[arrow] Update to 6.0.0 (#21113)

* Bump version to 6.0.0

* Update version files

* Update patch

* Update version files

* Update patch

* Update version files

* Update patch

* Update version files

* Fix deprecated functions

* Update version files

* Fix unused variable warning

* Update version files

* Enclose paths in quotes

Co-authored-by: JonLiu1993 <63675417+JonLiu1993@users.noreply.github.com>

* Enclose paths in quotes

* Update version files

* Fix S3 feature

* Update version files

* Remove quotes from empty DZSTD_MSVC_LIB_PREFIX

* Update version files

* Update version files

* Update version files

* Update version files

* Remove old attempt to fix AWS SDK issue

* Update versions files

Co-authored-by: JonLiu1993 <63675417+JonLiu1993@users.noreply.github.com>

ports/arrow/fix-dependencies.patch

@@ -1,8 +1,8 @@
 diff --git a/cpp/CMakeLists.txt b/cpp/CMakeLists.txt
-index b5a39bf33..af1e16a9b 100644
+index ba8c36e81..eac441dee 100644
 --- a/cpp/CMakeLists.txt
 +++ b/cpp/CMakeLists.txt
-@@ -668,7 +668,7 @@ endif()
+@@ -689,7 +689,7 @@ endif()
  
  if(ARROW_WITH_BROTLI)
    # Order is important for static linking
@@ -11,7 +11,7 @@ index b5a39bf33..af1e16a9b 100644
    list(APPEND ARROW_LINK_LIBS ${ARROW_BROTLI_LIBS})
    list(APPEND ARROW_STATIC_LINK_LIBS ${ARROW_BROTLI_LIBS})
    if(Brotli_SOURCE STREQUAL "SYSTEM")
-@@ -684,9 +684,9 @@ if(ARROW_WITH_BZ2)
+@@ -705,9 +705,9 @@ if(ARROW_WITH_BZ2)
  endif()
  
  if(ARROW_WITH_LZ4)
@@ -23,7 +23,7 @@ index b5a39bf33..af1e16a9b 100644
    endif()
  endif()
  
-@@ -735,10 +735,10 @@ if(ARROW_S3)
+@@ -764,10 +764,10 @@ if(ARROW_S3)
  endif()
  
  if(ARROW_WITH_UTF8PROC)
@@ -38,7 +38,7 @@ index b5a39bf33..af1e16a9b 100644
  endif()
  
 diff --git a/cpp/cmake_modules/ThirdpartyToolchain.cmake b/cpp/cmake_modules/ThirdpartyToolchain.cmake
-index e6852d9c2..fc0973ffa 100644
+index 673a58eed..8c2c1e2fb 100644
 --- a/cpp/cmake_modules/ThirdpartyToolchain.cmake
 +++ b/cpp/cmake_modules/ThirdpartyToolchain.cmake
 @@ -53,7 +53,7 @@ set(ARROW_THIRDPARTY_DEPENDENCIES
@@ -50,7 +50,7 @@ index e6852d9c2..fc0973ffa 100644
      BZip2
      c-ares
      gflags
-@@ -61,14 +61,14 @@ set(ARROW_THIRDPARTY_DEPENDENCIES
+@@ -62,14 +62,14 @@ set(ARROW_THIRDPARTY_DEPENDENCIES
      gRPC
      GTest
      LLVM
@@ -67,7 +67,26 @@ index e6852d9c2..fc0973ffa 100644
      xsimd
      ZLIB
      zstd)
-@@ -945,7 +945,7 @@ macro(build_snappy)
+@@ -872,17 +872,7 @@ set(Boost_ADDITIONAL_VERSIONS
+ # so we first need to determine whether we're building it
+ if(ARROW_WITH_THRIFT AND Thrift_SOURCE STREQUAL "AUTO")
+   find_package(Thrift 0.11.0 MODULE COMPONENTS ${ARROW_THRIFT_REQUIRED_COMPONENTS})
+-  if(Thrift_FOUND)
+-    find_package(PkgConfig QUIET)
+-    pkg_check_modules(THRIFT_PC
+-                      thrift
+-                      NO_CMAKE_PATH
+-                      NO_CMAKE_ENVIRONMENT_PATH
+-                      QUIET)
+-    if(THRIFT_PC_FOUND)
+-      string(APPEND ARROW_PC_REQUIRES_PRIVATE " thrift")
+-    endif()
+-  else()
++  if(NOT Thrift_FOUND AND NOT THRIFT_FOUND)
+     set(Thrift_SOURCE "BUNDLED")
+   endif()
+ endif()
+@@ -991,7 +981,7 @@ macro(build_snappy)
  endmacro()
  
  if(ARROW_WITH_SNAPPY)
@@ -76,7 +95,7 @@ index e6852d9c2..fc0973ffa 100644
    if(${Snappy_SOURCE} STREQUAL "SYSTEM" AND NOT snappy_PC_FOUND)
      get_target_property(SNAPPY_LIB Snappy::snappy IMPORTED_LOCATION)
      string(APPEND ARROW_PC_LIBS_PRIVATE " ${SNAPPY_LIB}")
-@@ -1014,10 +1014,16 @@ macro(build_brotli)
+@@ -1061,10 +1051,16 @@ macro(build_brotli)
  endmacro()
  
  if(ARROW_WITH_BROTLI)
@@ -96,7 +115,7 @@ index e6852d9c2..fc0973ffa 100644
    include_directories(SYSTEM ${BROTLI_INCLUDE_DIR})
  endif()
  
-@@ -1133,7 +1139,7 @@ macro(build_glog)
+@@ -1181,7 +1177,7 @@ macro(build_glog)
  endmacro()
  
  if(ARROW_USE_GLOG)
@@ -105,7 +124,7 @@ index e6852d9c2..fc0973ffa 100644
    # TODO: Don't use global includes but rather target_include_directories
    get_target_property(GLOG_INCLUDE_DIR glog::glog INTERFACE_INCLUDE_DIRECTORIES)
    include_directories(SYSTEM ${GLOG_INCLUDE_DIR})
-@@ -1206,8 +1212,7 @@ endmacro()
+@@ -1255,8 +1251,7 @@ endmacro()
  if(ARROW_NEED_GFLAGS)
    set(ARROW_GFLAGS_REQUIRED_VERSION "2.1.0")
    resolve_dependency(gflags
@@ -115,15 +134,34 @@ index e6852d9c2..fc0973ffa 100644
                       REQUIRED_VERSION
                       ${ARROW_GFLAGS_REQUIRED_VERSION}
                       IS_RUNTIME_DEPENDENCY
-@@ -1309,6 +1314,7 @@ if(ARROW_WITH_THRIFT)
-   if(NOT Thrift_FOUND AND NOT THRIFT_FOUND)
+@@ -1355,9 +1350,10 @@ endmacro()
+ if(ARROW_WITH_THRIFT)
+   # We already may have looked for Thrift earlier, when considering whether
+   # to build Boost, so don't look again if already found.
+-  if(NOT Thrift_FOUND)
++  if(TRUE)
      # Thrift c++ code generated by 0.13 requires 0.11 or greater
      resolve_dependency(Thrift
 +                       USE_CONFIG TRUE
                         REQUIRED_VERSION
                         0.11.0
                         PC_PACKAGE_NAMES
-@@ -1413,6 +1419,7 @@ if(ARROW_WITH_PROTOBUF)
+@@ -1366,6 +1362,14 @@ if(ARROW_WITH_THRIFT)
+   # TODO: Don't use global includes but rather target_include_directories
+   include_directories(SYSTEM ${THRIFT_INCLUDE_DIR})
+ 
++  if(THRIFT_INCLUDE_DIR)
++    file(READ "${THRIFT_INCLUDE_DIR}/config.h" THRIFT_CONFIG_H_CONTENT)
++    string(REGEX MATCH "#define PACKAGE_VERSION \"[0-9.]+\"" THRIFT_VERSION_DEFINITION
++                "${THRIFT_CONFIG_H_CONTENT}")
++    string(REGEX MATCH "[0-9.]+" THRIFT_VERSION "${THRIFT_VERSION_DEFINITION}")
++    set(THRIFT_VERSION "${THRIFT_VERSION}")
++  endif()
++
+   string(REPLACE "." ";" VERSION_LIST ${THRIFT_VERSION})
+   list(GET VERSION_LIST 0 THRIFT_VERSION_MAJOR)
+   list(GET VERSION_LIST 1 THRIFT_VERSION_MINOR)
+@@ -1480,6 +1484,7 @@ if(ARROW_WITH_PROTOBUF)
      set(ARROW_PROTOBUF_REQUIRED_VERSION "2.6.1")
    endif()
    resolve_dependency(Protobuf
@@ -131,7 +169,7 @@ index e6852d9c2..fc0973ffa 100644
                       REQUIRED_VERSION
                       ${ARROW_PROTOBUF_REQUIRED_VERSION}
                       PC_PACKAGE_NAMES
-@@ -1423,6 +1430,10 @@ if(ARROW_WITH_PROTOBUF)
+@@ -1490,6 +1495,10 @@ if(ARROW_WITH_PROTOBUF)
    endif()
  
    # TODO: Don't use global includes but rather target_include_directories
@@ -142,7 +180,7 @@ index e6852d9c2..fc0973ffa 100644
    include_directories(SYSTEM ${PROTOBUF_INCLUDE_DIR})
  
    if(TARGET arrow::protobuf::libprotobuf)
-@@ -1471,12 +1482,18 @@ if(ARROW_WITH_PROTOBUF)
+@@ -1538,12 +1547,18 @@ if(ARROW_WITH_PROTOBUF)
  
    # Log protobuf paths as we often see issues with mixed sources for
    # the libraries and protoc.
@@ -161,7 +199,7 @@ index e6852d9c2..fc0973ffa 100644
    message(STATUS "Found libprotobuf: ${PROTOBUF_LIBRARY}")
    message(STATUS "Found protobuf headers: ${PROTOBUF_INCLUDE_DIR}")
  endif()
-@@ -1889,7 +1906,7 @@ endmacro()
+@@ -1954,7 +1969,7 @@ endmacro()
  if(ARROW_WITH_RAPIDJSON)
    set(ARROW_RAPIDJSON_REQUIRED_VERSION "1.1.0")
    resolve_dependency(RapidJSON
@@ -170,7 +208,7 @@ index e6852d9c2..fc0973ffa 100644
                       TRUE
                       REQUIRED_VERSION
                       ${ARROW_RAPIDJSON_REQUIRED_VERSION}
-@@ -2024,10 +2041,11 @@ macro(build_lz4)
+@@ -2093,10 +2108,11 @@ macro(build_lz4)
  endmacro()
  
  if(ARROW_WITH_LZ4)
@@ -184,7 +222,7 @@ index e6852d9c2..fc0973ffa 100644
    include_directories(SYSTEM ${LZ4_INCLUDE_DIR})
  endif()
  
-@@ -2090,7 +2108,7 @@ endmacro()
+@@ -2160,7 +2176,7 @@ endmacro()
  if(ARROW_WITH_ZSTD)
    # ARROW-13384: ZSTD_minCLevel was added in v1.4.0, required by ARROW-13091
    resolve_dependency(zstd
@@ -193,7 +231,7 @@ index e6852d9c2..fc0973ffa 100644
                       libzstd
                       REQUIRED_VERSION
                       1.4.0)
-@@ -2262,9 +2280,8 @@ macro(build_utf8proc)
+@@ -2343,9 +2359,8 @@ macro(build_utf8proc)
  endmacro()
  
  if(ARROW_WITH_UTF8PROC)
@@ -205,7 +243,7 @@ index e6852d9c2..fc0973ffa 100644
                       PC_PACKAGE_NAMES
                       libutf8proc)
  
-@@ -2272,7 +2289,7 @@ if(ARROW_WITH_UTF8PROC)
+@@ -2353,7 +2368,7 @@ if(ARROW_WITH_UTF8PROC)
  
    # TODO: Don't use global definitions but rather
    # target_compile_definitions or target_link_libraries
@@ -214,7 +252,7 @@ index e6852d9c2..fc0973ffa 100644
                        INTERFACE_COMPILER_DEFINITIONS)
    if(UTF8PROC_COMPILER_DEFINITIONS)
      add_definitions(-D${UTF8PROC_COMPILER_DEFINITIONS})
-@@ -2280,7 +2297,7 @@ if(ARROW_WITH_UTF8PROC)
+@@ -2361,7 +2376,7 @@ if(ARROW_WITH_UTF8PROC)
  
    # TODO: Don't use global includes but rather
    # target_include_directories or target_link_libraries
@@ -223,7 +261,7 @@ index e6852d9c2..fc0973ffa 100644
                        INTERFACE_INCLUDE_DIRECTORIES)
    include_directories(SYSTEM ${UTF8PROC_INCLUDE_DIR})
  endif()
-@@ -2335,7 +2352,7 @@ endmacro()
+@@ -3270,7 +3285,7 @@ endmacro()
  
  macro(build_grpc)
    resolve_dependency(c-ares
@@ -232,7 +270,7 @@ index e6852d9c2..fc0973ffa 100644
                       TRUE
                       PC_PACKAGE_NAMES
                       libcares)
-@@ -2600,7 +2617,7 @@ endmacro()
+@@ -3493,7 +3508,7 @@ endmacro()
  if(ARROW_WITH_GRPC)
    set(ARROW_GRPC_REQUIRED_VERSION "1.17.0")
    resolve_dependency(gRPC
@@ -241,7 +279,7 @@ index e6852d9c2..fc0973ffa 100644
                       TRUE
                       REQUIRED_VERSION
                       ${ARROW_GRPC_REQUIRED_VERSION}
-@@ -2610,6 +2627,10 @@ if(ARROW_WITH_GRPC)
+@@ -3503,6 +3518,10 @@ if(ARROW_WITH_GRPC)
    # TODO: Don't use global includes but rather target_include_directories
    get_target_property(GRPC_INCLUDE_DIR gRPC::grpc++ INTERFACE_INCLUDE_DIRECTORIES)
    include_directories(SYSTEM ${GRPC_INCLUDE_DIR})
@@ -252,6 +290,16 @@ index e6852d9c2..fc0973ffa 100644
  
    if(GRPC_VENDORED)
      set(GRPCPP_PP_INCLUDE TRUE)
+@@ -4019,7 +4038,8 @@ if(ARROW_S3)
+                             s3
+                             transfer
+                             identity-management
+-                            sts)
++                            sts
++                 CONFIG)
+   endif()
+ 
+   # Restore previous value of BUILD_SHARED_LIBS
 diff --git a/cpp/src/arrow/adapters/orc/CMakeLists.txt b/cpp/src/arrow/adapters/orc/CMakeLists.txt
 index ca901b07d..4512a175f 100644
 --- a/cpp/src/arrow/adapters/orc/CMakeLists.txt

ports/arrow/portfile.cmake

@@ -1,25 +1,25 @@
 vcpkg_from_github(
     OUT_SOURCE_PATH SOURCE_PATH
     REPO apache/arrow
-    REF apache-arrow-5.0.0
-    SHA512 68f4377f654423e7ea47c8c0170ddb030d0b020b936ec435854e216392e6515d870287f410d0acd48b36dcfec61be4c4a95794857f1d91c66745f3c6ed748034
+    REF apache-arrow-6.0.0
+    SHA512 a5f89085f06a52c85ed9aadd5deee5988c28ab2abd775bb22b9c26efb99429ad3dead7e1976777c3fa159b8c07b536c04eac43fa81e655f40c08c07ba12f8b8b
     HEAD_REF master
     PATCHES
         all.patch
         fix-dependencies.patch
 )
 
-file(REMOVE ${SOURCE_PATH}/cpp/cmake_modules/Findzstd.cmake
-            ${SOURCE_PATH}/cpp/cmake_modules/FindBrotli.cmake
-            ${SOURCE_PATH}/cpp/cmake_modules/Find-c-aresAlt.cmake
-            ${SOURCE_PATH}/cpp/cmake_modules/FindLz4.cmake
-            ${SOURCE_PATH}/cpp/cmake_modules/FindSnappy.cmake
-            ${SOURCE_PATH}/cpp/cmake_modules/FindThrift.cmake
-            ${SOURCE_PATH}/cpp/cmake_modules/FindGLOG.cmake
-            ${SOURCE_PATH}/cpp/cmake_modules/Findutf8proc.cmake
-            ${SOURCE_PATH}/cpp/cmake_modules/FindRapidJSONAlt.cmake
-            ${SOURCE_PATH}/cpp/cmake_modules/FindgRPCAlt.cmake
-            ${SOURCE_PATH}/cpp/cmake_modules/FindgflagsAlt.cmake
+file(REMOVE "${SOURCE_PATH}/cpp/cmake_modules/Findzstd.cmake"
+            "${SOURCE_PATH}/cpp/cmake_modules/FindBrotli.cmake"
+            "${SOURCE_PATH}/cpp/cmake_modules/Find-c-aresAlt.cmake"
+            "${SOURCE_PATH}/cpp/cmake_modules/FindLz4.cmake"
+            "${SOURCE_PATH}/cpp/cmake_modules/FindSnappy.cmake"
+            "${SOURCE_PATH}/cpp/cmake_modules/FindThrift.cmake"
+            "${SOURCE_PATH}/cpp/cmake_modules/FindGLOG.cmake"
+            "${SOURCE_PATH}/cpp/cmake_modules/Findutf8proc.cmake"
+            "${SOURCE_PATH}/cpp/cmake_modules/FindRapidJSONAlt.cmake"
+            "${SOURCE_PATH}/cpp/cmake_modules/FindgRPCAlt.cmake"
+            "${SOURCE_PATH}/cpp/cmake_modules/FindgflagsAlt.cmake"
 )
 
 vcpkg_check_features(OUT_FEATURE_OPTIONS FEATURE_OPTIONS
@@ -60,11 +60,11 @@ else()
 endif()
 
 vcpkg_cmake_configure(
-    SOURCE_PATH ${SOURCE_PATH}/cpp
-    PREFER_NINJA
+    SOURCE_PATH "${SOURCE_PATH}/cpp"
     OPTIONS
         ${FEATURE_OPTIONS}
         ${MALLOC_OPTIONS}
+        ${S3_OPTIONS}
         -DCMAKE_SYSTEM_PROCESSOR=${VCPKG_TARGET_ARCHITECTURE}
         -DARROW_BUILD_SHARED=${ARROW_BUILD_SHARED}
         -DARROW_BUILD_STATIC=${ARROW_BUILD_STATIC}
@@ -80,24 +80,26 @@ vcpkg_cmake_configure(
         -DARROW_WITH_ZLIB=ON
         -DARROW_WITH_ZSTD=ON
         -DZSTD_MSVC_LIB_PREFIX=
+    MAYBE_UNUSED_VARIABLES
+        ZSTD_MSVC_LIB_PREFIX
 )
 
 vcpkg_cmake_install()
 
 vcpkg_copy_pdbs()
 
-if(EXISTS ${CURRENT_PACKAGES_DIR}/lib/arrow_static.lib)
+if(EXISTS "${CURRENT_PACKAGES_DIR}/lib/arrow_static.lib")
     message(FATAL_ERROR "Installed lib file should be named 'arrow.lib' via patching the upstream build.")
 endif()
 
 vcpkg_cmake_config_fixup(CONFIG_PATH lib/cmake/arrow)
 
-file(REMOVE_RECURSE ${CURRENT_PACKAGES_DIR}/debug/lib/cmake)
-file(REMOVE_RECURSE ${CURRENT_PACKAGES_DIR}/lib/cmake)
+file(REMOVE_RECURSE "${CURRENT_PACKAGES_DIR}/debug/lib/cmake")
+file(REMOVE_RECURSE "${CURRENT_PACKAGES_DIR}/lib/cmake")
 
-file(INSTALL ${SOURCE_PATH}/LICENSE.txt DESTINATION ${CURRENT_PACKAGES_DIR}/share/${PORT} RENAME copyright)
+file(INSTALL "${SOURCE_PATH}/LICENSE.txt" DESTINATION "${CURRENT_PACKAGES_DIR}/share/${PORT}" RENAME copyright)
 
-file(REMOVE_RECURSE ${CURRENT_PACKAGES_DIR}/debug/include)
-file(REMOVE_RECURSE ${CURRENT_PACKAGES_DIR}/debug/share)
+file(REMOVE_RECURSE "${CURRENT_PACKAGES_DIR}/debug/include")
+file(REMOVE_RECURSE "${CURRENT_PACKAGES_DIR}/debug/share")
 
 vcpkg_fixup_pkgconfig()

ports/arrow/vcpkg.json

@@ -1,7 +1,6 @@
 {
   "name": "arrow",
-  "version": "5.0.0",
-  "port-version": 2,
+  "version": "6.0.0",
   "description": "Cross-language development platform for in-memory analytics",
   "homepage": "https://arrow.apache.org",
   "supports": "x64 | (arm64 & !windows)",

versions/a-/arrow.json

@@ -1,5 +1,10 @@
 {
   "versions": [
+    {
+      "git-tree": "14ddd1079e0be1f6122e93f2103be465f893b37e",
+      "version": "6.0.0",
+      "port-version": 0
+    },
     {
       "git-tree": "8a30c7d6553216924aa3a3957ceb87d0fda39592",
       "version": "5.0.0",

versions/baseline.json

@@ -169,8 +169,8 @@
       "port-version": 0
     },
     "arrow": {
-      "baseline": "5.0.0",
-      "port-version": 2
+      "baseline": "6.0.0",
+      "port-version": 0
     },
     "ashes": {
       "baseline": "2021-06-18",