« Changelog History


Graphhopper v0.4 Release Notes

Release Date: 2015-03-18 // about 9 years ago
    • translation key turn changed and merged with left etc into turn_left, turn_right etc
    • create location index before preparation in the GraphHopper class
    • 🚚 encodingManager.getSingle() is removed and flagEncoder list is no longer sorted, the first vehicle is used for CH preparation
    • ✂ removed LocationIndexTreeSC, use new LocationIndexTree(levelGraph.getBaseGraph(), directory) instead
    • getLevel and setLevel do no longer automatically increase node count, use getNodeAccess.ensureNode for that
    • 👀 normal algorithms are now possible on prepared graph use getBaseGraph, see #116
    • GHResponse no longer has isFound method, use !hasErrors instead
    • 🔀 merged unused Edge class into EdgeEntry
    • astar and astarbi are now both none-heuristic and take parameters for beeline approximation: astar.approximation=BeelineSimplification|BeelineAccurate or astarbi.approximation=...
    • 👍 making GPX export according to the schema to support import from various tools like basecamp
    • 🔨 refactoring: AllEdgesIterator.getMaxId is now named getCount
    • 🚚 major change of internal API: moved method "Path RoutingAlgorithm.calcPath(QueryResult,QueryResult)" to a helper method QueryGraph.lookup, call queryResult.getClosestNode for the calcPath(nodeFrom,nodeTo) method
    • no cachedWays and cachedPoints in Path anymore
    • 🚚 Path.findInstruction was moved to InstructionList.find
    • if start and end point are identical an algorithm will find the path consisting only of one node, one point and one instruction (finish instruction), but without edges
    • 0️⃣ astarbi has new default values for approximation (false) and approximation_factor (1.2) in RoutingAlgorithmFactorySimple
    • instead of strings use the variables in AlgorithmOptions to specify an algorithm
    • 👉 use RoutingAlgorithmFactorySimple instead of RoutingAlgorithmFactory, also more consistent algorithm preparation handling due to new AlgorithmOptions, therefor removed NoOpAlgorithmPreparation
    • 👀 GHResponse.getXX methods now fail fast (throw an exception) if an error while route calculation occurred. See #287
    • renamed less often used URL parameter 'min_path_precision' to way_point_max_distance which makes it identical to the setWayPointMaxDistance method used for simplification at OSMImport
    • ✂ removed douglas.minprecision from Java API ghRequest.hints => use wayPointMaxDistance instead
    • 👍 encoder.supportTurnCost is replaced by encoder.supports(TurnWeighting.class)
    • 📜 CmdArgs is now a Map instead Map. The value will be parsed up on every getXY call, makes storing string vs. object less error-prone
    • ✂ removed GHRequest.getHint, instead use the provided methods in GHRequest.getHints().getXY and GHRequest.getHints().put
    • important graph incompatibility as properties cannot be loaded. renamed osmreader.bytesForFlags to graph.bytesForFlags, renamed config property osmreader.acceptWay to graph.flagEncoders
    • ✅ default weighting is now fastest, fixing #261
    • 👀 moved method GraphHopper.main into tools module and class com.graphhopper.tools.Import, see #250
    • 🔨 refactored GraphHopper.createWeighting to accept more than one configuration option, see #237
    • 🔨 refactored GraphHopper.disableCHShortcuts to setCHEnable(boolean)
    • moving the boolean parameter of GraphHopper.setInMemory into a separate method setStoreOnFlush
    • renaming of GraphHopper.setCHShortcuts to setCHWeighting, as well as the property prepare.chShortcuts to prepare.chWeighting
    • 👀 jsonp is disabled by default. You need to enable it in the config.properties, see the config-example.properties
    • 🔧 EncodingManager cannot be null in GraphHopperStorage. If you need to parse EncodingManager configuration from existing graph use EncodingManager.create
    • 🔧 no reflection done in EncodingManager which improves portability and makes configuration of encoders possible before adding to manager
    • ✂ removed dijkstraNativebi as no performance advantage but maintenance disadvantage and similar to oneToManyDijkstra
    • 📚 to provide context for turn costs we needed to add prevEdgeId into Weighting.calcWeight, see new documentation
    • 🔒 with the introduction of lock protection mechanism (see #112) GraphHopper needs always write access, see also #217
    • 🆕 new GraphHopper.clean method to remove the graph directory via Java API