From f6fa781fa49118fd83dede4a36d9551ba41b3264 Mon Sep 17 00:00:00 2001
From: Laurent Perron <lperron@google.com>
Date: Tue, 28 Sep 2021 15:58:37 +0200
Subject: [PATCH] split ortools/data into ortools/scheduling and
 ortools/packing

---
 cmake/cpp.cmake                               |  10 +-
 cmake/python.cmake                            |   9 +-
 examples/cpp/BUILD                            |   4 +-
 examples/cpp/jobshop_sat.cc                   |  18 +-
 examples/python/rcpsp_sat.py                  |   2 +-
 makefiles/Makefile.cpp.mk                     |  61 +--
 makefiles/Makefile.gen.mk                     | 136 +++---
 makefiles/Makefile.python.mk                  |  68 +--
 ortools/data/set_covering_data.h              |  59 ---
 ortools/data/set_covering_parser.cc           | 207 ---------
 ortools/data/set_covering_parser.h            | 113 -----
 ortools/packing/BUILD                         |   3 +
 ortools/packing/CMakeLists.txt                |  21 +
 ortools/{data => packing}/__init__.py         |   2 +-
 ortools/packing/arc_flow_builder.cc           | 408 ++++++++++++++++++
 ortools/packing/arc_flow_builder.h            |  92 ++++
 ortools/packing/arc_flow_solver.cc            | 192 +++++++++
 .../arc_flow_solver.h}                        |  33 +-
 .../vector_bin_packing.proto                  |   6 +-
 .../vector_bin_packing_parser.cc              |   8 +-
 .../vector_bin_packing_parser.h               |  14 +-
 ortools/{data => scheduling}/BUILD            |   0
 ortools/{data => scheduling}/CMakeLists.txt   |   4 +-
 ortools/scheduling/__init__.py                |  15 +
 .../jobshop_scheduling.proto                  |   6 +-
 .../jobshop_scheduling_parser.cc              |   8 +-
 .../jobshop_scheduling_parser.h               |  12 +-
 .../python/CMakeLists.txt                     |   4 +-
 ortools/{data => scheduling}/python/rcpsp.i   |  22 +-
 ortools/{data => scheduling}/rcpsp.proto      |   6 +-
 ortools/{data => scheduling}/rcpsp_parser.cc  |   8 +-
 ortools/{data => scheduling}/rcpsp_parser.h   |  14 +-
 tools/generate_all_deps.sh                    |   4 +-
 33 files changed, 956 insertions(+), 613 deletions(-)
 delete mode 100644 ortools/data/set_covering_data.h
 delete mode 100644 ortools/data/set_covering_parser.cc
 delete mode 100644 ortools/data/set_covering_parser.h
 create mode 100644 ortools/packing/BUILD
 create mode 100644 ortools/packing/CMakeLists.txt
 rename ortools/{data => packing}/__init__.py (97%)
 create mode 100644 ortools/packing/arc_flow_builder.cc
 create mode 100644 ortools/packing/arc_flow_builder.h
 create mode 100644 ortools/packing/arc_flow_solver.cc
 rename ortools/{data/set_covering_data.cc => packing/arc_flow_solver.h} (53%)
 rename ortools/{data => packing}/vector_bin_packing.proto (96%)
 rename ortools/{data => packing}/vector_bin_packing_parser.cc (95%)
 rename ortools/{data => packing}/vector_bin_packing_parser.h (83%)
 rename ortools/{data => scheduling}/BUILD (100%)
 rename ortools/{data => scheduling}/CMakeLists.txt (83%)
 create mode 100644 ortools/scheduling/__init__.py
 rename ortools/{data => scheduling}/jobshop_scheduling.proto (97%)
 rename ortools/{data => scheduling}/jobshop_scheduling_parser.cc (99%)
 rename ortools/{data => scheduling}/jobshop_scheduling_parser.h (89%)
 rename ortools/{data => scheduling}/python/CMakeLists.txt (94%)
 rename ortools/{data => scheduling}/python/rcpsp.i (53%)
 rename ortools/{data => scheduling}/rcpsp.proto (97%)
 rename ortools/{data => scheduling}/rcpsp_parser.cc (99%)
 rename ortools/{data => scheduling}/rcpsp_parser.h (87%)

diff --git a/cmake/cpp.cmake b/cmake/cpp.cmake
index 649422642e..633005c0cb 100644
--- a/cmake/cpp.cmake
+++ b/cmake/cpp.cmake
@@ -139,13 +139,14 @@ set(PROTO_SRCS)
 file(GLOB_RECURSE proto_files RELATIVE ${PROJECT_SOURCE_DIR}
   "ortools/bop/*.proto"
   "ortools/constraint_solver/*.proto"
-  "ortools/data/*.proto"
   "ortools/glop/*.proto"
   "ortools/graph/*.proto"
   "ortools/linear_solver/*.proto"
-  "ortools/sat/*.proto"
-  "ortools/util/*.proto"
   "ortools/linear_solver/*.proto"
+  "ortools/packing/*.proto"
+  "ortools/sat/*.proto"
+  "ortools/scheduling/*.proto"
+  "ortools/util/*.proto"
   )
 if(USE_SCIP)
   file(GLOB_RECURSE gscip_proto_files RELATIVE ${PROJECT_SOURCE_DIR} "ortools/gscip/*.proto")
@@ -209,7 +210,6 @@ foreach(SUBPROJECT IN ITEMS
  base
  bop
  constraint_solver
- data
  ${GSCIP_DIR}
  glop
  graph
@@ -217,8 +217,10 @@ foreach(SUBPROJECT IN ITEMS
  init
  linear_solver
  lp_data
+ packing
  port
  sat
+ scheduling
  util)
   add_subdirectory(ortools/${SUBPROJECT})
   #target_link_libraries(${PROJECT_NAME} PRIVATE ${PROJECT_NAME}_${SUBPROJECT})
diff --git a/cmake/python.cmake b/cmake/python.cmake
index c746de386a..dad8afceb1 100644
--- a/cmake/python.cmake
+++ b/cmake/python.cmake
@@ -68,10 +68,11 @@ search_python_module(mypy-protobuf)
 set(PROTO_PYS)
 file(GLOB_RECURSE proto_py_files RELATIVE ${PROJECT_SOURCE_DIR}
   "ortools/constraint_solver/*.proto"
-  "ortools/data/*.proto"
   "ortools/linear_solver/*.proto"
+  "ortools/packing/*.proto"
   "ortools/sat/*.proto"
   "ortools/util/*.proto"
+  "ortools/scheduling/*.proto"
   )
 list(REMOVE_ITEM proto_py_files "ortools/constraint_solver/demon_profiler.proto")
 foreach(PROTO_FILE IN LISTS proto_py_files)
@@ -106,7 +107,7 @@ if(USE_COINOR)
 endif()
 list(APPEND CMAKE_SWIG_FLAGS ${FLAGS} "-I${PROJECT_SOURCE_DIR}")
 
-foreach(SUBPROJECT IN ITEMS init algorithms graph linear_solver constraint_solver sat data util)
+foreach(SUBPROJECT IN ITEMS init algorithms graph linear_solver constraint_solver sat scheduling util)
   add_subdirectory(ortools/${SUBPROJECT}/python)
 endforeach()
 
@@ -122,7 +123,7 @@ file(COPY ortools/__init__.py DESTINATION python/${PROJECT_NAME}/linear_solver)
 file(COPY ortools/__init__.py DESTINATION python/${PROJECT_NAME}/constraint_solver)
 file(COPY ortools/__init__.py DESTINATION python/${PROJECT_NAME}/sat)
 file(COPY ortools/__init__.py DESTINATION python/${PROJECT_NAME}/sat/python)
-file(COPY ortools/__init__.py DESTINATION python/${PROJECT_NAME}/data)
+file(COPY ortools/__init__.py DESTINATION python/${PROJECT_NAME}/scheduling)
 file(COPY ortools/__init__.py DESTINATION python/${PROJECT_NAME}/util)
 
 file(COPY
@@ -167,7 +168,7 @@ add_custom_target(python_package ALL
   COMMAND ${CMAKE_COMMAND} -E copy $<TARGET_FILE:pywrapcp> ${PROJECT_NAME}/constraint_solver
   COMMAND ${CMAKE_COMMAND} -E copy $<TARGET_FILE:pywraplp> ${PROJECT_NAME}/linear_solver
   COMMAND ${CMAKE_COMMAND} -E copy $<TARGET_FILE:pywrapsat> ${PROJECT_NAME}/sat
-  COMMAND ${CMAKE_COMMAND} -E copy $<TARGET_FILE:pywraprcpsp> ${PROJECT_NAME}/data
+  COMMAND ${CMAKE_COMMAND} -E copy $<TARGET_FILE:pywraprcpsp> ${PROJECT_NAME}/scheduling
   COMMAND ${CMAKE_COMMAND} -E copy $<TARGET_FILE:sorted_interval_list> ${PROJECT_NAME}/util
   #COMMAND ${Python3_EXECUTABLE} setup.py bdist_egg bdist_wheel
   COMMAND ${Python3_EXECUTABLE} setup.py bdist_wheel
diff --git a/examples/cpp/BUILD b/examples/cpp/BUILD
index 5eaa63d565..b6ae1b265d 100644
--- a/examples/cpp/BUILD
+++ b/examples/cpp/BUILD
@@ -240,8 +240,6 @@ cc_binary(
         "//ortools/base:file",
         "@com_google_absl//absl/strings",
         "//ortools/base:timer",
-        "//ortools/data:jobshop_scheduling_cc_proto",
-        "//ortools/data:jobshop_scheduling_parser",
         "//ortools/sat:cp_model",
         "//ortools/sat:cp_model_solver",
         "//ortools/sat:disjunctive",
@@ -251,6 +249,8 @@ cc_binary(
         "//ortools/sat:optimization",
         "//ortools/sat:precedences",
         "//ortools/sat:sat_solver",
+        "//ortools/scheduling:jobshop_scheduling_cc_proto",
+        "//ortools/scheduling:jobshop_scheduling_parser",
     ],
 )
 
diff --git a/examples/cpp/jobshop_sat.cc b/examples/cpp/jobshop_sat.cc
index 15236248d6..fc749744d9 100644
--- a/examples/cpp/jobshop_sat.cc
+++ b/examples/cpp/jobshop_sat.cc
@@ -26,12 +26,12 @@
 #include "google/protobuf/wrappers.pb.h"
 #include "ortools/base/logging.h"
 #include "ortools/base/timer.h"
-#include "ortools/data/jobshop_scheduling.pb.h"
-#include "ortools/data/jobshop_scheduling_parser.h"
 #include "ortools/graph/connected_components.h"
 #include "ortools/sat/cp_model.h"
 #include "ortools/sat/cp_model.pb.h"
 #include "ortools/sat/model.h"
+#include "ortools/scheduling/jobshop_scheduling.pb.h"
+#include "ortools/scheduling/jobshop_scheduling_parser.h"
 
 ABSL_FLAG(std::string, input, "", "Jobshop data file name.");
 ABSL_FLAG(std::string, params, "", "Sat parameters in text proto format.");
@@ -49,12 +49,12 @@ ABSL_FLAG(bool, display_model, false, "Display jobshop proto before solving.");
 ABSL_FLAG(bool, display_sat_model, false, "Display sat proto before solving.");
 ABSL_FLAG(int, horizon, -1, "Override horizon computation.");
 
-using operations_research::data::jssp::Job;
-using operations_research::data::jssp::JobPrecedence;
-using operations_research::data::jssp::JsspInputProblem;
-using operations_research::data::jssp::Machine;
-using operations_research::data::jssp::Task;
-using operations_research::data::jssp::TransitionTimeMatrix;
+using operations_research::scheduling::jssp::Job;
+using operations_research::scheduling::jssp::JobPrecedence;
+using operations_research::scheduling::jssp::JsspInputProblem;
+using operations_research::scheduling::jssp::Machine;
+using operations_research::scheduling::jssp::Task;
+using operations_research::scheduling::jssp::TransitionTimeMatrix;
 
 namespace operations_research {
 namespace sat {
@@ -871,7 +871,7 @@ int main(int argc, char** argv) {
     LOG(FATAL) << "Please supply a data file with --input=";
   }
 
-  operations_research::data::jssp::JsspParser parser;
+  operations_research::scheduling::jssp::JsspParser parser;
   CHECK(parser.ParseFile(absl::GetFlag(FLAGS_input)));
   operations_research::sat::Solve(parser.problem());
   return EXIT_SUCCESS;
diff --git a/examples/python/rcpsp_sat.py b/examples/python/rcpsp_sat.py
index 57e9855a23..a8c83b8c11 100644
--- a/examples/python/rcpsp_sat.py
+++ b/examples/python/rcpsp_sat.py
@@ -18,7 +18,7 @@ import collections
 from absl import app
 from absl import flags
 from google.protobuf import text_format
-from ortools.data import pywraprcpsp
+from ortools.scheduling import pywraprcpsp
 from ortools.sat.python import cp_model
 
 FLAGS = flags.FLAGS
diff --git a/makefiles/Makefile.cpp.mk b/makefiles/Makefile.cpp.mk
index 234c383fe0..61cc709f2e 100644
--- a/makefiles/Makefile.cpp.mk
+++ b/makefiles/Makefile.cpp.mk
@@ -64,9 +64,6 @@ $(GEN_DIR)/ortools/bop: | $(GEN_DIR)/ortools
 $(GEN_DIR)/ortools/constraint_solver: | $(GEN_DIR)/ortools
 	-$(MKDIR) $(GEN_PATH)$Sortools$Sconstraint_solver
 
-$(GEN_DIR)/ortools/data: | $(GEN_DIR)/ortools
-	-$(MKDIR) $(GEN_PATH)$Sortools$Sdata
-
 $(GEN_DIR)/ortools/flatzinc: | $(GEN_DIR)/ortools
 	-$(MKDIR) $(GEN_PATH)$Sortools$Sflatzinc
 
@@ -79,18 +76,24 @@ $(GEN_DIR)/ortools/graph: | $(GEN_DIR)/ortools
 $(GEN_DIR)/ortools/gscip: | $(GEN_DIR)/ortools
 	-$(MKDIR) $(GEN_PATH)$Sortools$Sgscip
 
+$(GEN_DIR)/ortools/init: | $(GEN_DIR)/ortools
+	-$(MKDIR) $(GEN_PATH)$Sortools$Sinit
+
 $(GEN_DIR)/ortools/linear_solver: | $(GEN_DIR)/ortools
 	-$(MKDIR) $(GEN_PATH)$Sortools$Slinear_solver
 
+$(GEN_DIR)/ortools/packing: | $(GEN_DIR)/ortools
+	-$(MKDIR) $(GEN_PATH)$Sortools$Spacking
+
 $(GEN_DIR)/ortools/sat: | $(GEN_DIR)/ortools
 	-$(MKDIR) $(GEN_PATH)$Sortools$Ssat
 
+$(GEN_DIR)/ortools/scheduling: | $(GEN_DIR)/ortools
+	-$(MKDIR) $(GEN_PATH)$Sortools$Sscheduling
+
 $(GEN_DIR)/ortools/util: | $(GEN_DIR)/ortools
 	-$(MKDIR) $(GEN_PATH)$Sortools$Sutil
 
-$(GEN_DIR)/ortools/init: | $(GEN_DIR)/ortools
-	-$(MKDIR) $(GEN_PATH)$Sortools$Sinit
-
 $(GEN_DIR)/examples: | $(GEN_DIR)
 	-$(MKDIR) $(GEN_PATH)$Sexamples
 
@@ -118,9 +121,6 @@ $(OBJ_DIR)/bop: | $(OBJ_DIR)
 $(OBJ_DIR)/constraint_solver: | $(OBJ_DIR)
 	-$(MKDIR_P) $(OBJ_DIR)$Sconstraint_solver
 
-$(OBJ_DIR)/data: | $(OBJ_DIR)
-	-$(MKDIR_P) $(OBJ_DIR)$Sdata
-
 $(OBJ_DIR)/flatzinc: | $(OBJ_DIR)
 	-$(MKDIR_P) $(OBJ_DIR)$Sflatzinc
 
@@ -142,12 +142,18 @@ $(OBJ_DIR)/linear_solver: | $(OBJ_DIR)
 $(OBJ_DIR)/lp_data: | $(OBJ_DIR)
 	-$(MKDIR_P) $(OBJ_DIR)$Slp_data
 
+$(OBJ_DIR)/packing: | $(OBJ_DIR)
+	-$(MKDIR_P) $(OBJ_DIR)$Spacking
+
 $(OBJ_DIR)/port: | $(OBJ_DIR)
 	-$(MKDIR_P) $(OBJ_DIR)$Sport
 
 $(OBJ_DIR)/sat: | $(OBJ_DIR)
 	-$(MKDIR_P) $(OBJ_DIR)$Ssat
 
+$(OBJ_DIR)/scheduling: | $(OBJ_DIR)
+	-$(MKDIR_P) $(OBJ_DIR)$Sscheduling
+
 $(OBJ_DIR)/util: | $(OBJ_DIR)
 	-$(MKDIR_P) $(OBJ_DIR)$Sutil
 
@@ -159,24 +165,24 @@ $(OBJ_DIR)/swig: | $(OBJ_DIR)
 ###############
 # build from: $> grep "pb\.h:" makefiles/Makefile.gen.mk
 PROTO_DEPS = \
-$(GEN_DIR)/ortools/util/optional_boolean.pb.h \
-$(GEN_DIR)/ortools/data/jobshop_scheduling.pb.h \
-$(GEN_DIR)/ortools/data/rcpsp.pb.h \
-$(GEN_DIR)/ortools/data/vector_bin_packing.pb.h \
-$(GEN_DIR)/ortools/glop/parameters.pb.h \
-$(GEN_DIR)/ortools/graph/flow_problem.pb.h \
-$(GEN_DIR)/ortools/sat/boolean_problem.pb.h \
-$(GEN_DIR)/ortools/sat/cp_model.pb.h \
-$(GEN_DIR)/ortools/sat/sat_parameters.pb.h \
 $(GEN_DIR)/ortools/bop/bop_parameters.pb.h \
-$(GEN_DIR)/ortools/linear_solver/linear_solver.pb.h \
 $(GEN_DIR)/ortools/constraint_solver/assignment.pb.h \
 $(GEN_DIR)/ortools/constraint_solver/demon_profiler.pb.h \
 $(GEN_DIR)/ortools/constraint_solver/routing_enums.pb.h \
 $(GEN_DIR)/ortools/constraint_solver/routing_parameters.pb.h \
 $(GEN_DIR)/ortools/constraint_solver/search_limit.pb.h \
 $(GEN_DIR)/ortools/constraint_solver/search_stats.pb.h \
-$(GEN_DIR)/ortools/constraint_solver/solver_parameters.pb.h
+$(GEN_DIR)/ortools/constraint_solver/solver_parameters.pb.h \
+$(GEN_DIR)/ortools/glop/parameters.pb.h \
+$(GEN_DIR)/ortools/graph/flow_problem.pb.h \
+$(GEN_DIR)/ortools/linear_solver/linear_solver.pb.h \
+$(GEN_DIR)/ortools/packing/vector_bin_packing.pb.h \
+$(GEN_DIR)/ortools/sat/boolean_problem.pb.h \
+$(GEN_DIR)/ortools/sat/cp_model.pb.h \
+$(GEN_DIR)/ortools/sat/sat_parameters.pb.h \
+$(GEN_DIR)/ortools/scheduling/jobshop_scheduling.pb.h \
+$(GEN_DIR)/ortools/scheduling/rcpsp.pb.h \
+$(GEN_DIR)/ortools/util/optional_boolean.pb.h
 
 include $(OR_ROOT)makefiles/Makefile.gen.mk
 
@@ -607,7 +613,8 @@ install_ortools_dirs: install_dirs
 	-$(MKDIR) "$(DESTDIR)$(prefix)$Sinclude$Sortools$Sport"
 	-$(MKDIR) "$(DESTDIR)$(prefix)$Sinclude$Sortools$Ssat"
 	-$(MKDIR) "$(DESTDIR)$(prefix)$Sinclude$Sortools$Sutil"
-	-$(MKDIR) "$(DESTDIR)$(prefix)$Sinclude$Sortools$Sdata"
+	-$(MKDIR) "$(DESTDIR)$(prefix)$Sinclude$Sortools$Sscheduling"
+	-$(MKDIR) "$(DESTDIR)$(prefix)$Sinclude$Sortools$Spacking"
 
 .PHONY: install_cc # Install C++ OR-Tools to $(DESTDIR)$(prefix)
 install_cc: install_libortools install_third_party install_doc
@@ -633,8 +640,10 @@ install_libortools: $(OR_TOOLS_LIBS) install_ortools_dirs
 	$(COPY) $(GEN_PATH)$Sortools$Ssat$S*.pb.h "$(DESTDIR)$(prefix)$Sinclude$Sortools$Ssat"
 	$(COPY) ortools$Sutil$S*.h "$(DESTDIR)$(prefix)$Sinclude$Sortools$Sutil"
 	$(COPY) $(GEN_PATH)$Sortools$Sutil$S*.pb.h "$(DESTDIR)$(prefix)$Sinclude$Sortools$Sutil"
-	$(COPY) ortools$Sdata$S*.h "$(DESTDIR)$(prefix)$Sinclude$Sortools$Sdata"
-	$(COPY) $(GEN_PATH)$Sortools$Sdata$S*.pb.h "$(DESTDIR)$(prefix)$Sinclude$Sortools$Sdata"
+	$(COPY) ortools$Spacking$S*.h "$(DESTDIR)$(prefix)$Sinclude$Sortools$Spacking"
+	$(COPY) $(GEN_PATH)$Sortools$Spacking$S*.pb.h "$(DESTDIR)$(prefix)$Sinclude$Sortools$Spacking"
+	$(COPY) ortools$Sscheduling$S*.h "$(DESTDIR)$(prefix)$Sinclude$Sortools$Sscheduling"
+	$(COPY) $(GEN_PATH)$Sortools$Sscheduling$S*.pb.h "$(DESTDIR)$(prefix)$Sinclude$Sortools$Sscheduling"
 	$(COPY) $(LIB_DIR)$S$(LIB_PREFIX)ortools.$L "$(DESTDIR)$(prefix)$Slib"
 
 .PHONY: install_third_party
@@ -782,7 +791,8 @@ clean_cc:
 	-$(DEL) $(OBJ_DIR)$Sbase$S*.$O
 	-$(DEL) $(OBJ_DIR)$Sbop$S*.$O
 	-$(DEL) $(OBJ_DIR)$Sconstraint_solver$S*.$O
-	-$(DEL) $(OBJ_DIR)$Sdata$S*.$O
+	-$(DEL) $(OBJ_DIR)$Spacking$S*.$O
+	-$(DEL) $(OBJ_DIR)$Sscheduling$S*.$O
 	-$(DEL) $(OBJ_DIR)$Sflatzinc$S*.$O
 	-$(DEL) $(OBJ_DIR)$Sglop$S*.$O
 	-$(DEL) $(OBJ_DIR)$Sgraph$S*.$O
@@ -800,7 +810,8 @@ clean_cc:
 	-$(DEL) $(CC_TESTS)
 	-$(DEL) $(GEN_PATH)$Sortools$Sbop$S*.pb.*
 	-$(DEL) $(GEN_PATH)$Sortools$Sconstraint_solver$S*.pb.*
-	-$(DEL) $(GEN_PATH)$Sortools$Sdata$S*.pb.*
+	-$(DEL) $(GEN_PATH)$Sortools$Spacking$S*.pb.*
+	-$(DEL) $(GEN_PATH)$Sortools$Sscheduling$S*.pb.*
 	-$(DEL) $(GEN_PATH)$Sortools$Sflatzinc$S*.tab.*
 	-$(DEL) $(GEN_PATH)$Sortools$Sflatzinc$S*.yy.*
 	-$(DEL) $(GEN_PATH)$Sortools$Sflatzinc$Sparser.*
diff --git a/makefiles/Makefile.gen.mk b/makefiles/Makefile.gen.mk
index 6d98eff025..e0d329bff6 100644
--- a/makefiles/Makefile.gen.mk
+++ b/makefiles/Makefile.gen.mk
@@ -360,108 +360,91 @@ INIT_LIB_OBJS = \
 
 
 DATA_DEPS = \
- $(SRC_DIR)/ortools/data/jobshop_scheduling_parser.h \
- $(SRC_DIR)/ortools/data/rcpsp_parser.h \
- $(SRC_DIR)/ortools/data/set_covering_data.h \
- $(SRC_DIR)/ortools/data/set_covering_parser.h \
- $(SRC_DIR)/ortools/data/vector_bin_packing_parser.h \
- $(GEN_DIR)/ortools/data/jobshop_scheduling.pb.h \
- $(GEN_DIR)/ortools/data/rcpsp.pb.h \
- $(GEN_DIR)/ortools/data/vector_bin_packing.pb.h
+ $(SRC_DIR)/ortools/scheduling/jobshop_scheduling_parser.h \
+ $(SRC_DIR)/ortools/scheduling/rcpsp_parser.h \
+ $(SRC_DIR)/ortools/packing/vector_bin_packing_parser.h \
+ $(GEN_DIR)/ortools/scheduling/jobshop_scheduling.pb.h \
+ $(GEN_DIR)/ortools/scheduling/rcpsp.pb.h \
+ $(GEN_DIR)/ortools/packing/vector_bin_packing.pb.h
 
 DATA_LIB_OBJS = \
- $(OBJ_DIR)/data/jobshop_scheduling_parser.$O \
- $(OBJ_DIR)/data/rcpsp_parser.$O \
- $(OBJ_DIR)/data/set_covering_data.$O \
- $(OBJ_DIR)/data/set_covering_parser.$O \
- $(OBJ_DIR)/data/vector_bin_packing_parser.$O \
- $(OBJ_DIR)/data/jobshop_scheduling.pb.$O \
- $(OBJ_DIR)/data/rcpsp.pb.$O \
- $(OBJ_DIR)/data/vector_bin_packing.pb.$O
+ $(OBJ_DIR)/scheduling/jobshop_scheduling_parser.$O \
+ $(OBJ_DIR)/scheduling/rcpsp_parser.$O \
+ $(OBJ_DIR)/packing/vector_bin_packing_parser.$O \
+ $(OBJ_DIR)/scheduling/jobshop_scheduling.pb.$O \
+ $(OBJ_DIR)/scheduling/rcpsp.pb.$O \
+ $(OBJ_DIR)/packing/vector_bin_packing.pb.$O
 
-objs/data/jobshop_scheduling_parser.$O: \
- ortools/data/jobshop_scheduling_parser.cc \
- ortools/data/jobshop_scheduling_parser.h ortools/base/integral_types.h \
- ortools/gen/ortools/data/jobshop_scheduling.pb.h \
+objs/scheduling/jobshop_scheduling_parser.$O: \
+ ortools/scheduling/jobshop_scheduling_parser.cc \
+ ortools/scheduling/jobshop_scheduling_parser.h ortools/base/integral_types.h \
+ ortools/gen/ortools/scheduling/jobshop_scheduling.pb.h \
  ortools/base/commandlineflags.h ortools/base/filelineiter.h \
  ortools/base/file.h ortools/base/logging.h ortools/base/log_severity.h \
  ortools/base/logging_export.h ortools/base/macros.h \
- ortools/base/vlog_is_on.h | $(OBJ_DIR)/data
-	$(CCC) $(CFLAGS) -c $(SRC_DIR)$Sortools$Sdata$Sjobshop_scheduling_parser.cc $(OBJ_OUT)$(OBJ_DIR)$Sdata$Sjobshop_scheduling_parser.$O
+ ortools/base/vlog_is_on.h | $(OBJ_DIR)/scheduling
+	$(CCC) $(CFLAGS) -c $(SRC_DIR)$Sortools$Sscheduling$Sjobshop_scheduling_parser.cc $(OBJ_OUT)$(OBJ_DIR)$Sscheduling$Sjobshop_scheduling_parser.$O
 
-objs/data/rcpsp_parser.$O: ortools/data/rcpsp_parser.cc \
- ortools/data/rcpsp_parser.h ortools/base/integral_types.h \
- ortools/gen/ortools/data/rcpsp.pb.h ortools/base/filelineiter.h \
+objs/scheduling/rcpsp_parser.$O: ortools/scheduling/rcpsp_parser.cc \
+ ortools/scheduling/rcpsp_parser.h ortools/base/integral_types.h \
+ ortools/gen/ortools/scheduling/rcpsp.pb.h ortools/base/filelineiter.h \
  ortools/base/file.h ortools/base/logging.h \
  ortools/base/commandlineflags.h ortools/base/log_severity.h \
  ortools/base/logging_export.h ortools/base/macros.h \
- ortools/base/vlog_is_on.h | $(OBJ_DIR)/data
-	$(CCC) $(CFLAGS) -c $(SRC_DIR)$Sortools$Sdata$Srcpsp_parser.cc $(OBJ_OUT)$(OBJ_DIR)$Sdata$Srcpsp_parser.$O
+ ortools/base/vlog_is_on.h | $(OBJ_DIR)/scheduling
+	$(CCC) $(CFLAGS) -c $(SRC_DIR)$Sortools$Sscheduling$Srcpsp_parser.cc $(OBJ_OUT)$(OBJ_DIR)$Sscheduling$Srcpsp_parser.$O
 
-objs/data/set_covering_data.$O: ortools/data/set_covering_data.cc \
- ortools/data/set_covering_data.h ortools/base/integral_types.h | $(OBJ_DIR)/data
-	$(CCC) $(CFLAGS) -c $(SRC_DIR)$Sortools$Sdata$Sset_covering_data.cc $(OBJ_OUT)$(OBJ_DIR)$Sdata$Sset_covering_data.$O
-
-objs/data/set_covering_parser.$O: ortools/data/set_covering_parser.cc \
- ortools/data/set_covering_parser.h ortools/base/integral_types.h \
- ortools/data/set_covering_data.h ortools/base/filelineiter.h \
- ortools/base/file.h ortools/base/logging.h \
- ortools/base/commandlineflags.h ortools/base/log_severity.h \
- ortools/base/logging_export.h ortools/base/macros.h \
- ortools/base/vlog_is_on.h | $(OBJ_DIR)/data
-	$(CCC) $(CFLAGS) -c $(SRC_DIR)$Sortools$Sdata$Sset_covering_parser.cc $(OBJ_OUT)$(OBJ_DIR)$Sdata$Sset_covering_parser.$O
-
-objs/data/vector_bin_packing_parser.$O: \
- ortools/data/vector_bin_packing_parser.cc \
- ortools/data/vector_bin_packing_parser.h ortools/base/integral_types.h \
- ortools/gen/ortools/data/vector_bin_packing.pb.h \
+objs/packing/vector_bin_packing_parser.$O: \
+ ortools/packing/vector_bin_packing_parser.cc \
+ ortools/packing/vector_bin_packing_parser.h ortools/base/integral_types.h \
+ ortools/gen/ortools/packing/vector_bin_packing.pb.h \
  ortools/base/filelineiter.h ortools/base/file.h ortools/base/logging.h \
  ortools/base/commandlineflags.h ortools/base/log_severity.h \
  ortools/base/logging_export.h ortools/base/macros.h \
- ortools/base/vlog_is_on.h | $(OBJ_DIR)/data
-	$(CCC) $(CFLAGS) -c $(SRC_DIR)$Sortools$Sdata$Svector_bin_packing_parser.cc $(OBJ_OUT)$(OBJ_DIR)$Sdata$Svector_bin_packing_parser.$O
+ ortools/base/vlog_is_on.h | $(OBJ_DIR)/packing
+	$(CCC) $(CFLAGS) -c $(SRC_DIR)$Sortools$Spacking$Svector_bin_packing_parser.cc $(OBJ_OUT)$(OBJ_DIR)$Spacking$Svector_bin_packing_parser.$O
 
-ortools/data/jobshop_scheduling.proto: ;
+ortools/scheduling/jobshop_scheduling.proto: ;
 
-$(GEN_DIR)/ortools/data/jobshop_scheduling.pb.cc: \
- $(SRC_DIR)/ortools/data/jobshop_scheduling.proto | $(GEN_DIR)/ortools/data
-	$(PROTOC) --experimental_allow_proto3_optional --proto_path=$(INC_DIR) $(PROTOBUF_PROTOC_INC) --cpp_out=$(GEN_PATH) $(SRC_DIR)/ortools/data/jobshop_scheduling.proto
+$(GEN_DIR)/ortools/scheduling/jobshop_scheduling.pb.cc: \
+ $(SRC_DIR)/ortools/scheduling/jobshop_scheduling.proto | $(GEN_DIR)/ortools/scheduling
+	$(PROTOC) --experimental_allow_proto3_optional --proto_path=$(INC_DIR) $(PROTOBUF_PROTOC_INC) --cpp_out=$(GEN_PATH) $(SRC_DIR)/ortools/scheduling/jobshop_scheduling.proto
 
-$(GEN_DIR)/ortools/data/jobshop_scheduling.pb.h: \
- $(GEN_DIR)/ortools/data/jobshop_scheduling.pb.cc
-	$(TOUCH) $(GEN_PATH)$Sortools$Sdata$Sjobshop_scheduling.pb.h
+$(GEN_DIR)/ortools/scheduling/jobshop_scheduling.pb.h: \
+ $(GEN_DIR)/ortools/scheduling/jobshop_scheduling.pb.cc
+	$(TOUCH) $(GEN_PATH)$Sortools$Sscheduling$Sjobshop_scheduling.pb.h
 
-$(OBJ_DIR)/data/jobshop_scheduling.pb.$O: \
- $(GEN_DIR)/ortools/data/jobshop_scheduling.pb.cc | $(OBJ_DIR)/data
-	$(CCC) $(CFLAGS) -c $(GEN_PATH)$Sortools$Sdata$Sjobshop_scheduling.pb.cc $(OBJ_OUT)$(OBJ_DIR)$Sdata$Sjobshop_scheduling.pb.$O
+$(OBJ_DIR)/scheduling/jobshop_scheduling.pb.$O: \
+ $(GEN_DIR)/ortools/scheduling/jobshop_scheduling.pb.cc | $(OBJ_DIR)/scheduling
+	$(CCC) $(CFLAGS) -c $(GEN_PATH)$Sortools$Sscheduling$Sjobshop_scheduling.pb.cc $(OBJ_OUT)$(OBJ_DIR)$Sscheduling$Sjobshop_scheduling.pb.$O
 
-ortools/data/rcpsp.proto: ;
+ortools/scheduling/rcpsp.proto: ;
 
-$(GEN_DIR)/ortools/data/rcpsp.pb.cc: \
- $(SRC_DIR)/ortools/data/rcpsp.proto | $(GEN_DIR)/ortools/data
-	$(PROTOC) --experimental_allow_proto3_optional --proto_path=$(INC_DIR) $(PROTOBUF_PROTOC_INC) --cpp_out=$(GEN_PATH) $(SRC_DIR)/ortools/data/rcpsp.proto
+$(GEN_DIR)/ortools/scheduling/rcpsp.pb.cc: \
+ $(SRC_DIR)/ortools/scheduling/rcpsp.proto | $(GEN_DIR)/ortools/scheduling
+	$(PROTOC) --experimental_allow_proto3_optional --proto_path=$(INC_DIR) $(PROTOBUF_PROTOC_INC) --cpp_out=$(GEN_PATH) $(SRC_DIR)/ortools/scheduling/rcpsp.proto
 
-$(GEN_DIR)/ortools/data/rcpsp.pb.h: \
- $(GEN_DIR)/ortools/data/rcpsp.pb.cc
-	$(TOUCH) $(GEN_PATH)$Sortools$Sdata$Srcpsp.pb.h
+$(GEN_DIR)/ortools/scheduling/rcpsp.pb.h: \
+ $(GEN_DIR)/ortools/scheduling/rcpsp.pb.cc
+	$(TOUCH) $(GEN_PATH)$Sortools$Sscheduling$Srcpsp.pb.h
 
-$(OBJ_DIR)/data/rcpsp.pb.$O: \
- $(GEN_DIR)/ortools/data/rcpsp.pb.cc | $(OBJ_DIR)/data
-	$(CCC) $(CFLAGS) -c $(GEN_PATH)$Sortools$Sdata$Srcpsp.pb.cc $(OBJ_OUT)$(OBJ_DIR)$Sdata$Srcpsp.pb.$O
+$(OBJ_DIR)/scheduling/rcpsp.pb.$O: \
+ $(GEN_DIR)/ortools/scheduling/rcpsp.pb.cc | $(OBJ_DIR)/scheduling
+	$(CCC) $(CFLAGS) -c $(GEN_PATH)$Sortools$Sscheduling$Srcpsp.pb.cc $(OBJ_OUT)$(OBJ_DIR)$Sscheduling$Srcpsp.pb.$O
 
-ortools/data/vector_bin_packing.proto: ;
+ortools/packing/vector_bin_packing.proto: ;
 
-$(GEN_DIR)/ortools/data/vector_bin_packing.pb.cc: \
- $(SRC_DIR)/ortools/data/vector_bin_packing.proto | $(GEN_DIR)/ortools/data
-	$(PROTOC) --experimental_allow_proto3_optional --proto_path=$(INC_DIR) $(PROTOBUF_PROTOC_INC) --cpp_out=$(GEN_PATH) $(SRC_DIR)/ortools/data/vector_bin_packing.proto
+$(GEN_DIR)/ortools/packing/vector_bin_packing.pb.cc: \
+ $(SRC_DIR)/ortools/packing/vector_bin_packing.proto | $(GEN_DIR)/ortools/packing
+	$(PROTOC) --experimental_allow_proto3_optional --proto_path=$(INC_DIR) $(PROTOBUF_PROTOC_INC) --cpp_out=$(GEN_PATH) $(SRC_DIR)/ortools/packing/vector_bin_packing.proto
 
-$(GEN_DIR)/ortools/data/vector_bin_packing.pb.h: \
- $(GEN_DIR)/ortools/data/vector_bin_packing.pb.cc
-	$(TOUCH) $(GEN_PATH)$Sortools$Sdata$Svector_bin_packing.pb.h
+$(GEN_DIR)/ortools/packing/vector_bin_packing.pb.h: \
+ $(GEN_DIR)/ortools/packing/vector_bin_packing.pb.cc
+	$(TOUCH) $(GEN_PATH)$Sortools$Spacking$Svector_bin_packing.pb.h
 
-$(OBJ_DIR)/data/vector_bin_packing.pb.$O: \
- $(GEN_DIR)/ortools/data/vector_bin_packing.pb.cc | $(OBJ_DIR)/data
-	$(CCC) $(CFLAGS) -c $(GEN_PATH)$Sortools$Sdata$Svector_bin_packing.pb.cc $(OBJ_OUT)$(OBJ_DIR)$Sdata$Svector_bin_packing.pb.$O
+$(OBJ_DIR)/packing/vector_bin_packing.pb.$O: \
+ $(GEN_DIR)/ortools/packing/vector_bin_packing.pb.cc | $(OBJ_DIR)/packing
+	$(CCC) $(CFLAGS) -c $(GEN_PATH)$Sortools$Spacking$Svector_bin_packing.pb.cc $(OBJ_OUT)$(OBJ_DIR)$Spacking$Svector_bin_packing.pb.$O
 
 LP_DATA_DEPS = \
  $(SRC_DIR)/ortools/lp_data/lp_data.h \
@@ -5206,4 +5189,3 @@ $(GEN_DIR)/ortools/constraint_solver/solver_parameters.pb.h: \
 $(OBJ_DIR)/constraint_solver/solver_parameters.pb.$O: \
  $(GEN_DIR)/ortools/constraint_solver/solver_parameters.pb.cc | $(OBJ_DIR)/constraint_solver
 	$(CCC) $(CFLAGS) -c $(GEN_PATH)$Sortools$Sconstraint_solver$Ssolver_parameters.pb.cc $(OBJ_OUT)$(OBJ_DIR)$Sconstraint_solver$Ssolver_parameters.pb.$O
-
diff --git a/makefiles/Makefile.python.mk b/makefiles/Makefile.python.mk
index 48e9fd6b3c..a8f63503b6 100644
--- a/makefiles/Makefile.python.mk
+++ b/makefiles/Makefile.python.mk
@@ -47,7 +47,7 @@ PYGRAPH_LIBS = $(LIB_DIR)/_pywrapgraph.$(SWIG_PYTHON_LIB_SUFFIX)
 PYCP_LIBS = $(LIB_DIR)/_pywrapcp.$(SWIG_PYTHON_LIB_SUFFIX)
 PYLP_LIBS = $(LIB_DIR)/_pywraplp.$(SWIG_PYTHON_LIB_SUFFIX)
 PYSAT_LIBS = $(LIB_DIR)/_pywrapsat.$(SWIG_PYTHON_LIB_SUFFIX)
-PYDATA_LIBS = $(LIB_DIR)/_pywraprcpsp.$(SWIG_PYTHON_LIB_SUFFIX)
+PYSCHEDULING_LIBS = $(LIB_DIR)/_pywraprcpsp.$(SWIG_PYTHON_LIB_SUFFIX)
 PYSORTED_INTERVAL_LIST_LIBS = $(LIB_DIR)/_sorted_interval_list.$(SWIG_PYTHON_LIB_SUFFIX)
 PYTHON_OR_TOOLS_LIBS = \
  $(GEN_DIR)/ortools/__init__.py \
@@ -57,7 +57,7 @@ PYTHON_OR_TOOLS_LIBS = \
  $(PYCP_LIBS) \
  $(PYLP_LIBS) \
  $(PYSAT_LIBS) \
- $(PYDATA_LIBS) \
+ $(PYSCHEDULING_LIBS) \
  $(PYSORTED_INTERVAL_LIST_LIBS)
 
 # Main target
@@ -455,38 +455,38 @@ ifeq ($(PLATFORM),MACOSX)
 PYRCPSP_LDFLAGS = -install_name @rpath/_pywraprcpsp.$(SWIG_PYTHON_LIB_SUFFIX) #
 endif
 
-$(GEN_DIR)/ortools/data/rcpsp_pb2.py: \
- $(SRC_DIR)/ortools/data/rcpsp.proto \
+$(GEN_DIR)/ortools/scheduling/rcpsp_pb2.py: \
+ $(SRC_DIR)/ortools/scheduling/rcpsp.proto \
  $(PROTOBUF_PYTHON_DESC) \
- | $(GEN_DIR)/ortools/data
+ | $(GEN_DIR)/ortools/scheduling
 	$(PROTOC) --proto_path=$(INC_DIR) --python_out=$(GEN_PATH) $(MYPY_OUT) \
- $(SRC_DIR)/ortools/data/rcpsp.proto
+ $(SRC_DIR)/ortools/scheduling/rcpsp.proto
 
-$(GEN_DIR)/ortools/data/pywraprcpsp.py: \
- $(SRC_DIR)/ortools/data/rcpsp_parser.h \
+$(GEN_DIR)/ortools/scheduling/pywraprcpsp.py: \
+ $(SRC_DIR)/ortools/scheduling/rcpsp_parser.h \
  $(SRC_DIR)/ortools/base/base.i \
- $(SRC_DIR)/ortools/data/python/rcpsp.i \
- $(GEN_DIR)/ortools/data/rcpsp_pb2.py \
+ $(SRC_DIR)/ortools/scheduling/python/rcpsp.i \
+ $(GEN_DIR)/ortools/scheduling/rcpsp_pb2.py \
  $(DATA_DEPS) \
  $(PROTOBUF_PYTHON_DESC) \
- | $(GEN_DIR)/ortools/data
+ | $(GEN_DIR)/ortools/scheduling
 	$(SWIG_BINARY) $(SWIG_INC) -I$(INC_DIR) -c++ -python $(SWIG_PYTHON3_FLAG) \
- -o $(GEN_PATH)$Sortools$Sdata$Srcpsp_python_wrap.cc \
+ -o $(GEN_PATH)$Sortools$Sscheduling$Srcpsp_python_wrap.cc \
  -module pywraprcpsp \
- $(SRC_DIR)/ortools/data$Spython$Srcpsp.i
+ $(SRC_DIR)/ortools/scheduling$Spython$Srcpsp.i
 
-$(GEN_DIR)/ortools/data/rcpsp_python_wrap.cc: \
- $(GEN_DIR)/ortools/data/pywraprcpsp.py
+$(GEN_DIR)/ortools/scheduling/rcpsp_python_wrap.cc: \
+ $(GEN_DIR)/ortools/scheduling/pywraprcpsp.py
 
 $(OBJ_DIR)/swig/rcpsp_python_wrap.$O: \
- $(GEN_DIR)/ortools/data/rcpsp_python_wrap.cc \
+ $(GEN_DIR)/ortools/scheduling/rcpsp_python_wrap.cc \
  $(DATA_DEPS) \
  | $(OBJ_DIR)/swig
 	$(CCC) $(CFLAGS) $(PYTHON_INC) $(PYTHON3_CFLAGS) \
- -c $(GEN_PATH)$Sortools$Sdata$Srcpsp_python_wrap.cc \
+ -c $(GEN_PATH)$Sortools$Sscheduling$Srcpsp_python_wrap.cc \
  $(OBJ_OUT)$(OBJ_DIR)$Sswig$Srcpsp_python_wrap.$O
 
-$(PYDATA_LIBS): $(OBJ_DIR)/swig/rcpsp_python_wrap.$O $(OR_TOOLS_LIBS)
+$(PYSCHEDULING_LIBS): $(OBJ_DIR)/swig/rcpsp_python_wrap.$O $(OR_TOOLS_LIBS)
 	$(DYNAMIC_LD) \
  $(PYRCPSP_LDFLAGS) \
  $(LD_OUT)$(LIB_DIR)$S_pywraprcpsp.$(SWIG_PYTHON_LIB_SUFFIX) \
@@ -496,9 +496,9 @@ $(PYDATA_LIBS): $(OBJ_DIR)/swig/rcpsp_python_wrap.$O $(OR_TOOLS_LIBS)
  $(PYTHON_LNK) \
  $(PYTHON_LDFLAGS)
 ifeq ($(SYSTEM),win)
-	copy $(LIB_DIR)$S_pywraprcpsp.$(SWIG_PYTHON_LIB_SUFFIX) $(GEN_PATH)\\ortools\\data\\_pywraprcpsp.pyd
+	copy $(LIB_DIR)$S_pywraprcpsp.$(SWIG_PYTHON_LIB_SUFFIX) $(GEN_PATH)\\ortools\\scheduling\\_pywraprcpsp.pyd
 else
-	cp $(PYDATA_LIBS) $(GEN_PATH)/ortools/data
+	cp $(PYSCHEDULING_LIBS) $(GEN_PATH)/ortools/scheduling
 endif
 
 # sorted_interval_list
@@ -890,7 +890,7 @@ MISSING_PYPI_FILES = \
  $(PYPI_ARCHIVE_TEMP_DIR)/ortools/ortools/constraint_solver \
  $(PYPI_ARCHIVE_TEMP_DIR)/ortools/ortools/linear_solver \
  $(PYPI_ARCHIVE_TEMP_DIR)/ortools/ortools/sat \
- $(PYPI_ARCHIVE_TEMP_DIR)/ortools/ortools/data \
+ $(PYPI_ARCHIVE_TEMP_DIR)/ortools/ortools/scheduling \
  $(PYPI_ARCHIVE_TEMP_DIR)/ortools/ortools/util \
  $(PYPI_ARCHIVE_LIBS)
 
@@ -991,12 +991,12 @@ $(PYPI_ARCHIVE_TEMP_DIR)/ortools/ortools/sat: $(PYSAT_LIBS) | $(PYPI_ARCHIVE_TEM
 	-$(MKDIR) $(PYPI_ARCHIVE_TEMP_DIR)$Sortools$Sortools$Ssat$Spython
 	$(COPY) ortools$Ssat$Spython$S*.py $(PYPI_ARCHIVE_TEMP_DIR)$Sortools$Sortools$Ssat$Spython
 
-$(PYPI_ARCHIVE_TEMP_DIR)/ortools/ortools/data: $(PYDATA_LIBS) | $(PYPI_ARCHIVE_TEMP_DIR)/ortools/ortools
-	-$(DELREC) $(PYPI_ARCHIVE_TEMP_DIR)$Sortools$Sortools$Sdata
-	-$(MKDIR) $(PYPI_ARCHIVE_TEMP_DIR)$Sortools$Sortools$Sdata
-	$(TOUCH) $(PYPI_ARCHIVE_TEMP_DIR)$Sortools$Sortools$Sdata$S__init__.py
-	$(COPY) $(GEN_PATH)$Sortools$Sdata$S*.py* $(PYPI_ARCHIVE_TEMP_DIR)$Sortools$Sortools$Sdata
-	$(COPY) $(GEN_PATH)$Sortools$Sdata$S_pywraprcpsp.* $(PYPI_ARCHIVE_TEMP_DIR)$Sortools$Sortools$Sdata
+$(PYPI_ARCHIVE_TEMP_DIR)/ortools/ortools/scheduling: $(PYSCHEDULING_LIBS) | $(PYPI_ARCHIVE_TEMP_DIR)/ortools/ortools
+	-$(DELREC) $(PYPI_ARCHIVE_TEMP_DIR)$Sortools$Sortools$Sscheduling
+	-$(MKDIR) $(PYPI_ARCHIVE_TEMP_DIR)$Sortools$Sortools$Sscheduling
+	$(TOUCH) $(PYPI_ARCHIVE_TEMP_DIR)$Sortools$Sortools$Sscheduling$S__init__.py
+	$(COPY) $(GEN_PATH)$Sortools$Sscheduling$S*.py* $(PYPI_ARCHIVE_TEMP_DIR)$Sortools$Sortools$Sscheduling
+	$(COPY) $(GEN_PATH)$Sortools$Sscheduling$S_pywraprcpsp.* $(PYPI_ARCHIVE_TEMP_DIR)$Sortools$Sortools$Sscheduling
 
 $(PYPI_ARCHIVE_TEMP_DIR)/ortools/ortools/util: $(PYSORTED_INTERVAL_LIST_LIBS) | $(PYPI_ARCHIVE_TEMP_DIR)/ortools/ortools
 	-$(DELREC) $(PYPI_ARCHIVE_TEMP_DIR)$Sortools$Sortools$Sutil
@@ -1190,13 +1190,13 @@ clean_python:
 	-$(DELREC) ortools$Ssat$Spython$S__pycache__
 	-$(DEL) $(GEN_PATH)$Sortools$Ssat$S*_python_wrap.*
 	-$(DEL) $(GEN_PATH)$Sortools$Ssat$S_pywrap*
-	-$(DEL) $(GEN_PATH)$Sortools$Sdata$S*.py
-	-$(DEL) $(GEN_PATH)$Sortools$Sdata$S*.pyc
-	-$(DELREC) $(GEN_PATH)$Sortools$Sdata$S__pycache__
-	-$(DEL) ortools$Sdata$S*.pyc
-	-$(DELREC) ortools$Sdata$S__pycache__
-	-$(DEL) $(GEN_PATH)$Sortools$Sdata$S*_python_wrap.*
-	-$(DEL) $(GEN_PATH)$Sortools$Sdata$S_pywrap*
+	-$(DEL) $(GEN_PATH)$Sortools$Sscheduling$S*.py
+	-$(DEL) $(GEN_PATH)$Sortools$Sscheduling$S*.pyc
+	-$(DELREC) $(GEN_PATH)$Sortools$Sscheduling$S__pycache__
+	-$(DEL) ortools$Sscheduling$S*.pyc
+	-$(DELREC) ortools$Sscheduling$S__pycache__
+	-$(DEL) $(GEN_PATH)$Sortools$Sscheduling$S*_python_wrap.*
+	-$(DEL) $(GEN_PATH)$Sortools$Sscheduling$S_pywrap*
 	-$(DEL) $(GEN_PATH)$Sortools$Sutil$S*.py
 	-$(DEL) $(GEN_PATH)$Sortools$Sutil$S*.pyc
 	-$(DELREC) $(GEN_PATH)$Sortools$Sutil$S__pycache__
diff --git a/ortools/data/set_covering_data.h b/ortools/data/set_covering_data.h
deleted file mode 100644
index 4756f9e8fb..0000000000
--- a/ortools/data/set_covering_data.h
+++ /dev/null
@@ -1,59 +0,0 @@
-// Copyright 2010-2021 Google LLC
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-
-#ifndef OR_TOOLS_DATA_SET_COVERING_DATA_H_
-#define OR_TOOLS_DATA_SET_COVERING_DATA_H_
-
-#include <vector>
-
-#include "ortools/base/integral_types.h"
-
-namespace operations_research {
-namespace scp {
-
-class ScpData {
- public:
-  ScpData() : is_set_partitioning_(false) {}
-  // Getters.
-  int num_rows() const { return columns_per_row_.size(); }
-  int num_columns() const { return rows_per_column_.size(); }
-  // columns_per_row[i][j] returns the index of the jth column covering row i.
-  const std::vector<std::vector<int>>& columns_per_row() const {
-    return columns_per_row_;
-  }
-  // rows_per_column[i][j] returns the index of the jth row covering column i.
-  const std::vector<std::vector<int>>& rows_per_column() const {
-    return rows_per_column_;
-  }
-  const std::vector<int>& column_costs() const { return column_costs_; }
-
-  bool is_set_partitioning() const { return is_set_partitioning_; }
-  void set_is_set_partitioning(bool v) { is_set_partitioning_ = v; }
-
-  // Builders.
-  // Calling SetProblemSize() will clear all previous data.
-  void SetProblemSize(int num_rows, int num_columns);
-  void SetColumnCost(int column_id, int cost);
-  void AddRowInColumn(int row, int column);
-
- private:
-  std::vector<std::vector<int>> columns_per_row_;
-  std::vector<std::vector<int>> rows_per_column_;
-  std::vector<int> column_costs_;
-  bool is_set_partitioning_;
-};
-
-}  // namespace scp
-}  // namespace operations_research
-
-#endif  // OR_TOOLS_DATA_SET_COVERING_DATA_H_
diff --git a/ortools/data/set_covering_parser.cc b/ortools/data/set_covering_parser.cc
deleted file mode 100644
index 047d6ca0c5..0000000000
--- a/ortools/data/set_covering_parser.cc
+++ /dev/null
@@ -1,207 +0,0 @@
-// Copyright 2010-2021 Google LLC
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-
-#include "ortools/data/set_covering_parser.h"
-
-#include <cstdint>
-
-#include "absl/strings/numbers.h"
-#include "absl/strings/str_split.h"
-#include "ortools/base/filelineiter.h"
-
-namespace operations_research {
-namespace scp {
-
-ScpParser::ScpParser() : section_(INIT), line_(0), remaining_(0), current_(0) {}
-
-bool ScpParser::LoadProblem(const std::string& filename, Format format,
-                            ScpData* data) {
-  section_ = INIT;
-  line_ = 0;
-  remaining_ = 0;
-  current_ = 0;
-
-  for (const std::string& line : FileLines(filename)) {
-    ProcessLine(line, format, data);
-    if (section_ == ERROR) return false;
-  }
-  return section_ == END;
-}
-
-void ScpParser::ProcessLine(const std::string& line, Format format,
-                            ScpData* data) {
-  line_++;
-  const std::vector<std::string> words =
-      absl::StrSplit(line, absl::ByAnyChar(" :\t\r"), absl::SkipEmpty());
-  switch (section_) {
-    case INIT: {
-      if (words.size() != 2) {
-        LogError(line, "Problem reading the size of the problem");
-        return;
-      }
-      const int num_rows = strtoint32(words[0]);
-      const int num_columns = strtoint32(words[1]);
-      data->SetProblemSize(num_rows, num_columns);
-      current_ = 0;
-      switch (format) {
-        case SCP_FORMAT: {
-          section_ = COSTS;
-          break;
-        }
-        case RAILROAD_FORMAT: {
-          section_ = COLUMN;
-          break;
-        }
-        case TRIPLET_FORMAT: {
-          section_ = COLUMN;
-          break;
-        }
-        case SPP_FORMAT: {
-          section_ = COLUMN;
-          data->set_is_set_partitioning(true);
-          break;
-        }
-      }
-      break;
-    }
-    case COSTS: {
-      const int num_items = words.size();
-      if (current_ + num_items > data->num_columns()) {
-        LogError(line, "Too many cost items");
-        return;
-      }
-      for (int i = 0; i < num_items; ++i) {
-        data->SetColumnCost(current_++, strtoint32(words[i]));
-      }
-      if (current_ == data->num_columns()) {
-        section_ = NUM_COLUMNS_IN_ROW;
-        current_ = 0;
-      }
-      break;
-    }
-    case COLUMN: {
-      switch (format) {
-        case SCP_FORMAT: {
-          LogError(line, "Wrong state in the loader");
-          return;
-        }
-        case RAILROAD_FORMAT:
-          ABSL_FALLTHROUGH_INTENDED;
-        case SPP_FORMAT: {
-          if (words.size() < 2) {
-            LogError(line, "Column declaration too short");
-            return;
-          }
-          const int cost = strtoint32(words[0]);
-          data->SetColumnCost(current_, cost);
-          const int num_items = strtoint32(words[1]);
-          if (words.size() != 2 + num_items) {
-            LogError(line, "Mistatch in column declaration");
-            return;
-          }
-          for (int i = 0; i < num_items; ++i) {
-            const int row = strtoint32(words[i + 2]) - 1;  // 1 based.
-            data->AddRowInColumn(row, current_);
-          }
-          current_++;
-          if (current_ == data->num_columns()) {
-            section_ = format == RAILROAD_FORMAT ? END : NUM_NON_ZEROS;
-          }
-          break;
-        }
-        case TRIPLET_FORMAT: {
-          if (words.size() != 3) {
-            LogError(line, "Column declaration does not contain 3 rows");
-            break;
-          }
-          data->SetColumnCost(current_, 1);
-          for (int i = 0; i < 3; ++i) {
-            const int row = strtoint32(words[i]) - 1;  // 1 based.
-            data->AddRowInColumn(row, current_);
-          }
-          current_++;
-          if (current_ == data->num_columns()) {
-            section_ = END;
-          }
-          break;
-        }
-      }
-      break;
-    }
-    case NUM_COLUMNS_IN_ROW: {
-      if (words.size() != 1) {
-        LogError(line, "The header of a column should be one number");
-        return;
-      }
-      remaining_ = strtoint32(words[0]);
-      section_ = ROW;
-      break;
-    }
-    case ROW: {
-      const int num_items = words.size();
-      if (num_items > remaining_) {
-        LogError(line, "Too many columns in a row declaration");
-        return;
-      }
-      for (const std::string& w : words) {
-        remaining_--;
-        const int column = strtoint32(w) - 1;  // 1 based.
-        data->AddRowInColumn(current_, column);
-      }
-      if (remaining_ == 0) {
-        current_++;
-        if (current_ == data->num_rows()) {
-          section_ = END;
-        } else {
-          section_ = NUM_COLUMNS_IN_ROW;
-        }
-      }
-      break;
-    }
-    case NUM_NON_ZEROS: {
-      if (words.size() != 1) {
-        LogError(line, "The header of a column should be one number");
-        return;
-      }
-      section_ = END;
-      break;
-    }
-    case END: {
-      break;
-    }
-    case ERROR: {
-      break;
-    }
-  }
-}
-
-void ScpParser::LogError(const std::string& line, const std::string& message) {
-  LOG(ERROR) << "Error on line " << line_ << ": " << message << "(" << line
-             << ")";
-  section_ = ERROR;
-}
-
-int ScpParser::strtoint32(const std::string& word) {
-  int result;
-  CHECK(absl::SimpleAtoi(word, &result));
-  return result;
-}
-
-int64_t ScpParser::strtoint64(const std::string& word) {
-  int64_t result;
-  CHECK(absl::SimpleAtoi(word, &result));
-  return result;
-}
-
-}  // namespace scp
-}  // namespace operations_research
diff --git a/ortools/data/set_covering_parser.h b/ortools/data/set_covering_parser.h
deleted file mode 100644
index 7501b4d7d7..0000000000
--- a/ortools/data/set_covering_parser.h
+++ /dev/null
@@ -1,113 +0,0 @@
-// Copyright 2010-2021 Google LLC
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-
-#ifndef OR_TOOLS_DATA_SET_COVERING_PARSER_H_
-#define OR_TOOLS_DATA_SET_COVERING_PARSER_H_
-
-#include <cstdint>
-#include <string>
-#include <vector>
-
-#include "ortools/base/integral_types.h"
-#include "ortools/data/set_covering_data.h"
-
-namespace operations_research {
-namespace scp {
-
-// Set covering problem.
-//
-// We have a list of subsets of a set. Each subset has a cost.  The
-// goal is to select of solution set of subsets such that (1) all elements
-// of the set belongs to at least one subset of the solution set, and (2)
-// the sum of the cost of each subset in the solution set is minimal.
-//
-// To follow the standard literature, each element is called a row, and each
-// subset is called a column.
-
-class ScpParser {
- public:
-  enum Section {
-    INIT,
-    COSTS,
-    COLUMN,
-    NUM_COLUMNS_IN_ROW,
-    ROW,
-    NUM_NON_ZEROS,
-    END,
-    ERROR,
-  };
-
-  enum Format {
-    // The original scp format of these problem is:
-    //
-    // number of rows (m), number of columns (n)
-    //
-    // the cost of each column c(j),j=1,...,n
-    //
-    // for each row i (i=1,...,m): the number of columns which cover row
-    // i followed by a list of the columns which cover row i.
-    //
-    // The original problems (scp*) from the OR-LIB follow this format.
-    SCP_FORMAT,
-    // The railroad format is:
-    //   number of rows (m), number of columns (n)
-    //
-    //   for each column j (j=1,...,n): the cost of the column, the number
-    //   of rows that it covers followed by a list of the rows that it
-    //   covers.
-    //
-    // The railroad problems follow this format.
-    RAILROAD_FORMAT,
-    // The triplet format is:
-    //
-    // number of rows (m), number of columns (n)
-    //
-    // for each column, the 3 rows it contains.  Note that the cost of
-    // each column is 1.
-    //
-    // The Steiner triple covering problems follow this format.
-    TRIPLET_FORMAT,
-    // The spp format is:
-    //   number of rows (m), number of columns (n)
-    //
-    //   for each column j (j=1,...,n): the cost of the column, the number
-    //   of rows that it covers followed by a list of the rows that it
-    //   covers.
-    //
-    //   number of non_zeros
-    //
-    // The set partitioning problems follow this format.
-    SPP_FORMAT
-  };
-
-  ScpParser();
-
-  // This will clear the data before importing the file.
-  bool LoadProblem(const std::string& filename, Format format, ScpData* data);
-
- private:
-  void ProcessLine(const std::string& line, Format format, ScpData* data);
-  void LogError(const std::string& line, const std::string& error_message);
-  int strtoint32(const std::string& word);
-  int64_t strtoint64(const std::string& word);
-
-  Section section_;
-  int line_;
-  int remaining_;
-  int current_;
-};
-
-}  // namespace scp
-}  // namespace operations_research
-
-#endif  // OR_TOOLS_DATA_SET_COVERING_PARSER_H_
diff --git a/ortools/packing/BUILD b/ortools/packing/BUILD
new file mode 100644
index 0000000000..9e40fb978c
--- /dev/null
+++ b/ortools/packing/BUILD
@@ -0,0 +1,3 @@
+load("@rules_cc//cc:defs.bzl", "cc_proto_library")
+
+package(default_visibility = ["//visibility:public"])
diff --git a/ortools/packing/CMakeLists.txt b/ortools/packing/CMakeLists.txt
new file mode 100644
index 0000000000..3636dd06a1
--- /dev/null
+++ b/ortools/packing/CMakeLists.txt
@@ -0,0 +1,21 @@
+file(GLOB _SRCS "*.h" "*.cc")
+set(NAME ${PROJECT_NAME}_packing)
+
+# Will be merge in libortools.so
+#add_library(${NAME} STATIC ${_SRCS})
+add_library(${NAME} OBJECT ${_SRCS})
+set_target_properties(${NAME} PROPERTIES
+  CXX_STANDARD 17
+  CXX_STANDARD_REQUIRED ON
+  CXX_EXTENSIONS OFF
+  POSITION_INDEPENDENT_CODE ON
+  )
+target_include_directories(${NAME} PRIVATE
+  ${PROJECT_SOURCE_DIR}
+  ${PROJECT_BINARY_DIR})
+target_link_libraries(${NAME} PRIVATE
+  absl::flags
+  absl::strings
+  protobuf::libprotobuf
+  ${PROJECT_NAME}::proto)
+#add_library(${PROJECT_NAME}::packing ALIAS ${NAME})
diff --git a/ortools/data/__init__.py b/ortools/packing/__init__.py
similarity index 97%
rename from ortools/data/__init__.py
rename to ortools/packing/__init__.py
index 695694b10c..cb63857e22 100644
--- a/ortools/data/__init__.py
+++ b/ortools/packing/__init__.py
@@ -11,5 +11,5 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import os as _os
-__path__.append(_os.path.join(__path__[0],  '..', 'gen', 'ortools', 'data'))
+__path__.append(_os.path.join(__path__[0],  '..', 'gen', 'ortools', 'packing'))
 __path__.append(_os.path.join(__path__[0],  '..', '..', 'lib'))
diff --git a/ortools/packing/arc_flow_builder.cc b/ortools/packing/arc_flow_builder.cc
new file mode 100644
index 0000000000..3c79050ef9
--- /dev/null
+++ b/ortools/packing/arc_flow_builder.cc
@@ -0,0 +1,408 @@
+// Copyright 2010-2021 Google LLC
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include "ortools/packing/arc_flow_builder.h"
+
+#include <algorithm>
+#include <cstdint>
+
+#include "absl/container/flat_hash_map.h"
+#include "absl/strings/str_cat.h"
+#include "absl/strings/str_join.h"
+#include "ortools/base/commandlineflags.h"
+#include "ortools/base/map_util.h"
+#include "ortools/base/stl_util.h"
+#include "ortools/graph/topologicalsorter.h"
+
+namespace operations_research {
+namespace packing {
+namespace {
+
+class ArcFlowBuilder {
+ public:
+  // Same arguments as BuildArcFlowGraph(): see the .h.
+  ArcFlowBuilder(const std::vector<int>& bin_dimensions,
+                 const std::vector<std::vector<int>>& item_dimensions_by_type,
+                 const std::vector<int>& demand_by_type);
+
+  // Builds the arc-flow graph.
+  ArcFlowGraph BuildVectorBinPackingGraph();
+
+  // For debugging purposes.tring(
+  // Returns the number of states explored in the dynamic programming phase.
+  int64_t NumDpStates() const;
+
+ private:
+  // All items data, regrouped for sorting purposes.
+  struct Item {
+    std::vector<int> dimensions;
+    int demand;
+    int original_index;
+
+    // Used to sort items by relative size.
+    double NormalizedSize(const std::vector<int>& bin_dimensions) const;
+  };
+
+  // State of the dynamic programming algorithm.
+  struct DpState {
+    int cur_item_index;
+    int cur_item_quantity;
+    std::vector<int> used_dimensions;
+    // DP State indices of the states that can be obtained by moving
+    // either "right" to (cur_item_index, cur_item_quantity++) or "up"
+    // to (cur_item_index++, cur_item_quantity=0). -1 if impossible.
+    int right_child;
+    int up_child;
+  };
+
+  // Add item iteratively to create all possible nodes in a forward pass.
+  void ForwardCreationPass(DpState* dp_state);
+  // Scan DP-nodes backward to relabels each nodes by increasing them as much
+  // as possible.
+  void BackwardCompressionPass(int state_index);
+  // Relabel nodes by decreasing them as much as possible.
+  void ForwardCompressionPass(const std::vector<int>& source_node);
+
+  // Can we fit one more item in the bin?
+  bool CanFitNewItem(const std::vector<int>& used_dimensions, int item) const;
+  // Create a new used_dimensions that is used_dimensions + item dimensions.
+  std::vector<int> AddItem(const std::vector<int>& used_dimensions,
+                           int item) const;
+
+  // DpState helpers.
+  int LookupOrCreateDpState(int item, int quantity,
+                            const std::vector<int>& used_dimensions);
+
+  const std::vector<int> bin_dimensions_;
+  std::vector<Item> items_;
+
+  typedef absl::flat_hash_map<std::vector<int>, int> VectorIntIntMap;
+  int GetOrCreateNode(const std::vector<int>& used_dimensions);
+
+  // We store all DP states in a dense vector, and remember their index
+  // in the dp_state_index_ map (we use a tri-dimensional indexing because
+  // it's faster for the hash part).
+  std::vector<DpState*> dp_states_;
+  std::vector<std::vector<VectorIntIntMap>> dp_state_index_;
+
+  // The ArcFlowGraph will have nodes which will correspond to "some"
+  // of the vector<int> representing the partial bin usages encountered during
+  // the algo. These two data structures map one to the other (note that nodes
+  // are dense integers).
+  absl::flat_hash_map<std::vector<int>, int> node_indices_;
+  std::vector<std::vector<int>> nodes_;
+
+  std::set<ArcFlowGraph::Arc> arcs_;
+};
+
+double ArcFlowBuilder::Item::NormalizedSize(
+    const std::vector<int>& bin_dimensions) const {
+  double size = 0.0;
+  for (int i = 0; i < bin_dimensions.size(); ++i) {
+    size += static_cast<double>(dimensions[i]) / bin_dimensions[i];
+  }
+  return size;
+}
+
+int64_t ArcFlowBuilder::NumDpStates() const {
+  int64_t res = 1;  // We do not store the initial state.
+  for (const auto& it1 : dp_state_index_) {
+    for (const auto& it2 : it1) {
+      res += it2.size();
+    }
+  }
+  return res;
+}
+
+ArcFlowBuilder::ArcFlowBuilder(
+    const std::vector<int>& bin_dimensions,
+    const std::vector<std::vector<int>>& item_dimensions_by_type,
+    const std::vector<int>& demand_by_type)
+    : bin_dimensions_(bin_dimensions) {
+  // Checks dimensions.
+  for (int i = 0; i < bin_dimensions.size(); ++i) {
+    CHECK_GT(bin_dimensions[i], 0);
+  }
+
+  const int num_items = item_dimensions_by_type.size();
+  items_.resize(num_items);
+  for (int i = 0; i < num_items; ++i) {
+    items_[i].dimensions = item_dimensions_by_type[i];
+    items_[i].demand = demand_by_type[i];
+    items_[i].original_index = i;
+  }
+  std::sort(items_.begin(), items_.end(), [&](const Item& a, const Item& b) {
+    return a.NormalizedSize(bin_dimensions_) >
+           b.NormalizedSize(bin_dimensions_);
+  });
+}
+
+bool ArcFlowBuilder::CanFitNewItem(const std::vector<int>& used_dimensions,
+                                   int item) const {
+  for (int d = 0; d < bin_dimensions_.size(); ++d) {
+    if (used_dimensions[d] + items_[item].dimensions[d] > bin_dimensions_[d]) {
+      return false;
+    }
+  }
+  return true;
+}
+
+std::vector<int> ArcFlowBuilder::AddItem(
+    const std::vector<int>& used_dimensions, int item) const {
+  DCHECK(CanFitNewItem(used_dimensions, item));
+  std::vector<int> result = used_dimensions;
+  for (int d = 0; d < bin_dimensions_.size(); ++d) {
+    result[d] += items_[item].dimensions[d];
+  }
+  return result;
+}
+
+int ArcFlowBuilder::GetOrCreateNode(const std::vector<int>& used_dimensions) {
+  const auto& it = node_indices_.find(used_dimensions);
+  if (it != node_indices_.end()) {
+    return it->second;
+  }
+  const int index = node_indices_.size();
+  node_indices_[used_dimensions] = index;
+  nodes_.push_back(used_dimensions);
+  return index;
+}
+
+ArcFlowGraph ArcFlowBuilder::BuildVectorBinPackingGraph() {
+  // Initialize the DP states map.
+  dp_state_index_.resize(items_.size());
+  for (int i = 0; i < items_.size(); ++i) {
+    dp_state_index_[i].resize(items_[i].demand + 1);
+  }
+
+  // Explore all possible DP states (starting from the initial 'empty' state),
+  // and remember their ancestry.
+  std::vector<int> zero(bin_dimensions_.size(), 0);
+  dp_states_.push_back(new DpState({0, 0, zero, -1, -1}));
+  for (int i = 0; i < dp_states_.size(); ++i) {
+    ForwardCreationPass(dp_states_[i]);
+  }
+
+  // We can clear the dp_state_index map as it will not be used anymore.
+  // From now on, we will use the dp_states.used_dimensions to store the new
+  // labels in the backward pass.
+  const int64_t num_dp_states = NumDpStates();
+  dp_state_index_.clear();
+
+  // Backwards pass: "push" the bin dimensions as far as possible.
+  const int num_states = dp_states_.size();
+  std::vector<std::pair<int, int>> flat_deps;
+  for (int i = 0; i < dp_states_.size(); ++i) {
+    if (dp_states_[i]->up_child != -1) {
+      flat_deps.push_back(std::make_pair(dp_states_[i]->up_child, i));
+    }
+    if (dp_states_[i]->right_child != -1) {
+      flat_deps.push_back(std::make_pair(dp_states_[i]->right_child, i));
+    }
+  }
+  const std::vector<int> sorted_work =
+      util::graph::DenseIntStableTopologicalSortOrDie(num_states, flat_deps);
+  for (const int w : sorted_work) {
+    BackwardCompressionPass(w);
+  }
+
+  // ForwardCreationPass again, push the bin dimensions as low as possible.
+  const std::vector<int> source_node = dp_states_[0]->used_dimensions;
+  // We can now delete the states stored in dp_states_.
+  gtl::STLDeleteElements(&dp_states_);
+  ForwardCompressionPass(source_node);
+
+  // We need to connect all nodes that corresponds to at least one item selected
+  // to the sink node.
+  const int sink_node_index = nodes_.size() - 1;
+  for (int node = 1; node < sink_node_index; ++node) {
+    arcs_.insert({node, sink_node_index, -1});
+  }
+
+  ArcFlowGraph result;
+  result.arcs.assign(arcs_.begin(), arcs_.end());
+  result.nodes.assign(nodes_.begin(), nodes_.end());
+  result.num_dp_states = num_dp_states;
+  return result;
+}
+
+int ArcFlowBuilder::LookupOrCreateDpState(
+    int item, int quantity, const std::vector<int>& used_dimensions) {
+  VectorIntIntMap& map = dp_state_index_[item][quantity];
+  const int index =
+      map.insert({used_dimensions, dp_states_.size()}).first->second;
+  if (index == dp_states_.size()) {
+    dp_states_.push_back(
+        new DpState({item, quantity, used_dimensions, -1, -1}));
+  }
+  return index;
+}
+
+void ArcFlowBuilder::ForwardCreationPass(DpState* dp_state) {
+  const int item = dp_state->cur_item_index;
+  const int quantity = dp_state->cur_item_quantity;
+  const std::vector<int>& used_dimensions = dp_state->used_dimensions;
+
+  // Explore path up.
+  if (item < items_.size() - 1) {
+    dp_state->up_child = LookupOrCreateDpState(item + 1, 0, used_dimensions);
+  } else {
+    dp_state->up_child = -1;
+  }
+
+  // Explore path right.
+  if (quantity < items_[item].demand && CanFitNewItem(used_dimensions, item)) {
+    const std::vector<int> added = AddItem(used_dimensions, item);
+    dp_state->right_child = LookupOrCreateDpState(item, quantity + 1, added);
+  } else {
+    dp_state->right_child = -1;
+  }
+}
+
+void ArcFlowBuilder::BackwardCompressionPass(int state_index) {
+  // The goal of this function is to fill this.
+  std::vector<int>& result = dp_states_[state_index]->used_dimensions;
+
+  // Inherit our result from the result one step up.
+  const int up_index = dp_states_[state_index]->up_child;
+  const std::vector<int>& result_up =
+      up_index == -1 ? bin_dimensions_ : dp_states_[up_index]->used_dimensions;
+  result = result_up;
+
+  // Adjust our result from the result one step right.
+  const int right_index = dp_states_[state_index]->right_child;
+  if (right_index == -1) return;  // We're done.
+  const std::vector<int>& result_right =
+      dp_states_[right_index]->used_dimensions;
+  const Item& item = items_[dp_states_[state_index]->cur_item_index];
+  for (int d = 0; d < bin_dimensions_.size(); ++d) {
+    result[d] = std::min(result[d], result_right[d] - item.dimensions[d]);
+  }
+
+  // Insert the arc from the node to the "right" node.
+  const int node = GetOrCreateNode(result);
+  const int right_node = GetOrCreateNode(result_right);
+  DCHECK_NE(node, right_node);
+  arcs_.insert({node, right_node, item.original_index});
+  // Also insert the 'dotted' arc from the node to the "up" node (if different).
+  if (result != result_up) {
+    const int up_node = GetOrCreateNode(result_up);
+    arcs_.insert({node, up_node, -1});
+  }
+}
+
+// Reverse version of the backward pass.
+// Revisit states forward, and relabel nodes with the longest path in each
+// dimensions from the source. The only meaningfull difference is that we use
+// arcs and nodes, instead of dp_states.
+void ArcFlowBuilder::ForwardCompressionPass(
+    const std::vector<int>& source_node) {
+  const int num_nodes = node_indices_.size();
+  const int num_dims = bin_dimensions_.size();
+  std::set<ArcFlowGraph::Arc> new_arcs;
+  std::vector<std::vector<int>> new_nodes;
+  VectorIntIntMap new_node_indices;
+  std::vector<int> node_remap(num_nodes, -1);
+  // We need to revert the sorting of items as arcs store the original index.
+  std::vector<int> reverse_item_index_map(items_.size(), -1);
+  for (int i = 0; i < items_.size(); ++i) {
+    reverse_item_index_map[items_[i].original_index] = i;
+  }
+
+  std::vector<std::pair<int, int>> forward_deps;
+  std::vector<std::vector<ArcFlowGraph::Arc>> incoming_arcs(num_nodes);
+  for (const ArcFlowGraph::Arc& arc : arcs_) {
+    forward_deps.push_back(std::make_pair(arc.source, arc.destination));
+    incoming_arcs[arc.destination].push_back(arc);
+  }
+
+  const std::vector<int> sorted_work =
+      util::graph::DenseIntStableTopologicalSortOrDie(num_nodes, forward_deps);
+
+  const int old_source_node = GetOrCreateNode(source_node);
+  const int old_sink_node = GetOrCreateNode(bin_dimensions_);
+  CHECK_EQ(sorted_work.front(), old_source_node);
+  CHECK_EQ(sorted_work.back(), old_sink_node);
+
+  // Process nodes in order and remap state to max(previous_state + item
+  // dimensions).
+  for (const int w : sorted_work) {
+    std::vector<int> new_used(num_dims, 0);
+    if (w == sorted_work.back()) {  // Do not compress the sink node.
+      new_used = bin_dimensions_;
+    } else {
+      for (const ArcFlowGraph::Arc& arc : incoming_arcs[w]) {
+        const int item =
+            arc.item_index == -1 ? -1 : reverse_item_index_map[arc.item_index];
+        const int prev_node = node_remap[arc.source];
+        const std::vector<int>& prev = new_nodes[prev_node];
+        DCHECK_NE(prev_node, -1);
+        for (int d = 0; d < num_dims; ++d) {
+          if (item != -1) {
+            new_used[d] =
+                std::max(new_used[d], prev[d] + items_[item].dimensions[d]);
+          } else {
+            new_used[d] = std::max(new_used[d], prev[d]);
+          }
+        }
+      }
+    }
+    const auto& it = new_node_indices.find(new_used);
+    if (it != new_node_indices.end()) {
+      node_remap[w] = it->second;
+    } else {
+      const int new_index = new_nodes.size();
+      new_nodes.push_back(new_used);
+      new_node_indices[new_used] = new_index;
+      node_remap[w] = new_index;
+    }
+  }
+  // Remap arcs.
+  for (const ArcFlowGraph::Arc& arc : arcs_) {
+    CHECK_NE(node_remap[arc.source], -1);
+    CHECK_NE(node_remap[arc.destination], -1);
+    // Remove loss arcs between merged nodes.
+    if (arc.item_index == -1 &&
+        node_remap[arc.source] == node_remap[arc.destination])
+      continue;
+    new_arcs.insert(
+        {node_remap[arc.source], node_remap[arc.destination], arc.item_index});
+  }
+  VLOG(1) << "Reduced nodes from " << num_nodes << " to " << new_nodes.size();
+  VLOG(1) << "Reduced arcs from " << arcs_.size() << " to " << new_arcs.size();
+  nodes_ = new_nodes;
+  arcs_ = new_arcs;
+  CHECK_NE(node_remap[old_source_node], -1);
+  CHECK_EQ(0, node_remap[old_source_node]);
+  CHECK_NE(node_remap[old_sink_node], -1);
+  CHECK_EQ(nodes_.size() - 1, node_remap[old_sink_node]);
+}
+
+}  // namespace
+
+bool ArcFlowGraph::Arc::operator<(const ArcFlowGraph::Arc& other) const {
+  if (source != other.source) return source < other.source;
+  if (destination != other.destination) return destination < other.destination;
+  return item_index < other.item_index;
+}
+
+ArcFlowGraph BuildArcFlowGraph(
+    const std::vector<int>& bin_dimensions,
+    const std::vector<std::vector<int>>& item_dimensions_by_type,
+    const std::vector<int>& demand_by_type) {
+  ArcFlowBuilder afb(bin_dimensions, item_dimensions_by_type, demand_by_type);
+  return afb.BuildVectorBinPackingGraph();
+}
+
+}  // namespace packing
+}  // namespace operations_research
diff --git a/ortools/packing/arc_flow_builder.h b/ortools/packing/arc_flow_builder.h
new file mode 100644
index 0000000000..1cb0bc1e31
--- /dev/null
+++ b/ortools/packing/arc_flow_builder.h
@@ -0,0 +1,92 @@
+// Copyright 2010-2021 Google LLC
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+// This code builds an arc-flow generator for vector-bin-packing problems.
+//   see https://people.math.gatech.edu/~tetali/PUBLIS/CKPT.pdf
+// It implements a non-recursive version of algorithm 1 described in:
+//   http://www.dcc.fc.up.pt/~fdabrandao/papers/arcflow_manuscript.pdf
+// And in (poster version):
+//   http://www.dcc.fc.up.pt/~fdabrandao/papers/arcflow_poster.pdf
+// Available at:
+//   https://drive.google.com/open?id=1y-Vs1orv-QHO4lb2sjVWrZr9GQd5d2st
+//   https://drive.google.com/open?id=1fsWRqgNJ_3ClrhoKIeVc1EOd5s8Mj33i (poster)
+// Some improvements are not yet implemented:
+//   - Lifted stated: when storing a state of the dynamic programming forward
+//     pass, one can lift a state. A lifted state of a state S is a maximal
+//     increase of S that does not lose any state in the forward pass.
+//     A simple example is the following:
+//       bin, 1 dimension, capacity 5
+//       2 item of size 2.
+//       After adding item 1 in the DP pass, the state is (2).
+//       The lifted state is (3) that is (5) - (2) which is the maximal increase
+//       of (2) that does not loose any state.
+//     To limit time spent computing this, one can lift a state only if the
+//     remaining number of item is below a threshold.
+//   - Disable the backward pass (compress state towards the bin capacity).
+//     Although this reduces the graph a lot, this simplication is not valid
+//     when the cost is not the number of bins, but a function of the capacity
+//     used (useful for fair allocation).
+
+#ifndef OR_TOOLS_PACKING_ARC_FLOW_BUILDER_H_
+#define OR_TOOLS_PACKING_ARC_FLOW_BUILDER_H_
+
+#include <cstdint>
+#include <set>
+#include <vector>
+
+#include "absl/container/flat_hash_map.h"
+#include "ortools/base/integral_types.h"
+
+namespace operations_research {
+namespace packing {
+
+// Arc flow gragh built from a vector bin packing problem.
+// The first node will always be the source. The last will always be the sink
+// of the arc-flow graph.
+struct ArcFlowGraph {
+  struct Arc {
+    int source;
+    int destination;
+    int item_index;
+
+    // Needed for std::set.
+    bool operator<(const Arc& other) const;
+  };
+
+  std::vector<Arc> arcs;
+  // All the nodes explored during the DP phase.
+  // In the forward pass, these are the consumed capacity of the bin at this
+  // state. In the backward pass, this is pushed up towards the max capacity
+  // of the bin. In the final compression phase, this is pushed down towards
+  // the initial zero state.
+  std::vector<std::vector<int>> nodes;
+  // Debug info.
+  int64_t num_dp_states;
+};
+
+// Main method.
+
+// Arc flow builder. The input must enforce the following constraints:
+//  - item_dimensions_by_type.size() == demand_by_type.size() == num types
+//  - for each type t:
+//       item_dimensions_by_type[t].size() == bin_dimensions.size() ==
+//           num_dimensions
+ArcFlowGraph BuildArcFlowGraph(
+    const std::vector<int>& bin_dimensions,
+    const std::vector<std::vector<int>>& item_dimensions_by_type,
+    const std::vector<int>& demand_by_type);
+
+}  // namespace packing
+}  // namespace operations_research
+
+#endif  // OR_TOOLS_PACKING_ARC_FLOW_BUILDER_H_
diff --git a/ortools/packing/arc_flow_solver.cc b/ortools/packing/arc_flow_solver.cc
new file mode 100644
index 0000000000..248e751a76
--- /dev/null
+++ b/ortools/packing/arc_flow_solver.cc
@@ -0,0 +1,192 @@
+// Copyright 2010-2021 Google LLC
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include "ortools/packing/arc_flow_solver.h"
+
+#include "absl/flags/flag.h"
+#include "ortools/base/commandlineflags.h"
+#include "ortools/base/file.h"
+#include "ortools/base/timer.h"
+#include "ortools/packing/arc_flow_builder.h"
+
+ABSL_FLAG(std::string, arc_flow_dump_model, "",
+          "File to store the solver specific optimization proto.");
+
+namespace operations_research {
+namespace packing {
+
+namespace {
+double ConvertVectorBinPackingProblem(const vbp::VectorBinPackingProblem& input,
+                                      ArcFlowGraph* graph) {
+  WallTimer timer;
+  timer.Start();
+  const int num_items = input.item_size();
+  const int num_dims = input.resource_capacity_size();
+
+  // Collect problem data.
+  std::vector<std::vector<int>> shapes(num_items);
+  std::vector<int> demands(num_items);
+  std::vector<int> capacities(num_dims);
+  for (int i = 0; i < num_items; ++i) {
+    shapes[i].assign(input.item(i).resource_usage().begin(),
+                     input.item(i).resource_usage().end());
+    demands[i] = input.item(i).num_copies();
+  }
+  for (int i = 0; i < num_dims; ++i) {
+    capacities[i] = input.resource_capacity(i);
+  }
+
+  // Add extra dimensions to encode max_number_of_copies_per_bin.
+  for (int i = 0; i < num_items; ++i) {
+    const int max_copies = input.item(i).max_number_of_copies_per_bin();
+    if (max_copies == 0 || max_copies >= demands[i]) continue;
+    capacities.push_back(max_copies);
+    for (int j = 0; j < num_items; ++j) {
+      shapes[j].push_back(i == j);
+    }
+  }
+
+  *graph = BuildArcFlowGraph(capacities, shapes, demands);
+  const double arc_flow_time = timer.Get();
+
+  VLOG(1) << "The arc-flow grah has " << graph->nodes.size() << " nodes, and "
+          << graph->arcs.size() << " arcs. It was created by exploring "
+          << graph->num_dp_states
+          << " states in the dynamic programming phase in " << arc_flow_time
+          << " s";
+  return arc_flow_time;
+}
+}  // namespace
+
+vbp::VectorBinPackingSolution SolveVectorBinPackingWithArcFlow(
+    const vbp::VectorBinPackingProblem& problem,
+    MPSolver::OptimizationProblemType solver_type,
+    const std::string& mip_params, double time_limit, int num_threads,
+    bool log_statistics) {
+  ArcFlowGraph graph;
+  const double arc_flow_time = ConvertVectorBinPackingProblem(problem, &graph);
+
+  int max_num_bins = 0;
+  for (const auto& item : problem.item()) {
+    max_num_bins += item.num_copies();
+  }
+  const int num_types = problem.item_size();
+  std::vector<std::vector<MPVariable*>> incoming_vars(graph.nodes.size());
+  std::vector<std::vector<MPVariable*>> outgoing_vars(graph.nodes.size());
+  std::vector<MPVariable*> arc_to_var(graph.arcs.size());
+  std::vector<std::vector<MPVariable*>> item_to_vars(num_types);
+
+  MPSolver solver("VectorBinPacking", solver_type);
+  CHECK_OK(solver.SetNumThreads(num_threads));
+
+  for (int v = 0; v < graph.arcs.size(); ++v) {
+    const ArcFlowGraph::Arc& arc = graph.arcs[v];
+    MPVariable* const var =
+        solver.MakeIntVar(0, max_num_bins, absl::StrCat("a", v));
+    incoming_vars[arc.destination].push_back(var);
+    outgoing_vars[arc.source].push_back(var);
+    if (arc.item_index != -1) {
+      item_to_vars[arc.item_index].push_back(var);
+    }
+  }
+
+  // Per item demand constraint.
+  for (int i = 0; i < num_types; ++i) {
+    MPConstraint* const ct = solver.MakeRowConstraint(
+        problem.item(i).num_copies(), problem.item(i).num_copies());
+    for (MPVariable* const var : item_to_vars[i]) {
+      ct->SetCoefficient(var, 1.0);
+    }
+  }
+
+  // Flow conservation.
+  for (int n = 1; n < graph.nodes.size() - 1; ++n) {  // Ignore source and sink.
+    MPConstraint* const ct = solver.MakeRowConstraint(0.0, 0.0);
+    for (MPVariable* const var : incoming_vars[n]) {
+      ct->SetCoefficient(var, 1.0);
+    }
+    for (MPVariable* const var : outgoing_vars[n]) {
+      ct->SetCoefficient(var, -1.0);
+    }
+  }
+
+  MPVariable* const obj_var = solver.MakeIntVar(0, max_num_bins, "obj_var");
+  {  // Source.
+    MPConstraint* const ct = solver.MakeRowConstraint(0.0, 0.0);
+    for (MPVariable* const var : outgoing_vars[/*source*/ 0]) {
+      ct->SetCoefficient(var, 1.0);
+    }
+    ct->SetCoefficient(obj_var, -1.0);
+  }
+
+  {  // Sink.
+    MPConstraint* const ct = solver.MakeRowConstraint(0.0, 0.0);
+    const int sink_node = graph.nodes.size() - 1;
+    for (MPVariable* const var : incoming_vars[sink_node]) {
+      ct->SetCoefficient(var, 1.0);
+    }
+    ct->SetCoefficient(obj_var, -1.0);
+  }
+
+  MPObjective* const objective = solver.MutableObjective();
+  objective->SetCoefficient(obj_var, 1.0);
+
+  if (!absl::GetFlag(FLAGS_arc_flow_dump_model).empty()) {
+    MPModelProto output_model;
+    solver.ExportModelToProto(&output_model);
+    CHECK_OK(file::SetTextProto(absl::GetFlag(FLAGS_arc_flow_dump_model),
+                                output_model, file::Defaults()));
+  }
+
+  solver.EnableOutput();
+  solver.SetSolverSpecificParametersAsString(mip_params);
+  solver.SetTimeLimit(absl::Seconds(time_limit));
+  const MPSolver::ResultStatus result_status = solver.Solve();
+
+  vbp::VectorBinPackingSolution solution;
+  solution.set_solve_time_in_seconds(solver.wall_time() / 1000.0);
+  solution.set_arc_flow_time_in_seconds(arc_flow_time);
+  // Check that the problem has an optimal solution.
+  if (result_status == MPSolver::OPTIMAL) {
+    solution.set_status(vbp::OPTIMAL);
+    solution.set_objective_value(objective->Value());
+  } else if (result_status == MPSolver::FEASIBLE) {
+    solution.set_status(vbp::FEASIBLE);
+    solution.set_objective_value(objective->Value());
+  } else if (result_status == MPSolver::INFEASIBLE) {
+    solution.set_status(vbp::INFEASIBLE);
+  }
+
+  // TODO(user): Fill bins in the solution proto.
+
+  if (log_statistics) {
+    const bool has_solution = result_status == MPSolver::OPTIMAL ||
+                              result_status == MPSolver::FEASIBLE;
+    absl::PrintF("%-12s: %s\n", "Status",
+                 MPSolverResponseStatus_Name(
+                     static_cast<MPSolverResponseStatus>(result_status))
+                     .c_str());
+    absl::PrintF("%-12s: %15.15e\n", "Objective",
+                 has_solution ? solver.Objective().Value() : 0.0);
+    absl::PrintF("%-12s: %15.15e\n", "BestBound",
+                 has_solution ? solver.Objective().BestBound() : 0.0);
+    absl::PrintF("%-12s: %d\n", "Iterations", solver.iterations());
+    absl::PrintF("%-12s: %d\n", "Nodes", solver.nodes());
+    absl::PrintF("%-12s: %-6.4g\n", "Time", solution.solve_time_in_seconds());
+  }
+
+  return solution;
+}
+
+}  // namespace packing
+}  // namespace operations_research
diff --git a/ortools/data/set_covering_data.cc b/ortools/packing/arc_flow_solver.h
similarity index 53%
rename from ortools/data/set_covering_data.cc
rename to ortools/packing/arc_flow_solver.h
index eb41ea149a..394b81cfe4 100644
--- a/ortools/data/set_covering_data.cc
+++ b/ortools/packing/arc_flow_solver.h
@@ -11,27 +11,22 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "ortools/data/set_covering_data.h"
+#ifndef OR_TOOLS_PACKING_ARC_FLOW_SOLVER_H_
+#define OR_TOOLS_PACKING_ARC_FLOW_SOLVER_H_
+
+#include "ortools/linear_solver/linear_solver.h"
+#include "ortools/packing/vector_bin_packing.pb.h"
 
 namespace operations_research {
-namespace scp {
+namespace packing {
 
-void ScpData::SetProblemSize(int num_rows, int num_columns) {
-  columns_per_row_.clear();
-  columns_per_row_.resize(num_rows);
-  rows_per_column_.clear();
-  rows_per_column_.resize(num_columns);
-  column_costs_.resize(num_columns, 0);
-}
+vbp::VectorBinPackingSolution SolveVectorBinPackingWithArcFlow(
+    const vbp::VectorBinPackingProblem& problem,
+    MPSolver::OptimizationProblemType solver_type,
+    const std::string& mip_params, double time_limit, int num_threads,
+    bool log_statistics);
 
-void ScpData::SetColumnCost(int column_id, int cost) {
-  column_costs_[column_id] = cost;
-}
-
-void ScpData::AddRowInColumn(int row_id, int column_id) {
-  rows_per_column_[column_id].push_back(row_id);
-  columns_per_row_[row_id].push_back(column_id);
-}
-
-}  // namespace scp
+}  // namespace packing
 }  // namespace operations_research
+
+#endif  // OR_TOOLS_PACKING_ARC_FLOW_SOLVER_H_
diff --git a/ortools/data/vector_bin_packing.proto b/ortools/packing/vector_bin_packing.proto
similarity index 96%
rename from ortools/data/vector_bin_packing.proto
rename to ortools/packing/vector_bin_packing.proto
index d1f4614df5..ff0d1471a3 100644
--- a/ortools/data/vector_bin_packing.proto
+++ b/ortools/packing/vector_bin_packing.proto
@@ -36,11 +36,11 @@
 
 syntax = "proto3";
 
-option java_package = "com.google.ortools.data.vbp";
+option java_package = "com.google.ortools.packing.vbp";
 option java_multiple_files = true;
-option csharp_namespace = "Google.OrTools.Data.Vbp";
+option csharp_namespace = "Google.OrTools.Packing.Vbp";
 
-package operations_research.data.vbp;
+package operations_research.packing.vbp;
 
 message Item {
   // Optional name. This is only used for display/debugging purposes.
diff --git a/ortools/data/vector_bin_packing_parser.cc b/ortools/packing/vector_bin_packing_parser.cc
similarity index 95%
rename from ortools/data/vector_bin_packing_parser.cc
rename to ortools/packing/vector_bin_packing_parser.cc
index 5d1da4e0b5..6df8a4ce2c 100644
--- a/ortools/data/vector_bin_packing_parser.cc
+++ b/ortools/packing/vector_bin_packing_parser.cc
@@ -11,17 +11,17 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "ortools/data/vector_bin_packing_parser.h"
+#include "ortools/packing/vector_bin_packing_parser.h"
 
 #include <cstdint>
 
 #include "absl/strings/numbers.h"
 #include "absl/strings/str_split.h"
 #include "ortools/base/filelineiter.h"
-#include "ortools/data/vector_bin_packing.pb.h"
+#include "ortools/packing/vector_bin_packing.pb.h"
 
 namespace operations_research {
-namespace data {
+namespace packing {
 namespace vbp {
 
 bool VbpParser::ParseFile(const std::string& data_filename) {
@@ -117,5 +117,5 @@ int64_t VbpParser::strtoint64(const std::string& word) {
 }
 
 }  // namespace vbp
-}  // namespace data
+}  // namespace packing
 }  // namespace operations_research
diff --git a/ortools/data/vector_bin_packing_parser.h b/ortools/packing/vector_bin_packing_parser.h
similarity index 83%
rename from ortools/data/vector_bin_packing_parser.h
rename to ortools/packing/vector_bin_packing_parser.h
index 848e5385a8..9d6a1aad3c 100644
--- a/ortools/data/vector_bin_packing_parser.h
+++ b/ortools/packing/vector_bin_packing_parser.h
@@ -17,18 +17,18 @@
 //   - vector packing solver: (.vbp files)
 //         http://www.dcc.fc.up.pt/~fdabrandao/Vector_Packing_Solver
 
-#ifndef OR_TOOLS_DATA_VECTOR_BIN_PACKING_PARSER_H_
-#define OR_TOOLS_DATA_VECTOR_BIN_PACKING_PARSER_H_
+#ifndef OR_TOOLS_PACKING_VECTOR_BIN_PACKING_PARSER_H_
+#define OR_TOOLS_PACKING_VECTOR_BIN_PACKING_PARSER_H_
 
 #include <cstdint>
 #include <string>
 #include <vector>
 
 #include "ortools/base/integral_types.h"
-#include "ortools/data/vector_bin_packing.pb.h"
+#include "ortools/packing/vector_bin_packing.pb.h"
 
 namespace operations_research {
-namespace data {
+namespace packing {
 namespace vbp {
 
 class VbpParser {
@@ -38,7 +38,7 @@ class VbpParser {
   bool ParseFile(const std::string& data_filename);
 
   // We keep the fully qualified name for SWIG.
-  ::operations_research::data::vbp::VectorBinPackingProblem problem() const {
+  ::operations_research::packing::vbp::VectorBinPackingProblem problem() const {
     return vbp_;
   }
 
@@ -65,7 +65,7 @@ class VbpParser {
 };
 
 }  // namespace vbp
-}  // namespace data
+}  // namespace packing
 }  // namespace operations_research
 
-#endif  // OR_TOOLS_DATA_VECTOR_BIN_PACKING_PARSER_H_
+#endif  // OR_TOOLS_PACKING_VECTOR_BIN_PACKING_PARSER_H_
diff --git a/ortools/data/BUILD b/ortools/scheduling/BUILD
similarity index 100%
rename from ortools/data/BUILD
rename to ortools/scheduling/BUILD
diff --git a/ortools/data/CMakeLists.txt b/ortools/scheduling/CMakeLists.txt
similarity index 83%
rename from ortools/data/CMakeLists.txt
rename to ortools/scheduling/CMakeLists.txt
index 141a971fa1..e0c3a831df 100644
--- a/ortools/data/CMakeLists.txt
+++ b/ortools/scheduling/CMakeLists.txt
@@ -1,5 +1,5 @@
 file(GLOB _SRCS "*.h" "*.cc")
-set(NAME ${PROJECT_NAME}_data)
+set(NAME ${PROJECT_NAME}_scheduling)
 
 # Will be merge in libortools.so
 #add_library(${NAME} STATIC ${_SRCS})
@@ -17,4 +17,4 @@ target_link_libraries(${NAME} PRIVATE
   absl::strings
   protobuf::libprotobuf
   ${PROJECT_NAME}::proto)
-#add_library(${PROJECT_NAME}::data ALIAS ${NAME})
+#add_library(${PROJECT_NAME}::scheduling ALIAS ${NAME})
diff --git a/ortools/scheduling/__init__.py b/ortools/scheduling/__init__.py
new file mode 100644
index 0000000000..1ecf401ebf
--- /dev/null
+++ b/ortools/scheduling/__init__.py
@@ -0,0 +1,15 @@
+# Copyright 2010-2021 Google LLC
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+import os as _os
+__path__.append(_os.path.join(__path__[0],  '..', 'gen', 'ortools', 'scheduling'))
+__path__.append(_os.path.join(__path__[0],  '..', '..', 'lib'))
diff --git a/ortools/data/jobshop_scheduling.proto b/ortools/scheduling/jobshop_scheduling.proto
similarity index 97%
rename from ortools/data/jobshop_scheduling.proto
rename to ortools/scheduling/jobshop_scheduling.proto
index a8f91ee425..faf30dd73e 100644
--- a/ortools/data/jobshop_scheduling.proto
+++ b/ortools/scheduling/jobshop_scheduling.proto
@@ -39,11 +39,11 @@
 
 syntax = "proto3";
 
-option java_package = "com.google.ortools.data.jssp";
+option java_package = "com.google.ortools.scheduling.jssp";
 option java_multiple_files = true;
-option csharp_namespace = "Google.OrTools.Data.Jssp";
+option csharp_namespace = "Google.OrTools.scheduling.Jssp";
 
-package operations_research.data.jssp;
+package operations_research.scheduling.jssp;
 
 import "google/protobuf/wrappers.proto";
 
diff --git a/ortools/data/jobshop_scheduling_parser.cc b/ortools/scheduling/jobshop_scheduling_parser.cc
similarity index 99%
rename from ortools/data/jobshop_scheduling_parser.cc
rename to ortools/scheduling/jobshop_scheduling_parser.cc
index 5d1c82e06b..765277b898 100644
--- a/ortools/data/jobshop_scheduling_parser.cc
+++ b/ortools/scheduling/jobshop_scheduling_parser.cc
@@ -11,7 +11,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "ortools/data/jobshop_scheduling_parser.h"
+#include "ortools/scheduling/jobshop_scheduling_parser.h"
 
 #include <cmath>
 #include <cstdint>
@@ -24,13 +24,13 @@
 #include "ortools/base/filelineiter.h"
 #include "ortools/base/integral_types.h"
 #include "ortools/base/logging.h"
-#include "ortools/data/jobshop_scheduling.pb.h"
+#include "ortools/scheduling/jobshop_scheduling.pb.h"
 
 ABSL_FLAG(int64_t, jssp_scaling_up_factor, 100000L,
           "Scaling factor for floating point penalties.");
 
 namespace operations_research {
-namespace data {
+namespace scheduling {
 namespace jssp {
 
 void JsspParser::SetJobs(int job_count) {
@@ -541,5 +541,5 @@ int64_t JsspParser::strtoint64(const std::string& word) {
 }
 
 }  // namespace jssp
-}  // namespace data
+}  // namespace scheduling
 }  // namespace operations_research
diff --git a/ortools/data/jobshop_scheduling_parser.h b/ortools/scheduling/jobshop_scheduling_parser.h
similarity index 89%
rename from ortools/data/jobshop_scheduling_parser.h
rename to ortools/scheduling/jobshop_scheduling_parser.h
index 9d99849ef9..dc3d404103 100644
--- a/ortools/data/jobshop_scheduling_parser.h
+++ b/ortools/scheduling/jobshop_scheduling_parser.h
@@ -11,17 +11,17 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#ifndef OR_TOOLS_DATA_JOBSHOP_SCHEDULING_PARSER_H_
-#define OR_TOOLS_DATA_JOBSHOP_SCHEDULING_PARSER_H_
+#ifndef OR_TOOLS_SCHEDULING_JOBSHOP_SCHEDULING_PARSER_H_
+#define OR_TOOLS_SCHEDULING_JOBSHOP_SCHEDULING_PARSER_H_
 
 #include <cstdint>
 
 #include "absl/strings/match.h"
 #include "ortools/base/integral_types.h"
-#include "ortools/data/jobshop_scheduling.pb.h"
+#include "ortools/scheduling/jobshop_scheduling.pb.h"
 
 namespace operations_research {
-namespace data {
+namespace scheduling {
 namespace jssp {
 
 class JsspParser {
@@ -87,7 +87,7 @@ class JsspParser {
 };
 
 }  // namespace jssp
-}  // namespace data
+}  // namespace scheduling
 }  // namespace operations_research
 
-#endif  // OR_TOOLS_DATA_JOBSHOP_SCHEDULING_PARSER_H_
+#endif  // OR_TOOLS_SCHEDULING_JOBSHOP_SCHEDULING_PARSER_H_
diff --git a/ortools/data/python/CMakeLists.txt b/ortools/scheduling/python/CMakeLists.txt
similarity index 94%
rename from ortools/data/python/CMakeLists.txt
rename to ortools/scheduling/python/CMakeLists.txt
index d603d95ac2..6e25e09729 100644
--- a/ortools/data/python/CMakeLists.txt
+++ b/ortools/scheduling/python/CMakeLists.txt
@@ -5,7 +5,7 @@ set_property(SOURCE rcpsp.i PROPERTY COMPILE_DEFINITIONS
 swig_add_library(pywraprcpsp
   TYPE SHARED
   LANGUAGE python
-  OUTPUT_DIR  ${PROJECT_BINARY_DIR}/python/${PROJECT_NAME}/data
+  OUTPUT_DIR  ${PROJECT_BINARY_DIR}/python/${PROJECT_NAME}/scheduling
   SOURCES rcpsp.i)
 
 target_include_directories(pywraprcpsp PRIVATE ${Python3_INCLUDE_DIRS})
@@ -31,4 +31,4 @@ target_link_libraries(pywraprcpsp PRIVATE ortools::ortools)
 # see: https://cmake.org/cmake/help/git-stage/command/target_link_libraries.html#command:target_link_libraries
 if(MSVC)
   target_link_libraries(pywraprcpsp PRIVATE ${Python3_LIBRARIES})
-endif()
+endif()
\ No newline at end of file
diff --git a/ortools/data/python/rcpsp.i b/ortools/scheduling/python/rcpsp.i
similarity index 53%
rename from ortools/data/python/rcpsp.i
rename to ortools/scheduling/python/rcpsp.i
index af47c86323..f1f4982cef 100644
--- a/ortools/data/python/rcpsp.i
+++ b/ortools/scheduling/python/rcpsp.i
@@ -15,24 +15,24 @@
 %include "ortools/util/python/proto.i"
 
 %{
-#include "ortools/data/rcpsp_parser.h"
-#include "ortools/data/rcpsp.pb.h"
+#include "ortools/scheduling/rcpsp_parser.h"
+#include "ortools/scheduling/rcpsp.pb.h"
 %}
 
-PY_PROTO_TYPEMAP(ortools.data.rcpsp_pb2,
+PY_PROTO_TYPEMAP(ortools.scheduling.rcpsp_pb2,
                  RcpspProblem,
-                 ::operations_research::data::rcpsp::RcpspProblem);
+                 ::operations_research::scheduling::rcpsp::RcpspProblem);
 
 %ignoreall
 
 %unignore operations_research;
-%unignore operations_research::data;
-%unignore operations_research::data::rcpsp;
-%unignore operations_research::data::rcpsp::RcpspParser;
-%unignore operations_research::data::rcpsp::RcpspParser::RcpspParser;
-%rename (Problem) operations_research::data::rcpsp::RcpspParser::problem;
-%unignore operations_research::data::rcpsp::RcpspParser::ParseFile;
+%unignore operations_research::scheduling;
+%unignore operations_research::scheduling::rcpsp;
+%unignore operations_research::scheduling::rcpsp::RcpspParser;
+%unignore operations_research::scheduling::rcpsp::RcpspParser::RcpspParser();
+%rename (Problem) operations_research::scheduling::rcpsp::RcpspParser::problem;
+%unignore operations_research::scheduling::rcpsp::RcpspParser::ParseFile(const std::string& file_name);
 
-%include "ortools/data/rcpsp_parser.h"
+%include "ortools/scheduling/rcpsp_parser.h"
 
 %unignoreall
diff --git a/ortools/data/rcpsp.proto b/ortools/scheduling/rcpsp.proto
similarity index 97%
rename from ortools/data/rcpsp.proto
rename to ortools/scheduling/rcpsp.proto
index ef973e4417..3acad14b54 100644
--- a/ortools/data/rcpsp.proto
+++ b/ortools/scheduling/rcpsp.proto
@@ -67,11 +67,11 @@
 
 syntax = "proto3";
 
-option java_package = "com.google.ortools.data.rcpsp";
+option java_package = "com.google.ortools.scheduling.rcpsp";
 option java_multiple_files = true;
-option csharp_namespace = "Google.OrTools.Data.Rcpsp";
+option csharp_namespace = "Google.OrTools.Scheduling.Rcpsp";
 
-package operations_research.data.rcpsp;
+package operations_research.scheduling.rcpsp;
 
 message Resource {
   // The max capacity of the cumulative.
diff --git a/ortools/data/rcpsp_parser.cc b/ortools/scheduling/rcpsp_parser.cc
similarity index 99%
rename from ortools/data/rcpsp_parser.cc
rename to ortools/scheduling/rcpsp_parser.cc
index ca857a7493..97c74b1a15 100644
--- a/ortools/data/rcpsp_parser.cc
+++ b/ortools/scheduling/rcpsp_parser.cc
@@ -11,7 +11,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "ortools/data/rcpsp_parser.h"
+#include "ortools/scheduling/rcpsp_parser.h"
 
 #include <cstdint>
 
@@ -19,10 +19,10 @@
 #include "absl/strings/numbers.h"
 #include "absl/strings/str_split.h"
 #include "ortools/base/filelineiter.h"
-#include "ortools/data/rcpsp.pb.h"
+#include "ortools/scheduling/rcpsp.pb.h"
 
 namespace operations_research {
-namespace data {
+namespace scheduling {
 namespace rcpsp {
 
 RcpspParser::RcpspParser()
@@ -605,5 +605,5 @@ int64_t RcpspParser::strtoint64(const std::string& word) {
 }
 
 }  // namespace rcpsp
-}  // namespace data
+}  // namespace scheduling
 }  // namespace operations_research
diff --git a/ortools/data/rcpsp_parser.h b/ortools/scheduling/rcpsp_parser.h
similarity index 87%
rename from ortools/data/rcpsp_parser.h
rename to ortools/scheduling/rcpsp_parser.h
index 692118bf35..2c3658bf01 100644
--- a/ortools/data/rcpsp_parser.h
+++ b/ortools/scheduling/rcpsp_parser.h
@@ -14,17 +14,17 @@
 // A Project Scheduling Library parser.
 // See: http://www.om-db.wi.tum.de/psplib/  # PSP-Lib homepage.
 
-#ifndef OR_TOOLS_DATA_RCPSP_PARSER_H_
-#define OR_TOOLS_DATA_RCPSP_PARSER_H_
+#ifndef OR_TOOLS_SCHEDULING_RCPSP_PARSER_H_
+#define OR_TOOLS_SCHEDULING_RCPSP_PARSER_H_
 
 #include <string>
 #include <vector>
 
 #include "ortools/base/integral_types.h"
-#include "ortools/data/rcpsp.pb.h"
+#include "ortools/scheduling/rcpsp.pb.h"
 
 namespace operations_research {
-namespace data {
+namespace scheduling {
 namespace rcpsp {
 
 // RCPSP parser.
@@ -35,7 +35,7 @@ class RcpspParser {
   RcpspParser();
 
   // We keep the fully qualified name for swig.
-  ::operations_research::data::rcpsp::RcpspProblem problem() const {
+  ::operations_research::scheduling::rcpsp::RcpspProblem problem() const {
     return rcpsp_;
   }
 
@@ -78,7 +78,7 @@ class RcpspParser {
 };
 
 }  // namespace rcpsp
-}  // namespace data
+}  // namespace scheduling
 }  // namespace operations_research
 
-#endif  // OR_TOOLS_DATA_RCPSP_PARSER_H_
+#endif  // OR_TOOLS_SCHEDULING_RCPSP_PARSER_H_
diff --git a/tools/generate_all_deps.sh b/tools/generate_all_deps.sh
index 3cef98ea33..0a5b0918aa 100755
--- a/tools/generate_all_deps.sh
+++ b/tools/generate_all_deps.sh
@@ -4,15 +4,15 @@ tools/generate_deps.sh BASE base
 tools/generate_deps.sh PORT port base
 tools/generate_deps.sh UTIL util base port
 tools/generate_deps.sh INIT init
-tools/generate_deps.sh DATA data base port util
+tools/generate_deps.sh SCHEDULING scheduling base port util
 tools/generate_deps.sh LP_DATA lp_data util base algorithms linear_solver
 tools/generate_deps.sh GLOP glop util base lp_data linear_solver
 tools/generate_deps.sh GRAPH graph base util
 tools/generate_deps.sh ALGORITHMS algorithms base util graph linear_solver
 tools/generate_deps.sh SAT sat base util algorithms graph lp_data glop linear_solver
+tools/generate_deps.sh PACKING packing base graph port util linear_solver
 tools/generate_deps.sh BOP bop base util lp_data glop sat
 tools/generate_deps.sh GSCIP gscip base port
 tools/generate_deps.sh GUROBI gurobi base
 tools/generate_deps.sh LP linear_solver base util lp_data glop bop gscip gurobi
 tools/generate_deps.sh CP constraint_solver base util graph linear_solver sat
-