Merge pull request #252624 from tobim/pkgs/arrow-cpp-13

arrow-cpp: 12.0.1 -> 13.0.0

pkgs/development/libraries/arrow-cpp/cmake-find-protobuf.patch

@@ -1,18 +1,105 @@
-diff --git a/cmake_modules/FindProtobufAlt.cmake b/cmake_modules/FindProtobufAlt.cmake
-index d29f757ae..61c6e16e1 100644
+diff --git a/cmake_modules/FindProtobufAlt.cmake b/cmake_modules/FindprotobufAlt.cmake
+similarity index 60%
+rename from cmake_modules/FindProtobufAlt.cmake
+rename to cmake_modules/FindprotobufAlt.cmake
+index 15fe1b4f2..19bdd6291 100644
 --- a/cmake_modules/FindProtobufAlt.cmake
-+++ b/cmake_modules/FindProtobufAlt.cmake
-@@ -22,11 +22,8 @@ else()
++++ b/cmake_modules/FindprotobufAlt.cmake
+@@ -16,24 +16,24 @@
+ # under the License.
+ 
+ if(ARROW_PROTOBUF_USE_SHARED)
+-  set(Protobuf_USE_STATIC_LIBS OFF)
++  set(protobuf_USE_STATIC_LIBS OFF)
+ else()
+-  set(Protobuf_USE_STATIC_LIBS ON)
++  set(protobuf_USE_STATIC_LIBS ON)
  endif()
  
  set(find_package_args)
 -if(ProtobufAlt_FIND_VERSION)
 -  list(APPEND find_package_args ${ProtobufAlt_FIND_VERSION})
--endif()
- if(ProtobufAlt_FIND_QUIETLY)
++if(protobufAlt_FIND_VERSION)
++  list(APPEND find_package_args ${protobufAlt_FIND_VERSION})
+ endif()
+-if(ProtobufAlt_FIND_QUIETLY)
++if(protobufAlt_FIND_QUIETLY)
    list(APPEND find_package_args QUIET)
  endif()
 -find_package(Protobuf ${find_package_args})
 -set(ProtobufAlt_FOUND ${Protobuf_FOUND})
-+find_package(protobuf ${find_package_args})
-+set(ProtobufAlt_FOUND ${protobuf_FOUND})
+-if(ProtobufAlt_FOUND)
+-  set(ProtobufAlt_VERSION ${Protobuf_VERSION})
+-  set(ProtobufAlt_VERSION_MAJOR ${Protobuf_VERSION_MAJOR})
+-  set(ProtobufAlt_VERSION_MINOR ${Protobuf_VERSION_MINOR})
+-  set(ProtobufAlt_VERSION_PATCH ${Protobuf_VERSION_PATCH})
+-  set(ProtobufAlt_VERSION_TWEEK ${Protobuf_VERSION_TWEEK})
++find_package(protobuf ${find_package_args} CONFIG)
++set(protobufAlt_FOUND ${protobuf_FOUND})
++if(protobufAlt_FOUND)
++  set(protobufAlt_VERSION ${protobuf_VERSION})
++  set(protobufAlt_VERSION_MAJOR ${protobuf_VERSION_MAJOR})
++  set(protobufAlt_VERSION_MINOR ${protobuf_VERSION_MINOR})
++  set(protobufAlt_VERSION_PATCH ${protobuf_VERSION_PATCH})
++  set(protobufAlt_VERSION_TWEEK ${protobuf_VERSION_TWEEK})
+ endif()
+diff --git a/cmake_modules/ThirdpartyToolchain.cmake b/cmake_modules/ThirdpartyToolchain.cmake
+index 94f926039..8efc98323 100644
+--- a/cmake_modules/ThirdpartyToolchain.cmake
++++ b/cmake_modules/ThirdpartyToolchain.cmake
+@@ -63,7 +63,7 @@ set(ARROW_THIRDPARTY_DEPENDENCIES
+     opentelemetry-cpp
+     ORC
+     re2
+-    Protobuf
++    protobuf
+     RapidJSON
+     Snappy
+     Substrait
+@@ -189,7 +189,7 @@ macro(build_dependency DEPENDENCY_NAME)
+     build_opentelemetry()
+   elseif("${DEPENDENCY_NAME}" STREQUAL "ORC")
+     build_orc()
+-  elseif("${DEPENDENCY_NAME}" STREQUAL "Protobuf")
++  elseif("${DEPENDENCY_NAME}" STREQUAL "protobuf")
+     build_protobuf()
+   elseif("${DEPENDENCY_NAME}" STREQUAL "RapidJSON")
+     build_rapidjson()
+@@ -1722,7 +1722,7 @@ if(ARROW_WITH_PROTOBUF)
+   # "3" for 3.12.0 are different. So 23.4 is rejected with 3.12.0. If
+   # we use FORCE_ANY_NEWER_VERSION here, we can bypass the check and
+   # use 23.4.
+-  resolve_dependency(Protobuf
++  resolve_dependency(protobuf
+                      FORCE_ANY_NEWER_VERSION
+                      TRUE
+                      HAVE_ALT
+@@ -1732,7 +1732,7 @@ if(ARROW_WITH_PROTOBUF)
+                      PC_PACKAGE_NAMES
+                      protobuf)
+ 
+-  if(NOT Protobuf_USE_STATIC_LIBS AND MSVC_TOOLCHAIN)
++  if(NOT protobuf_USE_STATIC_LIBS AND MSVC_TOOLCHAIN)
+     add_definitions(-DPROTOBUF_USE_DLLS)
+   endif()
+ 
+@@ -1768,8 +1768,8 @@ if(ARROW_WITH_PROTOBUF)
+   message(STATUS "Found protoc: ${PROTOBUF_PROTOC_EXECUTABLE}")
+   get_target_property(PROTOBUF_TYPE ${ARROW_PROTOBUF_LIBPROTOBUF} TYPE)
+   if(NOT STREQUAL "INTERFACE_LIBRARY")
+-    # Protobuf_PROTOC_LIBRARY is set by all versions of FindProtobuf.cmake
+-    message(STATUS "Found libprotoc: ${Protobuf_PROTOC_LIBRARY}")
++    # protobuf_PROTOC_LIBRARY is set by all versions of FindProtobuf.cmake
++    message(STATUS "Found libprotoc: ${protobuf_PROTOC_LIBRARY}")
+     get_target_property(PROTOBUF_LIBRARY ${ARROW_PROTOBUF_LIBPROTOBUF} IMPORTED_LOCATION)
+     message(STATUS "Found libprotobuf: ${PROTOBUF_LIBRARY}")
+     message(STATUS "Found protobuf headers: ${PROTOBUF_INCLUDE_DIR}")
+@@ -3986,7 +3986,7 @@ endmacro()
+ 
+ if(ARROW_WITH_GRPC)
+   set(ARROW_GRPC_REQUIRED_VERSION "1.30.0")
+-  if(NOT Protobuf_SOURCE STREQUAL gRPC_SOURCE)
++  if(NOT protobuf_SOURCE STREQUAL gRPC_SOURCE)
+     # ARROW-15495: Protobuf/gRPC must come from the same source
+     message(STATUS "Forcing gRPC_SOURCE to Protobuf_SOURCE (${Protobuf_SOURCE})")
+     set(gRPC_SOURCE "${Protobuf_SOURCE}")

pkgs/development/libraries/arrow-cpp/darwin.patch

@@ -1,11 +0,0 @@
-diff --git a/cmake_modules/FindPythonLibsNew.cmake b/cmake_modules/FindPythonLibsNew.cmake
---- a/cmake_modules/FindPythonLibsNew.cmake
-+++ b/cmake_modules/FindPythonLibsNew.cmake
-@@ -117,6 +117,7 @@ list(GET _PYTHON_VALUES 6 PYTHON_SIZEOF_VOID_P)
- list(GET _PYTHON_VALUES 7 PYTHON_LIBRARY_SUFFIX)
- list(GET _PYTHON_VALUES 8 PYTHON_LIBRARY_PATH)
- list(GET _PYTHON_VALUES 9 PYTHON_OTHER_LIBS)
-+string(REPLACE "-lncurses" "" PYTHON_OTHER_LIBS "${PYTHON_OTHER_LIBS}")
- 
- # Make sure the Python has the same pointer-size as the chosen compiler
- # Skip the check on OS X, it doesn't consistently have CMAKE_SIZEOF_VOID_P defined

pkgs/development/libraries/arrow-cpp/default.nix

@@ -78,11 +78,11 @@ let
 in
 stdenv.mkDerivation rec {
   pname = "arrow-cpp";
-  version = "12.0.1";
+  version = "13.0.0";
 
   src = fetchurl {
     url = "mirror://apache/arrow/arrow-${version}/apache-arrow-${version}.tar.gz";
-    hash = "sha256-NIHEETk6oVx16I2Tz4MV+vf0PhgP4HkBKNOEDUF96Fg=";
+    hash = "sha256-Nd/aGRJip1a+k07viv7o0JdiytJQIdqmJuskniUayeY=";
   };
 
   sourceRoot = "apache-arrow-${version}/cpp";
@@ -120,8 +120,7 @@ stdenv.mkDerivation rec {
   };
 
   patches = [
-    # patch to fix python-test
-    ./darwin.patch
+    # Protobuf switched to lower case project name.
     ./cmake-find-protobuf.patch
   ];
 

pkgs/development/python-modules/dask-awkward/default.nix

@@ -3,6 +3,7 @@
 , buildPythonPackage
 , dask
 , fetchFromGitHub
+, fetchpatch
 , hatch-vcs
 , hatchling
 , pyarrow
@@ -25,6 +26,14 @@ buildPythonPackage rec {
     hash = "sha256-sSsd35Psf3VEydkNxtd9mSBzV23S7fRM/jhbC9T62kY=";
   };
 
+  patches = [
+    (fetchpatch {
+      name = "dask-awkward-pyarrow13-test-fixes.patch";
+      url = "https://github.com/dask-contrib/dask-awkward/commit/abe7f4504b4f926232e4d0dfa5c601d265773d85.patch";
+      hash = "sha256-IYlKTV6YasuKIJutB4cCmHeglGWUwBcvFgx1MZw4hjU=";
+    })
+  ];
+
   SETUPTOOLS_SCM_PRETEND_VERSION = version;
 
   pythonRelaxDeps = [

pkgs/development/python-modules/dask/default.nix

@@ -134,6 +134,8 @@ buildPythonPackage rec {
     # AttributeError: 'str' object has no attribute 'decode'
     "test_read_dir_nometa"
   ] ++ [
+    # https://github.com/dask/dask/issues/10347#issuecomment-1589683941
+    "test_concat_categorical"
     # AttributeError: 'ArrowStringArray' object has no attribute 'tobytes'. Did you mean: 'nbytes'?
     "test_dot"
     "test_dot_nan"

pkgs/development/python-modules/google-cloud-bigquery/default.nix

@@ -127,6 +127,8 @@ buildPythonPackage rec {
   ];
 
   meta = with lib; {
+    # Not compatible with pyarrow13 yet.
+    broken = true;
     description = "Google BigQuery API client library";
     homepage = "https://github.com/googleapis/python-bigquery";
     changelog = "https://github.com/googleapis/python-bigquery/blob/v${version}/CHANGELOG.md";

pkgs/development/python-modules/ibis-framework/default.nix

@@ -130,6 +130,19 @@ buildPythonPackage rec {
     # tries to download duckdb extensions
     "--deselect=ibis/backends/duckdb/tests/test_register.py::test_register_sqlite"
     "--deselect=ibis/backends/duckdb/tests/test_register.py::test_read_sqlite"
+    # pyarrow13 is not supported yet.
+    "--deselect=ibis/backends/tests/test_temporal.py::test_date_truncate"
+    "--deselect=ibis/backends/tests/test_temporal.py::test_integer_to_interval_timestamp"
+    "--deselect=ibis/backends/tests/test_temporal.py::test_integer_to_interval_timestamp"
+    "--deselect=ibis/backends/tests/test_temporal.py::test_interval_add_cast_column"
+    "--deselect=ibis/backends/tests/test_temporal.py::test_integer_to_interval_timestamp"
+    "--deselect=ibis/backends/tests/test_temporal.py::test_integer_to_interval_timestamp"
+    "--deselect=ibis/backends/tests/test_temporal.py::test_integer_to_interval_timestamp"
+    "--deselect=ibis/backends/tests/test_temporal.py::test_integer_to_interval_timestamp"
+    "--deselect=ibis/backends/tests/test_timecontext.py::test_context_adjustment_filter_before_window"
+    "--deselect=ibis/backends/tests/test_timecontext.py::test_context_adjustment_window_udf"
+    "--deselect=ibis/backends/tests/test_timecontext.py::test_context_adjustment_window_udf"
+    "--deselect=ibis/backends/tests/test_aggregation.py::test_aggregate_grouped"
   ];
 
   # patch out tests that check formatting with black

pkgs/development/python-modules/insightface/default.nix

@@ -69,6 +69,9 @@ buildPythonPackage rec {
   doCheck = false; # Upstream has no tests
 
   meta = with lib; {
+    # Both protobuf3 and protobuf4 in the build closure.
+    # related: https://github.com/onnx/onnx/issues/5563
+    broken = true;
     description = "State-of-the-art 2D and 3D Face Analysis Project";
     homepage = "https://github.com/deepinsight/insightface";
     license = licenses.mit;

pkgs/development/python-modules/intake/default.nix

@@ -107,6 +107,16 @@ buildPythonPackage rec {
     "test_python"
     # Timing-based, flaky on darwin and possibly others
     "TestServerV1Source.test_idle_timer"
+    # arrow-cpp-13 related
+    "test_read"
+    "test_pickle"
+    "test_read_dask"
+    "test_read_list"
+    "test_read_list_with_glob"
+    "test_to_dask"
+    "test_columns"
+    "test_df_transform"
+    "test_pipeline_apply"
   ] ++ lib.optionals (stdenv.isDarwin && lib.versionOlder stdenv.hostPlatform.darwinMinVersion "10.13") [
     # Flaky with older low-res mtime on darwin < 10.13 (#143987)
     "test_second_load_timestamp"

pkgs/development/python-modules/worldengine/default.nix

@@ -68,6 +68,7 @@ buildPythonPackage rec {
   ];
 
   meta = with lib; {
+    broken = true;
     homepage = "https://github.com/mindwerks/worldengine";
     description = "World generator using simulation of plates, rain shadow, erosion, etc";
     license = licenses.mit;