DarkflameServer/dCommon/CMakeLists.txt

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

78 lines
2.3 KiB
CMake
Raw Normal View History

set(DCOMMON_SOURCES
"AMFDeserialize.cpp"
"AmfSerialize.cpp"
"BinaryIO.cpp"
"dConfig.cpp"
"Diagnostics.cpp"
"Logger.cpp"
"Game.cpp"
"GeneralUtils.cpp"
"LDFFormat.cpp"
"Metrics.cpp"
"NiPoint3.cpp"
"NiQuaternion.cpp"
"Demangler.cpp"
Address Brick-By-Brick builds not properly saving and make migrations automatic (#725) * Properly store BBB in database Store the BBB data in the database as the received SD0 packet as opposed to just the raw lxfml. Addressed several memory leaks as well. * Add Sd0Conversion Add brick by brick conversion commands with 2 parameters to tell the program what to do with the data. Add zlib -> sd0 conversion. Files look good at a glance but should be tested in game to ensure stability. Tests to come. * moving to laptop ignore this commit. I need to move this to my laptop * Add functionality to delete bad models Adds functionality to delete bad models. Models are batched together and deleted in one commit. More testing is needed to ensure data safety. Positive tests on a live database reveal the broken models were truncated and complete ones were kept around successfully. Tests should be done to ensure larger sd0 models are properly saved and not truncated since this command should be able to be run any time. Valgrind tests need to be run as well to ensure no memory leaks exist. * Delete from query change Changed from delete to delete cascade and instead deleting from properties_contents as opposed to ugc. * Address numerous bugs DELETE CASCADE is not a valid SQL command so this was changed to a better delete statement. Added user confirmation before deleting a broken model. Address appending the string model appending bad data, causing excess deletion. Addressed memory leaks with sql::Blob * Error handling for string * Even more proper handling... * Add bounds check for cli command Output a message if a bad command is used. Update MasterServer.cpp * Remove user interference -Add back in mariadb build jobs so i dont nuke others systems - Remove all user interference and consolidate work into one command since 1 depends on the next. * Add comments test Revert "test" This reverts commit fb831f268b7a2f0ccd20595aff64902ab4f4b4ee. * Update CMakeMariaDBLists.txt Test * Improve migration runner Migration runner now runs automatically. - Resolved an issue where extremely large sql queries caused the database to go into an invalid state. - Made migrations run automatically on server start. - Resolved a tiny memory leak in migration runner? (discarded returned pointer) - Moved sd0 migrations of brick models to be run automatically with migration runner. - Created dummy file to tell when brick migrations have been run. * Update README Updated the README to reflect the new server migration state. * Make model deleter actually delete models My complicated sql actually did nothing... Tested that this new SQL properly gets rid of bad data. * Revert "Update CMakeMariaDBLists.txt" This reverts commit 8b859d8529755d7132cef072b2532589c098f683.
2022-10-24 22:20:36 +00:00
"ZCompression.cpp"
"BrickByBrickFix.cpp"
"BinaryPathFinder.cpp"
2022-12-05 08:57:58 +00:00
"FdbToSqlite.cpp"
Address Brick-By-Brick builds not properly saving and make migrations automatic (#725) * Properly store BBB in database Store the BBB data in the database as the received SD0 packet as opposed to just the raw lxfml. Addressed several memory leaks as well. * Add Sd0Conversion Add brick by brick conversion commands with 2 parameters to tell the program what to do with the data. Add zlib -> sd0 conversion. Files look good at a glance but should be tested in game to ensure stability. Tests to come. * moving to laptop ignore this commit. I need to move this to my laptop * Add functionality to delete bad models Adds functionality to delete bad models. Models are batched together and deleted in one commit. More testing is needed to ensure data safety. Positive tests on a live database reveal the broken models were truncated and complete ones were kept around successfully. Tests should be done to ensure larger sd0 models are properly saved and not truncated since this command should be able to be run any time. Valgrind tests need to be run as well to ensure no memory leaks exist. * Delete from query change Changed from delete to delete cascade and instead deleting from properties_contents as opposed to ugc. * Address numerous bugs DELETE CASCADE is not a valid SQL command so this was changed to a better delete statement. Added user confirmation before deleting a broken model. Address appending the string model appending bad data, causing excess deletion. Addressed memory leaks with sql::Blob * Error handling for string * Even more proper handling... * Add bounds check for cli command Output a message if a bad command is used. Update MasterServer.cpp * Remove user interference -Add back in mariadb build jobs so i dont nuke others systems - Remove all user interference and consolidate work into one command since 1 depends on the next. * Add comments test Revert "test" This reverts commit fb831f268b7a2f0ccd20595aff64902ab4f4b4ee. * Update CMakeMariaDBLists.txt Test * Improve migration runner Migration runner now runs automatically. - Resolved an issue where extremely large sql queries caused the database to go into an invalid state. - Made migrations run automatically on server start. - Resolved a tiny memory leak in migration runner? (discarded returned pointer) - Moved sd0 migrations of brick models to be run automatically with migration runner. - Created dummy file to tell when brick migrations have been run. * Update README Updated the README to reflect the new server migration state. * Make model deleter actually delete models My complicated sql actually did nothing... Tested that this new SQL properly gets rid of bad data. * Revert "Update CMakeMariaDBLists.txt" This reverts commit 8b859d8529755d7132cef072b2532589c098f683.
2022-10-24 22:20:36 +00:00
)
# Workaround for compiler bug where the optimized code could result in a memcpy of 0 bytes, even though that isnt possible.
# https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97185
if (CMAKE_CXX_COMPILER_ID STREQUAL "GNU")
set_source_files_properties("FdbToSqlite.cpp" PROPERTIES COMPILE_FLAGS "-Wno-stringop-overflow")
endif()
add_subdirectory(dClient)
foreach(file ${DCOMMON_DCLIENT_SOURCES})
set(DCOMMON_SOURCES ${DCOMMON_SOURCES} "dClient/${file}")
endforeach()
add_library(dCommon STATIC ${DCOMMON_SOURCES})
target_include_directories(dCommon
PUBLIC "." "dClient" "dEnums"
PRIVATE
"${PROJECT_SOURCE_DIR}/dDatabase/GameDatabase"
"${PROJECT_SOURCE_DIR}/dDatabase/GameDatabase/ITables"
"${PROJECT_SOURCE_DIR}/dDatabase/CDClientDatabase"
"${PROJECT_SOURCE_DIR}/thirdparty/mariadb-connector-cpp/include"
)
if (UNIX)
find_package(ZLIB REQUIRED)
elseif (WIN32)
include(FetchContent)
set(WITH_GTEST OFF CACHE BOOL "" FORCE)
set(ZLIB_ENABLE_TESTS OFF CACHE BOOL "" FORCE)
2024-12-11 15:39:57 +00:00
set(ZLIB_COMPAT ON CACHE BOOL "Enable ZLIB compatibility mode" FORCE)
# TODO Keep an eye on the zlib repository for an update to disable testing. Don't forget to update CMakePresets
FetchContent_Declare(
zlib
URL https://github.com/zlib-ng/zlib-ng/archive/refs/tags/2.2.2.zip
URL_HASH MD5=2cf9199fb785ea579a2a9905a75c38b3
)
# Disable warning about no project version.
set(CMAKE_POLICY_DEFAULT_CMP0048 NEW)
# Disable warning about the minimum version of cmake used for bcrypt being deprecated in the future
set(CMAKE_WARN_DEPRECATED OFF CACHE BOOL "" FORCE)
FetchContent_MakeAvailable(zlib)
set(ZLIB_INCLUDE_DIRS ${zlib_SOURCE_DIR} ${zlib_BINARY_DIR})
add_library(ZLIB::ZLIB ALIAS zlib)
else ()
message(
FATAL_ERROR
"This platform does not have a way to use zlib.\nCreate an issue on GitHub with your build system so it can be configured."
)
endif ()
target_link_libraries(dCommon
PRIVATE ZLIB::ZLIB bcrypt tinyxml2
INTERFACE dDatabase)