[arrow] Update to 8.0.1, fix usage (#25983)

* Use expressive patch name

* Fix BROTLI_INCLUDE_DIR import

* glog is unused (off by default)

* Cleanup portfile, fix 'arrow[core]'

* Add CI test port

* Update to 8.0.1 [skip actions]

* Declare dependency source for substituted packages

* Trim obsolete patch chunks

* Fix static link libs export

* Use substituted targets directly, for proper export

* Fix snappy lib export to pc file

* Update versions

* Add license for test port

* Add =pass entries for the new "vcpkg-ci-Xxx" port.

Co-authored-by: Billy O'Neal <bion@microsoft.com>

ports/arrow/fix-ThirdPartyToolchain.patch

@@ -1,3 +1,42 @@
+diff --git a/cpp/CMakeLists.txt b/cpp/CMakeLists.txt
+index 7d9c5c7..33afd70 100644
+--- a/cpp/CMakeLists.txt
++++ b/cpp/CMakeLists.txt
+@@ -705,7 +705,7 @@ endif()
+ 
+ if(ARROW_WITH_BROTLI)
+   # Order is important for static linking
+-  set(ARROW_BROTLI_LIBS Brotli::brotlienc Brotli::brotlidec Brotli::brotlicommon)
++  set(ARROW_BROTLI_LIBS unofficial::brotli::brotlienc unofficial::brotli::brotlidec unofficial::brotli::brotlicommon)
+   list(APPEND ARROW_LINK_LIBS ${ARROW_BROTLI_LIBS})
+   list(APPEND ARROW_STATIC_LINK_LIBS ${ARROW_BROTLI_LIBS})
+   if(Brotli_SOURCE STREQUAL "SYSTEM")
+@@ -721,9 +721,9 @@ if(ARROW_WITH_BZ2)
+ endif()
+ 
+ if(ARROW_WITH_LZ4)
+-  list(APPEND ARROW_STATIC_LINK_LIBS LZ4::lz4)
++  list(APPEND ARROW_STATIC_LINK_LIBS lz4::lz4)
+   if(Lz4_SOURCE STREQUAL "SYSTEM")
+-    list(APPEND ARROW_STATIC_INSTALL_INTERFACE_LIBS LZ4::lz4)
++    list(APPEND ARROW_STATIC_INSTALL_INTERFACE_LIBS lz4::lz4)
+   endif()
+ endif()
+ 
+@@ -793,10 +793,10 @@ if(ARROW_WITH_OPENTELEMETRY)
+ endif()
+ 
+ if(ARROW_WITH_UTF8PROC)
+-  list(APPEND ARROW_LINK_LIBS utf8proc::utf8proc)
+-  list(APPEND ARROW_STATIC_LINK_LIBS utf8proc::utf8proc)
++  list(APPEND ARROW_LINK_LIBS utf8proc)
++  list(APPEND ARROW_STATIC_LINK_LIBS utf8proc)
+   if(utf8proc_SOURCE STREQUAL "SYSTEM")
+-    list(APPEND ARROW_STATIC_INSTALL_INTERFACE_LIBS utf8proc::utf8proc)
++    list(APPEND ARROW_STATIC_INSTALL_INTERFACE_LIBS utf8proc)
+   endif()
+ endif()
+ 
 diff --git a/cpp/cmake_modules/ThirdpartyToolchain.cmake b/cpp/cmake_modules/ThirdpartyToolchain.cmake
 index f070323..290b5a7 100644
 --- a/cpp/cmake_modules/ThirdpartyToolchain.cmake
@@ -48,33 +87,45 @@ index f070323..290b5a7 100644
    else()
      set(PACKAGE_NAME ${DEPENDENCY_NAME})
    endif()
-@@ -1129,10 +1132,16 @@ macro(build_brotli)
+@@ -1061,7 +1064,21 @@ endmacro()
+ if(ARROW_WITH_SNAPPY)
+   resolve_dependency(Snappy PC_PACKAGE_NAMES snappy)
+   if(${Snappy_SOURCE} STREQUAL "SYSTEM" AND NOT snappy_PC_FOUND)
++    get_target_property(SNAPPY_IMPLIB_DEBUG Snappy::snappy IMPORTED_IMPLIB_DEBUG)
++    get_target_property(SNAPPY_LIB_DEBUG Snappy::snappy IMPORTED_LOCATION_DEBUG)
++    get_target_property(SNAPPY_IMPLIB_RELEASE Snappy::snappy IMPORTED_IMPLIB_RELEASE)
++    get_target_property(SNAPPY_LIB_RELEASE Snappy::snappy IMPORTED_LOCATION_RELEASE)
++    if(CMAKE_BUILD_TYPE STREQUAL "DEBUG" AND SNAPPY_IMPLIB_DEBUG)
++      set(SNAPPY_LIB "${SNAPPY_IMPLIB_DEBUG}")
++    elseif(CMAKE_BUILD_TYPE STREQUAL "DEBUG" AND SNAPPY_LIB_DEBUG)
++      set(SNAPPY_LIB "${SNAPPY_LIB_DEBUG}")
++    elseif(CMAKE_BUILD_TYPE STREQUAL "RELEASE" AND SNAPPY_IMPLIB_RELEASE)
++      set(SNAPPY_LIB "${SNAPPY_IMPLIB_RELEASE}")
++    elseif(CMAKE_BUILD_TYPE STREQUAL "RELEASE" AND SNAPPY_LIB_RELEASE)
++      set(SNAPPY_LIB "${SNAPPY_LIB_RELEASE}")
++    else()
+     get_target_property(SNAPPY_LIB Snappy::snappy IMPORTED_LOCATION)
++    endif()
+     string(APPEND ARROW_PC_LIBS_PRIVATE " ${SNAPPY_LIB}")
+   endif()
+   # TODO: Don't use global includes but rather target_include_directories
+@@ -1129,10 +1132,14 @@ macro(build_brotli)
  endmacro()
  
  if(ARROW_WITH_BROTLI)
 -  resolve_dependency(Brotli PC_PACKAGE_NAMES libbrotlidec libbrotlienc)
++  set(Brotli_SOURCE SYSTEM)
 +  resolve_dependency(unofficial-brotli USE_CONFIG TRUE PC_PACKAGE_NAMES libbrotlidec libbrotlienc)
-+  add_library(Brotli::brotlicommon ALIAS unofficial::brotli::brotlicommon)
-+  add_library(Brotli::brotlienc ALIAS unofficial::brotli::brotlienc)
-+  add_library(Brotli::brotlidec ALIAS unofficial::brotli::brotlidec)
-+  get_target_property(BROTLI_INCLUDE_DIR unofficial::brotli::brotlicommon-static INTERFACE_INCLUDE_DIRECTORIES)
    # TODO: Don't use global includes but rather target_include_directories
-+  if(NOT BROTLI_INCLUDE_DIR)
-   get_target_property(BROTLI_INCLUDE_DIR Brotli::brotlicommon
+-  get_target_property(BROTLI_INCLUDE_DIR Brotli::brotlicommon
++  get_target_property(BROTLI_INCLUDE_DIR unofficial::brotli::brotlicommon
                        INTERFACE_INCLUDE_DIRECTORIES)
++  if(NOT BROTLI_INCLUDE_DIR)
++    get_target_property(BROTLI_INCLUDE_DIR unofficial::brotli::brotlicommon-static INTERFACE_INCLUDE_DIRECTORIES)
 +  endif()
    include_directories(SYSTEM ${BROTLI_INCLUDE_DIR})
  endif()
  
-@@ -1249,7 +1261,7 @@ macro(build_glog)
- endmacro()
- 
- if(ARROW_USE_GLOG)
--  resolve_dependency(GLOG PC_PACKAGE_NAMES libglog)
-+  resolve_dependency(glog USE_CONFIG TRUE PC_PACKAGE_NAMES libglog)
-   # 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})
 @@ -1434,6 +1446,13 @@ if(ARROW_WITH_THRIFT)
    # TODO: Don't use global includes but rather target_include_directories
    include_directories(SYSTEM ${THRIFT_INCLUDE_DIR})
@@ -98,40 +149,51 @@ index f070323..290b5a7 100644
  
    if(NOT Protobuf_USE_STATIC_LIBS AND MSVC_TOOLCHAIN)
      add_definitions(-DPROTOBUF_USE_DLLS)
-@@ -2121,8 +2142,6 @@ endmacro()
- if(ARROW_WITH_RAPIDJSON)
-   set(ARROW_RAPIDJSON_REQUIRED_VERSION "1.1.0")
-   resolve_dependency(RapidJSON
--                     HAVE_ALT
--                     TRUE
-                      REQUIRED_VERSION
-                      ${ARROW_RAPIDJSON_REQUIRED_VERSION}
-                      IS_RUNTIME_DEPENDENCY
-@@ -2260,7 +2279,8 @@ macro(build_lz4)
+@@ -2260,10 +2276,11 @@ macro(build_lz4)
  endmacro()
  
  if(ARROW_WITH_LZ4)
 -  resolve_dependency(Lz4 PC_PACKAGE_NAMES liblz4)
++  set(Lz4_SOURCE SYSTEM)
 +  resolve_dependency(lz4 USE_CONFIG TRUE PC_PACKAGE_NAMES liblz4)
-+  add_library(LZ4::lz4 ALIAS lz4::lz4)
  
    # TODO: Don't use global includes but rather target_include_directories
-   get_target_property(LZ4_INCLUDE_DIR LZ4::lz4 INTERFACE_INCLUDE_DIRECTORIES)
-@@ -2517,11 +2537,10 @@ macro(build_utf8proc)
+-  get_target_property(LZ4_INCLUDE_DIR LZ4::lz4 INTERFACE_INCLUDE_DIRECTORIES)
++  get_target_property(LZ4_INCLUDE_DIR lz4::lz4 INTERFACE_INCLUDE_DIRECTORIES)
+   include_directories(SYSTEM ${LZ4_INCLUDE_DIR})
+ endif()
+ 
+@@ -2517,9 +2534,8 @@ macro(build_utf8proc)
  endmacro()
  
  if(ARROW_WITH_UTF8PROC)
 -  resolve_dependency(utf8proc
 -                     REQUIRED_VERSION
 -                     "2.2.0"
++  set(utf8proc_SOURCE SYSTEM)
 +  resolve_dependency(unofficial-utf8proc
                       PC_PACKAGE_NAMES
                       libutf8proc)
-+  add_library(utf8proc::utf8proc ALIAS utf8proc)
  
-   add_definitions(-DARROW_WITH_UTF8PROC)
+@@ -2527,7 +2543,7 @@ if(ARROW_WITH_UTF8PROC)
  
-@@ -3719,6 +3738,9 @@ if(ARROW_WITH_GRPC)
+   # TODO: Don't use global definitions but rather
+   # target_compile_definitions or target_link_libraries
+-  get_target_property(UTF8PROC_COMPILER_DEFINITIONS utf8proc::utf8proc
++  get_target_property(UTF8PROC_COMPILER_DEFINITIONS utf8proc
+                       INTERFACE_COMPILER_DEFINITIONS)
+   if(UTF8PROC_COMPILER_DEFINITIONS)
+     add_definitions(-D${UTF8PROC_COMPILER_DEFINITIONS})
+@@ -2535,7 +2551,7 @@ if(ARROW_WITH_UTF8PROC)
+ 
+   # TODO: Don't use global includes but rather
+   # target_include_directories or target_link_libraries
+-  get_target_property(UTF8PROC_INCLUDE_DIR utf8proc::utf8proc
++  get_target_property(UTF8PROC_INCLUDE_DIR utf8proc
+                       INTERFACE_INCLUDE_DIRECTORIES)
+   include_directories(SYSTEM ${UTF8PROC_INCLUDE_DIR})
+ endif()
+@@ -3719,6 +3735,9 @@ 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})
@@ -141,3 +203,16 @@ index f070323..290b5a7 100644
  
    if(GRPC_VENDORED)
      set(GRPCPP_PP_INCLUDE TRUE)
+diff --git a/cpp/src/arrow/adapters/orc/CMakeLists.txt b/cpp/src/arrow/adapters/orc/CMakeLists.txt
+index b1b6847..444a45e 100644
+--- a/cpp/src/arrow/adapters/orc/CMakeLists.txt
++++ b/cpp/src/arrow/adapters/orc/CMakeLists.txt
+@@ -30,7 +30,7 @@ set(ORC_MIN_TEST_LIBS
+     GTest::gtest_main
+     GTest::gtest
+     Snappy::snappy
+-    LZ4::lz4
++    lz4::lz4
+     ZLIB::ZLIB)
+ 
+ if(ARROW_BUILD_STATIC)

ports/arrow/portfile.cmake

@@ -1,13 +1,14 @@
 vcpkg_from_github(
     OUT_SOURCE_PATH SOURCE_PATH
     REPO apache/arrow
-    REF apache-arrow-8.0.0
-    SHA512 08b6ab4a3c5e0dd9c46402da8e7b9ef9f918eea177413cb31695192dfdb5a472ebbfef255b8343fe775d81e8b5eb268c3428a699fac48b36bf808f5b81e83a64
+    REF apache-arrow-8.0.1
+    SHA512 b32f5a3666de7d6d16ea828697bb42e1b6605f58719c42e670c9ec0a8782057dac933f6e14e97b46f82802fc38cc7f4cc825794a4a95ac641593c2ee26ac5bbe
     HEAD_REF master
     PATCHES
         vs-2022-fixes.patch
-        all.patch
+        msvc-static-name.patch
         fix-ThirdPartyToolchain.patch
+        static-link-libs.patch # https://github.com/apache/arrow/pull/13707
 )
 file(REMOVE "${SOURCE_PATH}/cpp/cmake_modules/Findzstd.cmake"
             "${SOURCE_PATH}/cpp/cmake_modules/FindBrotli.cmake"
@@ -87,32 +88,39 @@ vcpkg_cmake_configure(
 )
 
 vcpkg_cmake_install()
-
 vcpkg_copy_pdbs()
 
+vcpkg_fixup_pkgconfig()
+
 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")
-
-configure_file(${CMAKE_CURRENT_LIST_DIR}/vcpkg-cmake-wrapper.cmake ${CURRENT_PACKAGES_DIR}/share/${PORT} @ONLY)
 file(INSTALL "${CMAKE_CURRENT_LIST_DIR}/usage" DESTINATION "${CURRENT_PACKAGES_DIR}/share/${PORT}")
-file(INSTALL "${SOURCE_PATH}/LICENSE.txt" DESTINATION "${CURRENT_PACKAGES_DIR}/share/${PORT}" RENAME copyright)
+if("parquet" IN_LIST FEATURES)
+    file(GLOB PARQUET_FILES "${CURRENT_PACKAGES_DIR}/share/${PORT}/Parquet*")
+    file(COPY ${PARQUET_FILES} DESTINATION "${CURRENT_PACKAGES_DIR}/share/parquet")
+    file(REMOVE_RECURSE ${PARQUET_FILES})
+    file(RENAME "${CURRENT_PACKAGES_DIR}/share/${PORT}/FindParquet.cmake" "${CURRENT_PACKAGES_DIR}/share/parquet/FindParquet.cmake")
+    file(READ "${CMAKE_CURRENT_LIST_DIR}/usage-parquet" usage-parquet)
+    file(APPEND "${CURRENT_PACKAGES_DIR}/share/${PORT}/usage" "${usage-parquet}")
+else()
+    file(REMOVE "${CURRENT_PACKAGES_DIR}/share/${PORT}/FindParquet.cmake")
+endif()
 
-file(GLOB PARQUET_FILES ${CURRENT_PACKAGES_DIR}/share/${PORT}/Parquet*)
-file(COPY ${PARQUET_FILES} DESTINATION "${CURRENT_PACKAGES_DIR}/share/parquet")
-file(REMOVE_RECURSE ${PARQUET_FILES})
-file(RENAME ${CURRENT_PACKAGES_DIR}/share/${PORT}/FindParquet.cmake ${CURRENT_PACKAGES_DIR}/share/parquet/FindParquet.cmake)
-
-file(REMOVE_RECURSE "${CURRENT_PACKAGES_DIR}/debug/include")
-file(REMOVE_RECURSE "${CURRENT_PACKAGES_DIR}/debug/share")
+if("example" IN_LIST FEATURES)
+    file(INSTALL "${SOURCE_PATH}/cpp/examples/minimal_build/" DESTINATION "${CURRENT_PACKAGES_DIR}/share/${PORT}/example")
+endif()
 
 if ("plasma" IN_LIST FEATURES)
     vcpkg_copy_tools(TOOL_NAMES plasma-store-server AUTO_CLEAN)
 endif ()
 
-vcpkg_fixup_pkgconfig()
+file(REMOVE_RECURSE "${CURRENT_PACKAGES_DIR}/debug/include")
+file(REMOVE_RECURSE "${CURRENT_PACKAGES_DIR}/debug/share")
+file(REMOVE_RECURSE "${CURRENT_PACKAGES_DIR}/share/doc")
+
+configure_file("${CMAKE_CURRENT_LIST_DIR}/vcpkg-cmake-wrapper.cmake" "${CURRENT_PACKAGES_DIR}/share/${PORT}" @ONLY)
+vcpkg_install_copyright(FILE_LIST "${SOURCE_PATH}/LICENSE.txt" "${SOURCE_PATH}/NOTICE.txt")

ports/arrow/static-link-libs.patch

@@ -0,0 +1,12 @@
+diff --git a/cpp/src/arrow/CMakeLists.txt b/cpp/src/arrow/CMakeLists.txt
+index 690c51a..2f7d26b 100644
+--- a/cpp/src/arrow/CMakeLists.txt
++++ b/cpp/src/arrow/CMakeLists.txt
+@@ -570,6 +570,7 @@ add_arrow_lib(arrow
+               ${ARROW_SHARED_PRIVATE_LINK_LIBS}
+               STATIC_LINK_LIBS
+               ${ARROW_STATIC_LINK_LIBS}
++              STATIC_INSTALL_INTERFACE_LIBS
+               ${ARROW_STATIC_INSTALL_INTERFACE_LIBS}
+               SHARED_INSTALL_INTERFACE_LIBS
+               ${ARROW_SHARED_INSTALL_INTERFACE_LIBS})

ports/arrow/usage

@@ -2,6 +2,3 @@ The package arrow provides CMake targets:
 
     find_package(Arrow CONFIG REQUIRED)
     target_link_libraries(main PRIVATE ${ARROW_LIBRARIES})
-
-    find_package(Parquet CONFIG REQUIRED)
-    target_link_libraries(main PRIVATE ${PARQUET_LIBRARIES})

ports/arrow/usage-parquet

@@ -0,0 +1,3 @@
+
+    find_package(Parquet CONFIG REQUIRED)
+    target_link_libraries(main PRIVATE ${PARQUET_LIBRARIES})

ports/arrow/vcpkg.json

@@ -1,6 +1,6 @@
 {
   "name": "arrow",
-  "version": "8.0.0",
+  "version": "8.0.1",
   "description": "Cross-language development platform for in-memory analytics",
   "homepage": "https://arrow.apache.org",
   "license": "Apache-2.0",
@@ -12,7 +12,6 @@
     "brotli",
     "bzip2",
     "gflags",
-    "glog",
     "lz4",
     "openssl",
     "re2",
@@ -49,6 +48,9 @@
     "dataset": {
       "description": "Dataset support"
     },
+    "example": {
+      "description": "Install the minimal example (source code)"
+    },
     "filesystem": {
       "description": "Filesystem support"
     },

scripts/ci.baseline.txt

@@ -1253,6 +1253,11 @@ cmake-user:x64-windows-static-md=pass
 cmake-user:x64-windows-static=pass
 cmake-user:x64-windows=pass
 cmake-user:x86-windows=pass
+vcpkg-ci-arrow:x64-windows=pass
+vcpkg-ci-arrow:x64-windows-static=pass
+vcpkg-ci-arrow:x64-windows-static-md=pass
+vcpkg-ci-arrow:x64-osx=pass
+vcpkg-ci-arrow:x64-linux=pass
 vcpkg-ci-boost:x64-linux=pass
 vcpkg-ci-boost:x64-windows-static-md=pass
 vcpkg-ci-boost:x64-windows-static=pass

scripts/test_ports/vcpkg-ci-arrow/portfile.cmake

@@ -0,0 +1,10 @@
+set(VCPKG_POLICY_EMPTY_PACKAGE enabled)
+
+string(COMPARE EQUAL "${VCPKG_LIBRARY_LINKAGE}" "dynamic" ARROW_LINK_SHARED)
+
+vcpkg_cmake_configure(
+    SOURCE_PATH "${CURRENT_INSTALLED_DIR}/share/arrow/example"
+    OPTIONS
+        -DARROW_LINK_SHARED=${ARROW_LINK_SHARED}
+)
+vcpkg_cmake_build()

scripts/test_ports/vcpkg-ci-arrow/vcpkg.json

@@ -0,0 +1,20 @@
+{
+  "name": "vcpkg-ci-arrow",
+  "version-date": "2022-07-25",
+  "description": "Builds an example app in order to validate the arrow port.",
+  "license": "Apache-2.0",
+  "dependencies": [
+    {
+      "name": "vcpkg-cmake",
+      "host": true
+    },
+    {
+      "name": "arrow",
+      "default-features": false,
+      "features": [
+        "csv",
+        "example"
+      ]
+    }
+  ]
+}

versions/a-/arrow.json

@@ -1,5 +1,10 @@
 {
   "versions": [
+    {
+      "git-tree": "6935bb6b1a309931f2f85bd8da334b9777d51a4a",
+      "version": "8.0.1",
+      "port-version": 0
+    },
     {
       "git-tree": "4334074d163357f9002145ca722dc8dbcc546dbc",
       "version": "8.0.0",

versions/baseline.json

@@ -177,7 +177,7 @@
       "port-version": 0
     },
     "arrow": {
-      "baseline": "8.0.0",
+      "baseline": "8.0.1",
       "port-version": 0
     },
     "ashes": {