maxDistanceFromRoadOption = maxDistanceFromRoad.map(MapServiceOptions::getMaxDistanceFromRoadOption).orElse("");
-        String transportTypeOption = transportType.map(MapServiceOptions::getTransportTypeOption).orElse("");
+        String transportTypeOption = transportType == null || transportType.isAutoSelect()
+                ? ""


So, for the map service, the default is auto-select?

rsynek · 2026-10-01T13:23:00Z

+
+    public String getOptions(TransportType transportType) {
+        return getOptions((String) null, transportType);
+    }


Instead of adding overloaded methods and passing nulls to override what's coming from the configuration parameters, we may decouple the config from these getOptions() parameters by introducing a builder that registers the overrides and only then builds the options String.

rsynek · 2026-10-01T13:24:10Z

    List<Location> getLocations();

+    default List<TransportType> getTransportTypes() {
+        return List.of();


If this leads to auto-select, let's return it here already; it will become more obvious.

rsynek · 2026-10-01T13:58:14Z

+A deployment may additionally be restricted to a subset of transport types with
+`timefold.platform.map-service.allowed-transport-types`, a comma-separated list; when it is not set, every transport
+type is allowed.


Who restricts that?

I assume this comes as a platform config, in which case the model developer cannot much influence that if they don't run the model locally.

rsynek · 2026-10-01T13:59:47Z

+If your model stores matrices itself rather than relying on the enricher, the setters take the transport type too:
+
+[source,java,options="nowrap"]
+----
+location.setTravelTimeMatrix(TransportType.BICYCLE, travelTimeMatrix);
+location.setDistanceMatrix(TransportType.BICYCLE, distanceMatrix);


Let's not expose this as an API. We should take care of the matrices, not the user.

rsynek · 2026-10-01T14:10:14Z

+Each of those transport types is validated against the allowed transport types and then fetched in its own map service
+request, so a dataset using two transport types results in two requests and two sets of matrices per `Location`.
+A dataset that uses a transport type the deployment is not allowed to route with is rejected before any request is made.
+When the solution returns an empty list, the default transport type is used.
+
+One of the transport types is the *primary* one: the configured one when it is fixed, otherwise `car` if the dataset
+uses it, and the first one in declaration order otherwise. The primary transport type is special in two ways:
+
+* The model-level map metadata — the locations that are not in the map, and the resolved map region — comes from its
+request only.
+* Its matrices also back the lookups that do not name a transport type, so those keep answering on a deployment
+configured for a single non-default transport type.


Why should the user care? This text explains how we implement the transport types, but that's the complexity we are trying to remove from the user.

rsynek · 2026-10-01T14:28:11Z

+    private DistanceMatrix travelTimeMatrixForMode(TransportType transportType, OffsetDateTime departureTime) {
+        if (isDefaultMode(transportType)) {
+            return hasTimeframeMatrices(travelTimesByTimeframe)
+                    ? resolveTimeframeMatrix(travelTimesByTimeframe, departureTime, "travel time")
+                    : travelTimeMatrix;
+        }
+        DistanceMatrix[] byTimeframe = travelTimesByTimeframeByMode == null
+                ? null
+                : travelTimesByTimeframeByMode[transportType.ordinal()];
+        if (byTimeframe != null && timeframeIndexResolver != null) {
+            return resolveTimeframeMatrix(byTimeframe, departureTime, "travel time");
+        }
+        return travelTimeMatrixByMode == null ? null : travelTimeMatrixByMode[transportType.ordinal()];
+    }
+
+    private DistanceMatrix distanceMatrixForMode(TransportType transportType) {
+        if (isDefaultMode(transportType)) {
+            return distanceMatrix;
+        }
+        return distanceMatrixByMode == null ? null : distanceMatrixByMode[transportType.ordinal()];
+    }
+
+    private DistanceMatrix distanceMatrixForMode(TransportType transportType, OffsetDateTime departureTime) {
+        if (isDefaultMode(transportType)) {
+            return hasTimeframeMatrices(distancesByTimeframe)
+                    ? resolveTimeframeMatrix(distancesByTimeframe, departureTime, "distance")
+                    : distanceMatrix;
+        }
+        DistanceMatrix[] byTimeframe = distancesByTimeframeByMode == null
+                ? null
+                : distancesByTimeframeByMode[transportType.ordinal()];
+        if (byTimeframe != null && timeframeIndexResolver != null) {
+            return resolveTimeframeMatrix(byTimeframe, departureTime, "distance");
+        }
+        return distanceMatrixByMode == null ? null : distanceMatrixByMode[transportType.ordinal()];
+    }


According to SonarCloud, these methods that resolve the distance matrix are not covered by tests. We likely miss more tests that would indirectly trigger them.

rsynek · 2026-10-01T14:31:16Z

+        return travelTimeMatrixByMode == null ? null : travelTimeMatrixByMode[transportType.ordinal()];
+    }
+
+    private DistanceMatrix travelTimeMatrixForMode(TransportType transportType, OffsetDateTime departureTime) {


Since we are adding more logic to resolve the matrix on the hot path, do we have any benchmarks showing the impact on move evaluation speed for multiple vs. a single transport type?

diogodanielsoaresferreira requested review from cristianonicolai, mswiderski and winklerm August 10, 2026 17:32

diogodanielsoaresferreira temporarily deployed to internal August 10, 2026 17:32 — with GitHub Actions Inactive

diogodanielsoaresferreira temporarily deployed to internal August 24, 2026 11:02 — with GitHub Actions Inactive

diogodanielsoaresferreira and others added 2 commits September 4, 2026 07:45

feat: remove comments

862e09d

winklerm force-pushed the draft/transport-type-aware-location branch from ea51de8 to 862e09d Compare September 4, 2026 05:46

winklerm temporarily deployed to internal September 4, 2026 05:46 — with GitHub Actions Inactive

chore: change TransportType to enum to quick fix OpenAPI spec

9173ac6

winklerm temporarily deployed to internal September 4, 2026 09:32 — with GitHub Actions Inactive

feat: add Transport types and replace map of distance matrixes by array

0f53d1a

diogodanielsoaresferreira temporarily deployed to internal September 10, 2026 15:12 — with GitHub Actions Inactive

feat: remove options from argument in MapsServiceOptionsSupplier

829157e

diogodanielsoaresferreira temporarily deployed to internal September 10, 2026 15:29 — with GitHub Actions Inactive

diogodanielsoaresferreira deployed to internal September 10, 2026 15:29 — with GitHub Actions Active

Merge branch 'main' into draft/transport-type-aware-location

772ab5e

diogodanielsoaresferreira deployed to internal September 16, 2026 09:09 — with GitHub Actions Active

Update transport types

2a66083

diogodanielsoaresferreira deployed to internal September 16, 2026 12:23 — with GitHub Actions Active

feat: add allowed transport type to maps service client

a5f4a27

diogodanielsoaresferreira deployed to internal September 21, 2026 15:22 — with GitHub Actions Active

diogodanielsoaresferreira deployed to internal September 24, 2026 09:13 — with GitHub Actions Active

diogodanielsoaresferreira marked this pull request as ready for review September 24, 2026 09:13

diogodanielsoaresferreira requested a review from a team September 24, 2026 09:13

diogodanielsoaresferreira requested review from rsynek and triceo as code owners September 24, 2026 09:13

diogodanielsoaresferreira deployed to internal September 24, 2026 09:13 — with GitHub Actions Active

diogodanielsoaresferreira changed the title ~~Draft/transport type aware location~~ feat: implement multi transport types for the same dataset Sep 24, 2026

This comment has been minimized.

Sign in to view

feat: make empty transport type the default and convert to transport …

192f7da

…type only in solver

diogodanielsoaresferreira deployed to internal September 24, 2026 13:00 — with GitHub Actions Active

This comment has been minimized.

Sign in to view

feat: add documentation about transport types

bbf4d33

diogodanielsoaresferreira requested a review from TomCools as a code owner September 24, 2026 13:25

diogodanielsoaresferreira deployed to internal September 24, 2026 13:25 — with GitHub Actions Active

diogodanielsoaresferreira deployed to documentation (preview) September 24, 2026 13:25 — with GitHub Actions View deployment

This comment has been minimized.

Sign in to view

Merge branch 'main' into draft/transport-type-aware-location

b2a8e83

diogodanielsoaresferreira deployed to internal October 1, 2026 11:19 — with GitHub Actions Active

diogodanielsoaresferreira deployed to documentation (preview) October 1, 2026 11:20 — with GitHub Actions View deployment

rsynek requested changes Oct 1, 2026

View reviewed changes

rsynek reviewed Oct 1, 2026

View reviewed changes

Conversation

diogodanielsoaresferreira commented Aug 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sonarqubecloud Bot commented Aug 10, 2026

Quality Gate failed

Uh oh!

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

timefold-automations Bot commented Oct 1, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

This branch was successfully deployed

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

diogodanielsoaresferreira commented Aug 10, 2026 •

edited

Loading