feat: Add fedora runs to CI (facebookincubator#14575)

Summary:
This PR adds a fedora build to our default set of linux builds. This should ensure that we continue to build and work on Fedora in the future.


Differential Revision: D81728281

Pulled By: kgpai

Author: kgpai

.github/workflows/linux-build-base.yml

@@ -36,7 +36,6 @@ jobs:
     env:
       CCACHE_DIR: ${{ github.workspace }}/ccache
       VELOX_DEPENDENCY_SOURCE: SYSTEM
-      GTest_SOURCE: BUNDLED
       cudf_SOURCE: BUNDLED
       CUDA_VERSION: '12.8'
       faiss_SOURCE: BUNDLED
@@ -206,3 +205,62 @@ jobs:
       - name: Run Tests
         run: |
           cd _build/debug && ctest -j 8 --output-on-failure --no-tests=error
+
+  fedora-debug:
+    runs-on: 8-core-ubuntu-22.04
+    container: ghcr.io/facebookincubator/velox-dev:fedora
+    # prevent errors when forks ff their main branch
+    if: ${{ github.repository == 'facebookincubator/velox' }}
+    name: Fedora debug
+    env:
+      CCACHE_DIR: ${{ github.workspace }}/ccache
+    defaults:
+      run:
+        shell: bash
+        working-directory: velox
+    steps:
+      - name: Get Ccache Stash
+        uses: apache/infrastructure-actions/stash/restore@3354c1565d4b0e335b78a76aedd82153a9e144d4
+        with:
+          path: ${{ env.CCACHE_DIR }}
+          key: ccache-fedora-debug-default-gcc
+
+      - name: Ensure Stash Dirs Exists
+        working-directory: ${{ github.workspace }}
+        run: |
+          mkdir -p "$CCACHE_DIR"
+
+      - uses: actions/checkout@v4
+        with:
+          path: velox
+          persist-credentials: false
+
+      - name: Clear CCache Statistics
+        run: |
+          ccache -sz
+
+      - name: Make Debug Build
+        env:
+          VELOX_DEPENDENCY_SOURCE: SYSTEM
+          faiss_SOURCE: BUNDLED
+          fmt_SOURCE: BUNDLED
+          gRPC_SOURCE: SYSTEM
+          MAKEFLAGS: NUM_THREADS=8 MAX_HIGH_MEM_JOBS=4 MAX_LINK_JOBS=3
+          EXTRA_CMAKE_FLAGS: >-
+            -DVELOX_ENABLE_PARQUET=ON
+            -DARROW_THRIFT_USE_SHARED=ON
+            -DVELOX_ENABLE_EXAMPLES=ON
+        run: |
+          uv tool install --force cmake@3.31.1
+          dnf install -y -q --setopt=install_weak_deps=False grpc-devel grpc-plugins
+          export EXTRA_CMAKE_FLAGS="${EXTRA_CMAKE_FLAGS} -DVELOX_ENABLE_FAISS=ON"
+          make debug
+
+      - name: CCache after
+        run: |
+          ccache -vs
+
+      - uses: apache/infrastructure-actions/stash/save@3354c1565d4b0e335b78a76aedd82153a9e144d4
+        with:
+          path: ${{ env.CCACHE_DIR }}
+          key: ccache-fedora-debug-default-gcc

CMakeLists.txt

@@ -445,7 +445,8 @@ if(ENABLE_ALL_WARNINGS)
     endif()
     if(CMAKE_CXX_COMPILER_VERSION VERSION_GREATER_EQUAL "14.0.0")
       string(APPEND KNOWN_COMPILER_SPECIFIC_WARNINGS " -Wno-error=template-id-cdtor")
-      string(APPEND KNOWN_COMPILER_SPECIFIC_WARNINGS " -Wno-error=overloaded-virtual")
+      string(APPEND KNOWN_COMPILER_SPECIFIC_WARNINGS " -Wno-overloaded-virtual")
+      string(APPEND KNOWN_COMPILER_SPECIFIC_WARNINGS " -Wno-error=tautological-compare")
     endif()
   endif()
 

scripts/setup-centos9.sh

@@ -60,7 +60,7 @@ function install_build_prerequisites {
     ninja-build python3-pip python3-devel wget which
 
   install_uv
-  uv_install cmake
+  uv_install cmake@3.31.1
 
   if [[ ${USE_CLANG} != "false" ]]; then
     install_clang15

scripts/setup-fedora.sh

@@ -55,6 +55,35 @@ function install_build_prerequisites {
   fi
 }
 
+function install_velox_deps_from_dnf {
+  dnf_install \
+    bison boost-devel c-ares-devel curl-devel double-conversion-devel \
+    elfutils-libelf-devel flex fmt-devel gflags-devel glog-devel gmock-devel \
+    gtest-devel libdwarf-devel libevent-devel libicu-devel \
+    libsodium-devel libzstd-devel lz4-devel openssl-devel-engine \
+    re2-devel snappy-devel thrift-devel xxhash-devel zlib-devel grpc-devel grpc-plugins
+
+  install_faiss_deps
+}
+
+function install_velox_deps {
+  run_and_time install_velox_deps_from_dnf
+  run_and_time install_gcs-sdk-cpp #grpc, abseil, protobuf
+  run_and_time install_fast_float
+  run_and_time install_folly
+  run_and_time install_fizz
+  run_and_time install_wangle
+  run_and_time install_mvfst
+  run_and_time install_fbthrift
+  run_and_time install_duckdb
+  run_and_time install_stemmer
+  run_and_time install_arrow
+  run_and_time install_xsimd    # to new in fedora repos
+  run_and_time install_simdjson # to new in fedora repos
+  run_and_time install_geos     # to new in fedora repos
+  run_and_time install_faiss
+}
+
 (return 2>/dev/null) && return # If script was sourced, don't run commands.
 
 (
@@ -90,6 +119,8 @@ function install_build_prerequisites {
       set -u
     fi
     install_velox_deps
+    # BUILD_TESTING requires grpc
+    dnf_install grpc
     echo "All dependencies for Velox installed!"
     if [[ ${USE_CLANG} != "false" ]]; then
       echo "To use clang for the Velox build set the CC and CXX environment variables in your session."

velox/connectors/tpcds/TpcdsConnectorSplit.h

@@ -53,8 +53,8 @@ template <>
 struct fmt::formatter<facebook::velox::connector::tpcds::TpcdsConnectorSplit>
     : formatter<std::string> {
   auto format(
-      facebook::velox::connector::tpcds::TpcdsConnectorSplit s,
-      format_context& ctx) {
+      facebook::velox::connector::tpcds::TpcdsConnectorSplit const& s,
+      format_context& ctx) const {
     return formatter<std::string>::format(s.toString(), ctx);
   }
 };
@@ -64,8 +64,9 @@ struct fmt::formatter<
     std::shared_ptr<facebook::velox::connector::tpcds::TpcdsConnectorSplit>>
     : formatter<std::string> {
   auto format(
-      std::shared_ptr<facebook::velox::connector::tpcds::TpcdsConnectorSplit> s,
-      format_context& ctx) {
+      std::shared_ptr<
+          facebook::velox::connector::tpcds::TpcdsConnectorSplit> const& s,
+      format_context& ctx) const {
     return formatter<std::string>::format(s->toString(), ctx);
   }
 };

velox/exec/fuzzer/WriterFuzzer.cpp

@@ -16,6 +16,7 @@
 #include "velox/exec/fuzzer/WriterFuzzer.h"
 
 #include <boost/random/uniform_int_distribution.hpp>
+#include <fmt/ranges.h>
 
 #include <re2/re2.h>
 #include <algorithm>

velox/functions/prestosql/fuzzer/NoisySumResultVerifier.h

@@ -16,6 +16,7 @@
 
 #pragma once
 
+#include <boost/algorithm/string/join.hpp>
 #include <cstdint>
 #include <string>
 #include "velox/core/PlanNode.h"
@@ -239,8 +240,8 @@ class NoisySumResultVerifier : public ResultVerifier {
     // Build partition by clause for window function
     std::string partitionBy;
     if (!partitionKeys.empty()) {
-      partitionBy =
-          fmt::format("partition by {}", fmt::join(partitionKeys, ", "));
+      partitionBy = fmt::format(
+          "partition by {}", boost::algorithm::join(partitionKeys, ", "));
     }
 
     // Use row_number() window function to identify first occurrence of each