GH-49948: [CI][C++] Revert PR 49931 (Pin MinGW MSYS2 packages) but keep bumped minIO version

.github/workflows/cpp.yml

@@ -366,40 +366,6 @@ jobs:
       - name: Setup MSYS2
         shell: msys2 {0}
         run: ci/scripts/msys2_setup.sh cpp
-      - name: Pin MSYS2 packages
-        # Temporary workaround for #49930: gcc 16 surfaces a cluster of 5
-        # MINGW64 test failures. Pinning gcc-libs to 15.2 (and C++ packages
-        # rebuilt against gcc-libs 16.1, for ABI compatibility) avoids all
-        # of them. #49272/#49462 cover one (arrow-json-test); the other 4
-        # (async-utility-test, threading-utility-test, dataset-writer-test
-        # `bad_weak_ptr`, dataset-file-test) need separate fixes — see #49930.
-        # Remove once all 5 pass on current upstream MSYS2 without these pins
-        if: matrix.msystem_upper == 'MINGW64'
-        shell: msys2 {0}
-        run: |
-          set -ex
-          base="https://repo.msys2.org/mingw/mingw64"
-          urls=(
-            "$base/mingw-w64-x86_64-gcc-libs-15.2.0-14-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-gcc-15.2.0-14-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-aws-crt-cpp-0.38.4-1-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-boost-libs-1.91.0-1-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-boost-1.91.0-1-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-ccache-4.13.2-1-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-clang-libs-22.1.4-1-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-clang-22.1.4-1-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-cmake-4.3.2-2-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-icu-78.3-1-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-llvm-libs-22.1.4-1-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-llvm-tools-22.1.4-1-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-llvm-22.1.4-1-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-tbb-2022.3.0-1-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-thrift-0.22.0-1-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-zstd-1.5.7-1-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-gflags-2.2.2-7-any.pkg.tar.zst"
-            "$base/mingw-w64-x86_64-aws-sdk-cpp-1.11.479-1-any.pkg.tar.zst"
-          )
-          pacman -U --noconfirm "${urls[@]}"
       - name: Cache ccache
         uses: actions/cache@v5
         with:

cpp/src/arrow/dataset/CMakeLists.txt

@@ -181,6 +181,17 @@ add_arrow_dataset_test(scanner_test)
 add_arrow_dataset_test(subtree_test)
 add_arrow_dataset_test(write_node_test)
 
+# Diagnostic for issue GH-49958: build two tests as C++23 with libstdc++ backend
+# so the terminate handler prints stack
+if(MINGW
+   AND CMAKE_CXX_COMPILER_ID STREQUAL "GNU"
+   AND CMAKE_CXX_COMPILER_VERSION VERSION_GREATER_EQUAL 14)
+  set_target_properties(arrow-dataset-dataset-writer-test arrow-dataset-file-test
+                        PROPERTIES CXX_STANDARD 23)
+  target_link_libraries(arrow-dataset-dataset-writer-test PRIVATE stdc++exp)
+  target_link_libraries(arrow-dataset-file-test PRIVATE stdc++exp)
+endif()
+
 if(ARROW_CSV)
   add_arrow_dataset_test(file_csv_test)
 endif()

cpp/src/arrow/dataset/dataset_writer_test.cc

@@ -18,10 +18,20 @@
 #include "arrow/dataset/dataset_writer.h"
 
 #include <chrono>
+#include <cstdio>
+#include <cstdlib>
+#include <exception>
 #include <mutex>
 #include <optional>
+#include <sstream>
 #include <thread>
+#include <typeinfo>
 #include <vector>
+#include <version>
+#if defined(__MINGW32__) && defined(__cpp_lib_stacktrace)
+#  include <stacktrace>
+#  define ARROW_TEST_HAVE_STACKTRACE 1
+#endif
 
 #include "arrow/array/builder_primitive.h"
 #include "arrow/dataset/file_ipc.h"
@@ -37,6 +47,48 @@
 
 using namespace std::string_view_literals;  // NOLINT
 
+namespace {
+[[noreturn]] void ArrowTerminateHandler() noexcept {
+  std::ostringstream tid;
+  tid << std::this_thread::get_id();
+  std::fprintf(stderr, "\n=== std::terminate on thread %s ===\n", tid.str().c_str());
+  if (auto eptr = std::current_exception()) {
+    try {
+      std::rethrow_exception(eptr);
+    } catch (const std::exception& e) {
+      std::fprintf(stderr, "  type:   %s\n  what(): %s\n", typeid(e).name(), e.what());
+    } catch (...) {
+      std::fprintf(stderr, "  non-std exception\n");
+    }
+  } else {
+    std::fprintf(stderr, "  (no in-flight exception)\n");
+  }
+  std::fflush(stderr);
+
+#ifdef ARROW_TEST_HAVE_STACKTRACE
+  std::fprintf(stderr, "Stack trace:\n");
+  std::fflush(stderr);
+  try {
+    for (const auto& frame : std::stacktrace::current(/*skip=*/1)) {
+      std::fprintf(stderr, "  %s\n", std::to_string(frame).c_str());
+      std::fflush(
+          stderr);  // per-frame flush so a mid-trace crash still leaves frames 0..N-1
+    }
+  } catch (...) {
+    std::fprintf(stderr, "  <stacktrace unavailable>\n");
+  }
+#endif
+
+  std::fflush(stderr);
+  std::_Exit(134);  // do NOT abort() on MinGW — it can pop a WER dialog
+}
+
+[[maybe_unused]] const auto kInstallTerminateHandler = [] {
+  std::set_terminate(ArrowTerminateHandler);
+  return 0;
+}();
+}  // namespace
+
 namespace arrow {
 namespace dataset {
 namespace internal {

cpp/src/arrow/dataset/file_test.cc

@@ -18,11 +18,21 @@
 #include <arrow/compute/function.h>
 #include <arrow/compute/registry.h>
 #include <cstdint>
+#include <cstdio>
+#include <cstdlib>
+#include <exception>
 #include <memory>
+#include <sstream>
 #include <string>
 #include <thread>
 #include <tuple>
+#include <typeinfo>
 #include <vector>
+#include <version>
+#if defined(__MINGW32__) && defined(__cpp_lib_stacktrace)
+#  include <stacktrace>
+#  define ARROW_TEST_HAVE_STACKTRACE 1
+#endif
 
 #include <gmock/gmock.h>
 #include <gtest/gtest.h>
@@ -49,6 +59,48 @@
 
 namespace cp = arrow::compute;
 
+namespace {
+[[noreturn]] void ArrowTerminateHandler() noexcept {
+  std::ostringstream tid;
+  tid << std::this_thread::get_id();
+  std::fprintf(stderr, "\n=== std::terminate on thread %s ===\n", tid.str().c_str());
+  if (auto eptr = std::current_exception()) {
+    try {
+      std::rethrow_exception(eptr);
+    } catch (const std::exception& e) {
+      std::fprintf(stderr, "  type:   %s\n  what(): %s\n", typeid(e).name(), e.what());
+    } catch (...) {
+      std::fprintf(stderr, "  non-std exception\n");
+    }
+  } else {
+    std::fprintf(stderr, "  (no in-flight exception)\n");
+  }
+  std::fflush(stderr);
+
+#ifdef ARROW_TEST_HAVE_STACKTRACE
+  std::fprintf(stderr, "Stack trace:\n");
+  std::fflush(stderr);
+  try {
+    for (const auto& frame : std::stacktrace::current(/*skip=*/1)) {
+      std::fprintf(stderr, "  %s\n", std::to_string(frame).c_str());
+      std::fflush(
+          stderr);  // per-frame flush so a mid-trace crash still leaves frames 0..N-1
+    }
+  } catch (...) {
+    std::fprintf(stderr, "  <stacktrace unavailable>\n");
+  }
+#endif
+
+  std::fflush(stderr);
+  std::_Exit(134);  // do NOT abort() on MinGW — it can pop a WER dialog
+}
+
+[[maybe_unused]] const auto kInstallTerminateHandler = [] {
+  std::set_terminate(ArrowTerminateHandler);
+  return 0;
+}();
+}  // namespace
+
 namespace arrow {
 
 using compute::ExecBatchFromJSON;