v1.8.0

Added

• Support skills in periodic heuristic (first_solution_strategy='periodic') #194
• Implementation of vehicle_trips relation: the routes can be successive or with a minimum duration lapse in between #123
• CSV headers adapts to the language provided through HTTP_ACCEPT_LANGUAGE header to facilitate import in Mapotempo-Web #196
• Return route day/date and visits' index in result #196
• Treat complex shipments (multi-pickup-single-delivery and single-pickup-multi-delivery) as multiple simple shipments internally to increase performance #261
• Prioritize the vehicles (and trips of vehicle_trips relation) via modifying the fixed costs so that first vehicles (and trips) are preferred over the latter ones #266
• Document return codes #224
• OR-Tools wrapper can use initial capacity value #245

Changed

• Improve cases where a service has two visits in periodic heuristic: ensure that the second visit can be assigned to the right day #227
• Relation lapse becomes lapses : we can now provide a specific lapse for every consecutive ID in the relation #265

Removed

• Field trips in vehicle model. Use vehicle_trips relation instead #123

Fixed

• VROOM was used incorrectly in various cases: negative quantities, vehicle duration, activity position #223 #242
• Capacity violation in periodic heuristic algorithm (first_solution_strategy='periodic') #227
• Service timewindows without an end were not respected #262
• total_time, total_travel_time and related values are correctly calculated #237

v1.7.0

Added

• Corresponding vehicle_id is returned within each service's skills if problem is partitioned with vehicle entity #110
• Support initial routes and skills in split_solve (max_split_size) algorithm #140
• Support relations (order, same_route, sequence, shipment) in split_solve algorithm (max_split_size) and partitions ([:configuration][:preprocessing][:partitions]) PR145
• split_solve algorithm (max_split_size) respects relations (vehicle_trips, meetup, minimum_duration_lapse, maximum_duration_lapse, minimum_day_lapse, maximum_day_lapse) PR145
• Return geojsons structure according to geometry parameter #165

Changed

• Bump grape to v1.5.0 - It speeds up the processing of POST requests for nested problems (up to 20 times faster, ruby-grape/grape#2096) #107
• Bump grape-swagger to v1.3.0 - The documentation is now correctly generated. It allows SDK generation. #107
• The fields costs introduced in v0.1.5 is renamed cost_details to avoid confusion with field cost #107
• Bump VROOM to v1.8.0 and start using the features integrated since v1.3.0 #107
• Bump OR-Tools v7.8 in docker image #107
• VROOM were previously always called synchronously, it is now reserved to a set of effective router_mode (:car, :truck_medium) within a limit of points (<200). #107
• Heuristic selection (first_solution_strategy='self_selection') takes into account the supplied initial routes (routes) and the best solution is used as the initial route #159

Removed

• geometry_polyline parameter now be provided through geometry parameter #165

Fixed

• unassigned output were in some cases returning the key shipment_id instead of pickup_shipment_id and delivery_shipment_id #107
• Uniformize route content and always return original_vehicle_id #107
• Infeasibility detection of services with negative quantity #111

v1.6.0

Added

• Allow support for semicolumn and tabulation in CSV files
• Initial routes are now correctly grouped when using partitions (Especially for Scheduling Problems). However, this feature is still under development within split based algorithms (split_solve and dichotomous)
• position field for activities
• Unfeasible missions are removed from routes #31
• Detailed route costs both for routes and solutions #46
• [WIP] Allow multidepot within clustering. For now, this only influences clusters initialization #56
• Introduce direct shipments: pickup and delivery must be consecutive #51
• Return original ids for missions and vehicles #73
• Return total_waiting_time #83
• Introduce minimize_days_worked parameter for scheduling heuristic. The default behavior of scheduling heuristic is now to balance work load over the period, the introduced parameter allows to return to the previous behavior #89
• Detect inconsistent same_point_day definition #76

Changed

• Cache is now part of wrappers hash
• Collect current distance from OR-Tools
• Refactoring of the chaining of the various solving procedures (Rework define_process)
• Now use OR-Tools asserts instead or recompiling the entire project
• Reject too small lapses
• Bump OR-Tools to v7.8
• Various edits, improvements and refactoring within scheduling heuristic
• Scheduling heuristic now empties under-filled routes to reassign them #28
• Improve unfeasible service detection performance #28 #65
• Avoid heuristic selection when unnecessary #31
• Shipments are now tolerated within split_clustering if the pickup or the delivery is located at a depot #46
• Reduce clustering default restarts from 50 to 10 #49
• Replace balanced_kmeans by balanced_vrp_clustering gem #49
• Factorize get operation status codes #52
• VROOM is no more refused due to initial routes #59
• Improved kill of jobs within scheduling heuristic #57
• CSV is now only returned for completed jobs #60
• Update fronts accordingly to CSV edits #61
• Ensure matrix, matrix_id and matrix_index consistency #62
• Dump filtered problems #62
• Default multi_trips value is now 1 #47
• Bump VROOM to v1.5.0 #66
• Reduce memory usage for scheduling problems #71
• Reduce OR-Tools computation time within scheduling heuristic #78
• Rework route referents definition within scheduling heuristic, in order to improve routes initialization #89
• Bump Ruby to 2.5.5 #97
• Delegate quantity rounding to optimizer-ortools #98
• Improve Force start & breaks through optimizer-ortools #98

Removed

• solver_parameter from the internal model, it is replaced by solver and first_solution_strategy parameters
• Relation id was mandatory, but was never used or returned #31
• Vehicle sort by day index #31
• overall_duration parameter from API as it is not functional #46

Fixed

• Resolution time when using split_independent or max_split procedures
• Alternative activities with empties and fills
• Shift computation for sheduling problems
• Sheduling heuristic now rejects shipments and rests #28
• Reject empty problems #35
• Negative time limits #35
• Expand multi_trips only once #36
• Remove data from already started jobs #38
• Clean interrupted working jobs at startup #38
• Error when delete operation is invoked immedialty after get #38
• Coerce first_solution_strategy into array #38
• Avoid unnecessary route generation #46
• Correctly handle consecutive delete operations #52
• Returned http error codes #62
• Split by skills generating unexpected sub-problems #47
• Remove duplicated empty routes for the same vehicle #50
• Return total_time and total_travel_distance from scheduling heuristic #63
• Infinite loop due to impossible split within dichotomous #67
• Avoid depot duplication #72
• Wrong number of visits #86
• Parsing of intermediate protobuf file returned by optimizer-ortools #87
• Uniformize wrappers output accordingly to API documentation #83

Deprecated

• Positions are no longer relations and must be defined at activity level
• Route indice is now index #31