Fatal error. error #3684, null

fatal error. error #3684, null

au3 functions. Fixed: _WinAPI_CreateEvent() return error on already define $sName. Fixed: use "wstr" for "ptr" with Null value. Fixed. Fixed cannot exit directly when a fatal error occurs in PHP8 environment Fixed null socket error in ioException (swoole/[email protected]) (@twose). # IPython notebook fail with “ Internal Server Error” PR # log fatal error when PDF conversion fails.

Fatal error. error #3684, null - all became

Release Notes

Changelog¶

Check sprers.eu for issues and more details about development, contributors, etc.

Important

Conan shouldn’t break any existing recipe or command line invocation. If it does, please submit a report on GitHub. Read more about the Conan stability commitment.

(Feb)¶

  • Feature: Add sprers.eue_manager tools to conan config list. # . Docs here
  • Feature: Use system package manager helpers from sprers.eue_manager. # . Docs here
  • Feature: Add section to profiles. #
  • Feature: Add and , conan new templates. # . Docs here
  • Feature: Add is_msvc_static_runtime method to sprers.eu to identify when using msvc with static runtime. # . Docs here
  • Feature: Improve support for Visual Studio in . #
  • Feature: Make pkg-config tooling accessible under sprers.eufig and sprers.eufigDeps. #
  • Feature: Use .bazel suffix for generated Bazel files. # . Docs here
  • Feature: New tools in sprers.eu for invoking system package managers in recipes. # . Docs here
  • Feature: Testing the expected PC files created when the component name matches with the root package one using either pkg_config or PkgConfigDeps generators. # . Docs here
  • Feature: Better definition of clang compiler in Windows in CMakeToolchain. #
  • Feature: Add VxWorks to OSs in default sprers.eu # . Docs here
  • Feature: Add is_msvc to validate if sprers.euer is Visual Studio and msvc compilers. # . Docs here
  • Feature: sprers.eu field is mandatory for CMakeToolchain and OS in (‘Macos’, ‘iOS’, ‘watchOS’, ‘tvOS’). #
  • Feature: Adding and to and conan install to work with . # . Docs here
  • Feature: Adding clang 14 to . Needed for emsdk package in Conan Center Index. #
  • Feature: PkgConfigDeps shows WARN messages if there are duplicated pkg_config_name and/or pkg_config_aliases. # . Docs here
  • Feature: Improvements in , including some cross-building functionality. #
  • Feature: Update content created by the markdown generator. # . Docs here
  • Fix: Remove auto-detection of VS as compiler, detect it as version . # . Docs here
  • Fix: Do not report warning for duplicated component names in CMakeDeps. #
  • Fix: Let legacy Meson build helper use other backends apart from ninja. #
  • Fix: msvc_runtime_flag returns empty string instead of None. # . Docs here
  • Fix: Parsing a url with query args in results in a bad filename that could fail. #
  • Fix: The argument patch_file from sprers.eu is now relative to sprers.eu_folder`by default, unless an absolute path to another location is provided, for example, to a path in the `sprers.eu_folder. # . Docs here
  • Fix: Use install folder for Bazel dependency paths. # . Docs here
  • Fix: Enforce CMP policy to NEW in CMakeToolchain. #
  • Fix: Add quotes around conan_message output variable so it is not modified. #
  • Fix: Add as hidden-import to sprers.eu, so it is bundled with the installer. #
  • Fix: Move imports of pre-defined layouts to their build-system domain. # . Docs here
  • Fix: Allow generator checks for Visual Studio #
  • Fix: Do not generate transitive .props for MSbuildDeps tool-requires. #
  • Fix: Manage spaces in profile definition. #
  • Fix: Add to LDFLAGS in when necessary. #
  • Fix: Fix extra } characters in cppstd info message. #
  • Fix: Fix quotes in generated environment deactivation scripts. #
  • Fix: Fix the generated code, so it doesnt fail for . #
  • Fix: Fix spaces in sprers.eu to prevent the YAML linter from complaining. #
  • Fix: Convert NewCppInfo folders to absolute. #
  • Fix: Improved CMakeToolchain robustness regarding , and commands allowing better cross-build scenarios and better differentiation of the right context where to get, for example, executables (build vs host). # . Docs here
  • Bugfix: Fix using absolute paths instead of relative. #
  • BugFix: Avoid BazelDeps exception when depending on a package without libs. Close sprers.eu#
  • Bugfix: Fix that happened when sing with packages (and lockfiles). #
  • Bugfix: Visual Studio auto-detected profile was incomplete. #
  • Bugfix: Fix the caching of sprers.euencies at validate() time. #
  • Bugfix: Avoid errors when using of repeated references (which can happen if using dependencies). #

(Jan)¶

  • Bugfix: The CMakeDeps generator now uses the property cmake_build_modules declared in components of the required packages not only in the root cpp_info. #
  • Bugfix: Adding missing hidden-imports to pyinstaller. Close sprers.eu#
  • Bugfix: Make pkg_config generator listen to root cpp_info properties. #

(Dec)¶

  • Feature: Add <PackageName>_LIBRARIES, <PackageName>_INCLUDE_DIRS, <PackageName>_INCLUDE_DIR, <PackageName>_DEFINITIONS and <PackageName>_VERSION_STRING variables in CMakeDeps. #
  • Feature: Adding a new Block to the CMakeToolchain now doesn’t require inheriting CMakeToolchainBlock. # . Docs here
  • Feature: Add build_modules and build_modules_paths to JsonGenerator. # . Docs here
  • Feature: The CMakeToolchain is now prepared to apply several user toolchains. # . Docs here
  • Feature: In the sprers.eu of the test_package, the reference being tested is always available at sprers.eu_reference_str. # . Docs here
  • Feature: Introduced a new test_type value explicit so a user can declare explicitly the requires or build_requires manually (using sprers.eu_reference_str), it won’t be automatically injected as a require. In Conan the test_type attribute will be ignored, the behavior will be always explicit, so declaring test_type=”explicit” will make the test recipe compatible with Conan # . Docs here
  • Feature: Introduced tool_requires attribute to provide a compatible way to migrate to Conan , where the current concept of build_requires has been renamed to tool_requires. # . Docs here
  • Feature: Upgrade Conan python jinja requirement to v3.x. #
  • Feature: Provided several sprers.eu functions to manage symlinks: Transform absolute to relative symlinks, remove broken symlinks, remove external symlinks and get the symlinks in a folder. These tools will help migrate to Conan where the package files won’t be automatically cleaned from broken absolute symlinks or external symlinks. # . Docs here
  • Feature: Remove legacy folder setters in only used for testing. #
  • Feature: Add property to check the current git version, aligned with . # . Docs here
  • Feature: Add build_requires support in BazelDeps generators. #
  • Fix: Fix variable names set by CMakeDeps modules. #
  • Fix: Call to find_dependency in module mode to find transitive dependencies. #
  • Fix: remove rpath from .pc files generated by pkg_config & PkgConfigDeps generators. # . Docs here
  • Fix: Deleted CMake warning for already existing targets. #
  • Bugfix: Fix passing component’s linkflags in CMakeDepes generator #
  • Bugfix: AutotoolsToolchain was not passing the compiler to get_gnu_triplet function. #

(Jan)¶

  • Bugfix: The CMakeDeps generator now uses the property cmake_build_modules declared in components of the required packages not only in the root cpp_info. #
  • Bugfix: Make pkg_config generator listen to root cpp_info properties. #

(Dec)¶

  • Fix: Remove argument from method. # . Docs here
  • Fix: Do not convert to property the legacy . #
  • Bugfix: Compiler msvc was not working for CMake legacy generators. #

(Dec)¶

  • Bugfix: Making aggregate_components non-destructive, which was causing errors in generators with components. #
  • Bugfix: Fix the definition of D_GLIBCXX_USE_CXX11_ABI in gcc-like compilers for and . Define it only to for new compilers, assuming that the default is alread 1. #

(Dec)¶

  • Feature: Remove cmake_target_namespace and cmake_module_target_namespace properties. # . Docs here
  • Feature: Allow CMakeDeps to set cmake_target_name property as an absolute target. # . Docs here
  • Feature: Add warning in CMakeDeps generated CMake files when target names collide. # . Docs here
  • Feature: Legacy cmake generators (cmake_find_package, cmake_find_package_multi) don’t listen to new set_properties model anymore. # . Docs here
  • Feature: pkg_config_name is used as the main name for a package/component and it will be used as the name for the *.pc file. # . Docs here
  • Feature: Added new property`pkg_config_aliases` which admits a list of strings to define different aliases for any package/component. # . Docs here
  • Feature: Define in to represent platforms without OS “bare metal”. # . Docs here
  • Feature: Modern to supersede legacy . Includes management of PKG_CONFIG_PATH and mapping to a structure # . Docs here
  • Feature: Add for migration of force_host_context. # . Docs here
  • Feature: Added C++23 support for Visual Studio and GCC one. #
  • Feature: Enable to prepare for future namespace. # . Docs here
  • Feature: Add backend support to MesonToolchain. # . Docs here
  • Fix: Fix <PackageName>_FIND_COMPONENTS CMake generated variable to the correct value associated with the filename, not the package name. # . Docs here
  • Fix: Updated the ConanException in sprers.eu to improve the error message handling. # . Docs here
  • Fix: Simplify definition in new , leaving only . Use max number of CPUs by default to build in parallel. # . Docs here
  • Fix: Fix the version model, which is comparison broken with the “main” version < # . Docs here
  • Fix: Fix the sprers.eu_ps1() method. #
  • Fix: CMakeToolchain will not crash if build_type not defined. #
  • Fix: Avoid raising an exception for when there are editables in the graph. #
  • Fix: Fix wrong parameter name in CMakeDeps find_library function. #
  • Fix: Improve error message when unzipping Conan .tgz artifacts. #
  • Fix: Respect error code 6 in some situations. #
  • Bugfix: Fix parallel package downloading. While downloading conan locks incorrect package ref. #
  • Bugfix: Add import for custom Blocks in . # . Docs here
  • Bugfix: Option –require-override is not working for sprers.eu. #
  • Bugfix: Fix unescaped double-quotes for defines in generator. #
  • Bugfix: Use new sprers.eu_building in MesonToolchain . #
  • Bugfix: CMakeDeps generated *sprers.eu was not including properly the set of link flags. #
  • Bugfix: CMakeDeps was not populating INTERFACE_LINK_OPTIONS to each target. #
  • Bugfix: PkgConfigDeps was not adding correctly the Requires for all the package dependencies. #
  • Bugfix: properly included and path quoted in . #
  • Bugfix: Solved an issue with the conan config install whereby a cryptic error was raised when a user tried to install a directory that previously was a file and vice-versa. #
  • Bugfix: Missing framework for Xcode generator with no compiler setting. #

(Nov)¶

  • Bugfix: Legacy cmake_multi generator is not affected by set_property. #

(Nov)¶

  • Fix: Fix XcodeDeps architecture name translation from Conan to Apple identifiers. #
  • Bugfix: Fix XcodeDeps bad xcconfig generation when using dash-case-named packages #
  • Bugfix: Avoid exception if compiler not defined in sprers.eu file. #
  • Bugfix: legacy cmake generator is not affected by set_property. #

(Oct)¶

  • Feature: Remove sdk condition in _.xcconfig_ files generated by _XcodeDeps_. #
  • Feature: Add new Macos version (Monterey). #
  • Feature: Add CMAKE_POSITION_INDEPENDENT_CODE in CMakeToolchain if it’s set in the conanfile independently from the value of the shared option. #
  • Feature: Generate aggregated deactivation file for aggregated environments. # . Docs here
  • Feature: adding preprocessor definitions to ResourceCompile. #
  • Feature: Support cmake_module_target_name and cmake_module_file_name properties in _cmake_find_package_ generator. # . Docs here
  • Feature: Define new for the MSBuild toolchain and remove generation of intel_vars files by . # . Docs here
  • Feature: New Conan _XcodeDeps_ multi-config generator for _Xcode_. # . Docs here
  • Feature: Refactor sprers.eufigdeps module. #
  • Feature: Decoupled as an abstract environment representation (do not depend on conanfile), and , as the specialization for a given conanfile, settings/settings_build, and . # . Docs here
  • Feature: Enabled patch_string when using apply_conandata_patches. # . Docs here
  • Feature: Add property “cmake_target_aliases” to create some CMake alias targets using CMakeDeps. #
  • Fix: Do not fall back for the property “filename” to target properties, only to legacy “names” definition. #
  • Fix: Fix build settings conditional logic to work correctly with Xcode IDE. #
  • Fix: Show a more specific error message on ‘conan search’ when a package version is not valid #
  • Fix: Removed the new layout sprers.eue property as it was not clear the value/usage. # . Docs here
  • Fix: The “conan package” method now raises an exception when declaring the layout() method. This is part of the migration to Conan where the local method “conan package” is removed. # . Docs here
  • Fix: Moved experimental LayoutPackager to a sprers.euckager with a new interface. Removed also the sprers.euns attribute that now is configured directly in the AutoPackager. # . Docs here
  • Fix: When the layout() method is declared inside the sprers.eu of a test-package, Conan won’t generate temporary folders at build/_HASH_ anymore. The build folder will follow the structure declared at the layout() method. # . Docs here
  • Fix: Fix frameworks aggregation in XcodeDeps. #
  • Fix: Remove python version from header, it can be problematic sometimes and adds no value. #
  • Fix: Change to define as , because Meson now raises errors on it defined in . #
  • Fix: Rename to (being it a object). The ConanFileInterface object is only used when accessing sprers.euencies in a conanfile, so we can make sure that the new develop generators use the final cpp_info name. # . Docs here
  • Fix: Renamed argument in classes to , that can get “build” and “run” values. # . Docs here
  • Fix: New behavior failed for dual profile, the new computation of subsystem paths completely depend now on defined settings, but not on or auto-detection. # . Docs here
  • Bugfix: Avoid lockfile raising because of incorrect options when using . Close sprers.eu#
  • Bugfix: Avoid VCVars defining argument for VS<= Close sprers.eu#
  • Bugfix: Respect the previous value of . #
  • Bugfix: conan export-pkg now raises an error (ConanInvalidConfiguration exception) if the method raise that error and results in an invalid package_id. #
  • Bugfix: Respect the even for cli argument. #
  • Bugfix: Avoid crash in for custom generator when compiler is not defined. #
  • Bugfix: Do not force new authentication when a token is expired, it might not be needed. #
  • Bugfix: The CMakeToolchain generator always sets fPIC enabled due to a typo where the actual parameter is not templated but instead hard-coded as ON. #
  • Bugfix: The path to a patch file in a sprers.eu is also relative to the base source folder when the patches are not associated with a version (list of patches). # . Docs here

(Oct)¶

  • Feature: Added IntelCC as public generator. # . Docs here
  • feature: Prepare for adoption. # . Docs here
  • Feature: Support multiple toolchains in one recipe. # . Docs here
  • Feature: generator learned how to handle to use executables from them. # . Docs here
  • Feature: helper now honors PKG_CONFIG environment variable. #
  • Feature: Make new environment generators multi-config (Release/Debug and arch). # . Docs here
  • Feature: Environment scripts will generate scripts by default. #
  • Feature: Support remote archives other than zip in . #
  • Feature: New “compiler” intel-cc with different modes (icx, dpcpp, classic) for Intel oneAPI. # . Docs here
  • Feature: Add Intel oneAPI support for CMakeToolChain, MSBuildToolchain and VCVars. # . Docs here
  • Feature: Add objects attribute to the cpp_info so that we can add object files to the linker without having to add those mixed with linker flags. # . Docs here
  • Feature: New setttings: _xtensalx6_ and _xtensalx_ for ESP32/ESP platforms. #
  • Fix: New environment deactivate scripts fail in Windows because env-var different casing. #
  • Fix: Avoid “overcrowding” Windows environment variables with LD_LIBRARY_PATH and DYLD_LIBRARY_PATH, not used in Windows. #
  • Fix: Command conan package now works for new with generators folder. #
  • Fix: Don’t just print “ERROR: True” on unresolvable conflict. #
  • Fix: Fix migrations settings comparison. #
  • Fix: Removed unused dependency from Python . #
  • Fix: Make automatic call of and independent. # . Docs here
  • Bugfix: was using the instead of a relative folder to the conanfile, producing layout errors. #
  • Bugfix: Support components with symbolic imports. #
  • Bugfix: Fix bug with argument. #
  • Bugfix: Add support for Xcode 13/Apple clang #
  • Bugfix: Fixed bad Requires declaration in *.pc files. #
  • Bugfix: Fixed bug whereby when a conflict of requirements happens, in some special situations, just a message ERROR: True was printed in the terminal, making it hard to guess what was going on. #
  • BugFix: Fixed a bug where using the new layout() the exports went to the sprers.eu_folder instead of the base source folder in the cache and only the exports_sources should go there. #

(Oct)¶

  • Fix: Check current _sprers.eu_ file when updating Conan and only migrate if the user has not modified the file. If the local file is modified then create a new cacert file but don’t overwrite current. #
  • Fix: Update Conan debian package to fix ssl certificates problem. #

(Sept)¶

  • Bugfix: Added root certificate for Let’s encrypt. #

(Sept)¶

  • Bugfix: Add support for Xcode 13/Apple clang #

(Sept)¶

  • Feature: Change default cmake_layout() source folder from ‘src’ to ‘.’ # . Docs here
  • Feature: Recovered base_path argument for sprers.eu and sprers.eu_conandata_patches to be able to specify a relative folder from the sprers.eu_folder directory (that follows the layout() method). # . Docs here
  • Fix: Allow user definition of CMAKE_XXX_INIT variables in user toolchains when using CMakeToolchain. #
  • Fix: Upgrade minimum in sprers.eu to make it compatible with latest upgrade of to #
  • Bugfix: Aggregate [conf] from build_requires earlier so it is available for generators declared as generators. attribute. Close sprers.eu#
  • Bugfix: The qmake generator now assigns QMAKE_LFLAGS_SHLIB and QMAKE_LFLAGS_APP variables instead of the incorrect QMAKE_LFLAGS following the official docs. # . Docs here

(Sept)¶

  • Feature: Update conan new modern templates and . # . Docs here
  • Feature: Introduced a new cpp_info property cmake_target_namespace to declare the target namespace for the CMakeDeps generator. This feature allows declaring a global target with a different namespace like Foo::Bar. # . Docs here
  • Feature: Detect Visual Studio as msvc. #
  • Feature: Add Clang 13 support. # . Docs here
  • Feature: Testing support for Windows CMake + Clang (independent LLVM, not VS) + Ninja/MinGW builds, and CMake + Clang (Visual Studio 16 internal LLVM 11 via ClangCL toolset). #
  • Feature: Provide new to enable the usage of the build profile as default, and to allow definition of the host profile default in new . # . Docs here
  • Feature: CMakeToolchain new member find_builddirs defaulted to True to add the cpp_sprers.euirs from the requirements to the CMAKE_PREFIX_PATH/CMAKE_MODULE_PATH. That would allow finding the config files packaged and to be able to include() them from the consumer sprers.eu. # . Docs here
  • Feature: CMakeDeps. Added a new property cmake_find_mode with possible values to config`(default), `module, both or none to control the files to be generated from a package itself. The none replaces the current skip_deps_file property. # . Docs here
  • Feature: CMakeDeps: Added two new properties cmake_module_file_name and cmake_module_target_name, analog to cmake_file_name and cmake_target_name, but to configure the name of sprers.eu file and the target declared inside. # . Docs here
  • Feature: Remove conan-center (sprers.eu) default remote. # . Docs here
  • Feature: Implement round trip new profile section, necessary for lockfiles, and specially stdout printing. #
  • Feature: Allow wildcard for consumer, too, same it was done for settings in # . Docs here
  • Fix: Adding management of private dependencies, via new trait compatible with for new and . #
  • Fix: Remove unused pip dependency #
  • Fix: Upgrade dependency to allow #
  • Fix: Make conan remove accept package reference syntax. # . Docs here
  • Fix: Fixed old CMake build helper to cross-build to iOS when two profiles are specified. #
  • Fix: Fix conan export typo in help message. # . Docs here
  • Fix: Relax python six dependency to allow #
  • Fix: Bump urllib3 version to #
  • Fix: The new Autotools build helper accepts a build_script_folder ` argument in the `configure() method to specify are subfolder where the configure script is. # . Docs here
  • Fix: Use frameworks in Premake generator. # . Docs here
  • Fix: The tool sprers.eu_conandata_patches will use the root source folder to find the patch file and the tool sprers.eu will take the current source folder declared in the layout() method to know where is the source to apply the patches. # . Docs here
  • Fix: Avoid checking other remotes when is defined and revisions are activated and binary is not found in the defined remote. #
  • Bugfix: Setting the CMAKE_OSX_DEPLOYMENT_TARGET variable as a cache entry. #
  • BugFix: Use topological ordering to define VirtualBuildEnv composition and precedence of appending variables. #
  • Bugfix: Bazel build files have an extra ] if there are no dependencies. #
  • Bugfix: Add AlmaLinux to with_yum. #
  • Bugfix: CMakeToolchain. Fixed a bugfix whereby a variable declared at the .variables containing a boolean ended at CMake with a quoted “True” or “False” values, instead of ON / OFF# . Docs here
  • Bugfix: Fixed bug whereby Conan failed when using compiler=gcc with sprers.eun=5 (without specifying a minor version) and sprers.eu=17. #
  • Bugfix: No verbose traceback was been printed for sprers.eu() method. #
  • Bugfix: Fix Bazel cc_library: deps and linkopts. #
  • Bugfix: Fixed bug whereby using new layout() method together with sprers.euents in the package_info method caused an exception. #
  • Bugfix: Fix PkgConfigDeps that was failing in the case of components with requirements. #

(Jul)¶

  • Feature: Use CMAKE_OSX_DEPLOYMENT_TARGET to get -version-min set in _CMakeToolchain_. #
  • Feature: Display information in the conan info output. #
  • Feature: Now it is possible to define settings for a downstream consumer using -s &:setting=value or the same syntax in the profile even if the consumer is a sprers.eu or a sprers.eu not declaring a name. e.g: Building a Debug application with Release dependencies: -s build_type=Release -s &:build_type=Debug# . Docs here
  • Feature: The AutotoolsDeps allows to alter the generated environment corresponding to the information read from the dependencies before calling the generate() method. # . Docs here
  • Feature: new remove and items methods for the Environment objects. # . Docs here
  • Feature: New VCVars generator that generates a sprers.eu that activates the Visual Studio Developer Command Prompt. # . Docs here
  • Feature: Skip build helper test and using [conf]. # . Docs here
  • Feature: Implement a new syntax to be able to dissambiguate alias requirements and resolve them earlier in the flow, solving some limitations of the previous alias definition. This approach is intended to be the one in Conan (issue backported from sprers.eu). # . Docs here
  • Feature: Introduce the argument to define dependency overrides directly on command line. # . Docs here
  • Feature: New sprers.eu_bash mechanism to enable running commands in a bash shell in Windows. It works only with the new environment definition from the dependencies (env_buildinfo and run_buildinfo) as long as the new AutotoolsToolchain, AutotoolsDeps and Autotools build helper. It supports automatic conversion of the environment variables values declared as “path” according to the declared subsystem in the conf sprers.eu:subsystem, that is not being auto-detected anymore. # . Docs here
  • Feature: A unique environment launcher (sprers.eu) is generated to aggregate all the environment generators (VirtualRunEnv, VirtualBuildEnv, AutotoolsToolchain and AutotoolsDeps) that had been generated so the user can easily activate all of them with one command. # . Docs here
  • Feature: Use _CMake_ File API. #
  • Fix: Add definition to . #
  • Fix: Improve error message when a package in editable mode. #
  • Fix: Add to model. #
  • Fix: Fix _CMake_ rejecting library name with special characters. #
  • Fix: Use filename [PKG-NAME]-[COMP-NAME] for PkgConfigDeps. # . Docs here
  • Fix: Saving all the toolchain args information into sprers.eu instead of json file. # . Docs here
  • Fix: Added warning in the new toolchains (the used in the generate() method) if no build profile is being used. # . Docs here
  • Fix: Implemented check that will raise an error in the CMakeDeps generator when using the build_context_activated, build_context_suffix or build_context_build_modules attributes if no build profile is being used. # . Docs here
  • Fix: The`CMakeDeps` generator will check if the targets specified in the find_package(foo components x y z) exist instead of checking against an internal variable. Also, this check will be done at the end of the sprers.eu so any included build_module can declare the needed targets. # . Docs here
  • Fix: Consistent help message for conan profile (sub-command part). # . Docs here
  • Fix: Consistently put short arguments (-a) before long ones (–args). #
  • Fix: CC=clang–gcc-toolchain is now identified as _clang_. #
  • Fix: The new VirtualEnv generator has been split into VirtualRunEnv and VirtualBuildEnv. Both are automatically generated as before but only VirtualBuildEnv will be activated by default. # . Docs here
  • Bugfix: Fixing when conanfile has . #
  • Bugfix: Fix QbsProfile toolchain KeyError. #
  • Bugfix: Do not define CMAKE_GENERATOR_TOOLSET in CMakeToolchain for Ninja generator, and define it in instead. #
  • BugFix: in host context, like gtest, are being propagated downstream by generators in the model. # . Docs here
  • Bugfix: Fix that overridden requirements _”cannot be found in lockfile”_. #

(Jun)¶

  • Feature: New generator. # . Docs here
  • Feature: Proposal of jinja2 templates for profiles. # . Docs here
  • Feature: Add support for CMAKE_CXX_STANDARD_REQUIRED in _CMakeToolchain_. #
  • Feature: Add information to conan info output both to stdout and json outputs. # . Docs here
  • Feature: Improved the new AutotoolsToolchain, AutotoolsDeps and Autotools build helper. # . Docs here
  • Feature: Initial cross-build support in with definition of , and , deduced from (only using the 2 profiles) and from new items. # . Docs here
  • Feature: Easier access to modify or update context values in blocks. # . Docs here
  • Feature: Provide [conf] command line support. # . Docs here
  • Feature: Added support for using server config from a custom location, setting CONAN_SERVER_HOME env variable or using -d or –server_dir flag when launching the server with conan_server command. # . Docs here
  • Feature: Support for CMakeDeps generator of a new property “skip_deps_file” to be declared in the cpp_info of a package to skip creating sprers.eu files for it, allowing to create “system wrapper” recipes easily. # . Docs here
  • Feature: New model, using a dict {requirement: ConanFileInterface} to prepare for Conan # . Docs here
  • Feature: Allow a explicit to update cache revision without . # . Docs here
  • Feature: New layout helper to define a multi-platform CMake layout that will work for different generators (ninja, xcode, visual, unix), and is multi-config. # . Docs here
  • Feature: The conan_sprers.eu now includes xxx_DIR variables for the dependencies to ease the find_package mechanism to locate them. The declaration of these directories is a must when cross-building in OSX where CMake ignores CMAKE_PREFIX_PATH and CMAKE_MODULE_PATH to look only at the system framework directories. # . Docs here
  • Feature: Provide access in the recipes to the environment declared with the [new environment system](sprers.eu). # . Docs here
  • Fix: Fix Bazel build string defines. #
  • Fix: Fixed behavior in the sprers.eus feature whereby the sources saved or downloaded inside the source(self) were saved at the sprers.eu folder. But sprers.eu is intended to describe where the sources are instead of forcing where the sources are saved. # . Docs here
  • Fix: Properly generate qbs profile for msvc. #
  • Fix: Configuration sprers.eu_home_short works with “None” value. #
  • Fix: Avoid to generate OSX and Apple config for non Apple builds. #
  • Fix: The new MesonToolchain now takes the declared environment variables (CC, CXX…) from build-requires and profiles to set the variables c, cpp, c_ld, cpp_ld etc, into the conan_meson_sprers.eu# . Docs here
  • Fix: Added new preprocessor_definitions to new Meson build helper. # . Docs here
  • Fix: The new MesonToolchain now allows adjusting any variable before generating the conan_meson_sprers.eu file. # . Docs here
  • Bugfix: Disabled remotes shouldn’t fail if not used at all #
  • BugFix: was retrieving the dependency if existing, or failing otherwise, because the default requires was hardcoded to fetch the host (build=False) dependency. # . Docs here
  • Bugfix: Now, conan profile {show, update, get, remove} is working fine with new experimental [conf] section. #

(Jun)¶

  • Bugfix: Avoid crash when using with the conan test command. #
  • Bugfix: The CMakeDeps generator variables were named wrongly when a component had the same name as the package. #

(Jun)¶

  • Fix: Update the experimental template to start using the new basic info. #
  • Fix: Do not fail in when there are with same name as host requires, unless is enabled for those and a different suffix has not been defined. #
  • Fix: When using the new sprers.eu (at layout(self) method) the sources (from export, export_sources and scm) are copied to the base source folder and not to the sprers.eu that is intended to describe where the sources are after fetching them. # . Docs here
  • BugFix: Do not quote all values and allow integer and macro referencing in #
  • Bugfix: The new generators like CMakeDeps and CMakeToolchain`write the generated files defaulting to the `install folder if no sprers.eutors is specified in the layout() method. #
  • Bugfix: The CMakeToolchain generator now manages correctly a recipe without arch declared in an Apple system. #

(May)¶

  • Feature: Remove by default to True in , it is not necessary by default, users can opt-in, and new test validates shared libs will work with generator . # . Docs here
  • Feature: simplified with only 1 category of blocks, made template code as another block, and reordered blocks so relevant flags for try-compile are taken into account. # . Docs here
  • Feature: Add new default conancenter remote for sprers.eu as first in the list. # . Docs here
  • Feature: Implements a new experimental Bazel integration with , and . # . Docs here
  • Feature: Introduced new options for the CMakeDeps generator allowing to manage build_requires even declaring the same package as a require and build_require avoiding the collision of the config cmake files and enabling to specify which build_modules should be included (e.g protobuf issue) # . Docs here
  • Feature: Expand user-agent string to include OS info. #
  • Feature: Implement for conan export-pkg packages that cannot be rebuilt with . # . Docs here
  • Feature: Define for defining the user toolchain for , both for injecting a user toolchain in the generated and for completely replacing . # . Docs here
  • Feature: add GCC 11 to _sprers.eu_. #
  • Feature: Add new sprers.eu() interface. # . Docs here
  • Feature: Update urlib3 Conan dependency setting version >= to avoid CVE #
  • Feature: Build-requires can define [conf] for its consumers. # . Docs here
  • Feature: support M1 Catalyst. #
  • Feature: New and (when not using ) arguments to explicitly define that the installed or created package has to be a , receiving the build profile instead of the host one. # . Docs here
  • Feature: Introduced the layout() method to the recipe to be able to declare the folder structure both for the local development methods (conan source, conan build…) and in the cache. Also, associated to the folders, cppinfo objects to be used in editable packages and file pattern descriptions to enable “auto packaging”. # . Docs here
  • Fix: CMakeDeps generator: The transitive requirements for a build_require are not included in the sprers.eu files generated. #
  • Fix: The CMakeToolchain now supports Apple M1 cross-building with a profile without environment declared pointing to the system toolchain. #
  • Fix: Set env_sprers.eu_FRAMEWORK_PATH correctly. #
  • Fix: Fix some typos in the code. #
  • Fix: Improve error message when a directory doesn’t contain a valid repository. #
  • Fix: The build_modules defined per generator in cpp_info now are rendered properly using the markdown generator. #
  • Fix: Simplify code access to [conf] variables removing attribute based access. #
  • Bugfix: Prevent unintended evil insertions into sprers.eu resulted in corrupted package and inability to install. #
  • Bugfix: Allow to correctly process packages with dots in the package name. #
  • Bugfix: Avoid errors because of mismatch in lockfiles when using feature. #
  • BugFix: Respect order of declared directories when using components. #
  • Bugfix: Raise an exception when response header Content-type is different than application/json or application/json; charset=utf-8. #
  • Bugfix: Fix exception in CMakeToolchain when settings remove known compilers. #
  • Bugfix: Fix current directory definition in commands in new toolchains. #
  • Bugfix: AptTool: add repo key before running apt-add-repository. #
  • BugFix: Prevent evil insertions into sprers.eu resulted in corrupted package and inability to install. #

(Apr)¶

  • Feature: Add support to for Ninja toolchain defined with . #
  • Feature: The CMakeDeps generator will print CMake traces with the declared targets. e.g: Target declared: ‘OpenSSL::Crypto’. #
  • Feature: Add clang 12 support. #
  • Feature: List tools and core from profile and _sprers.eu_. # . Docs here
  • Feature: Add cross-building tests for new AutoTools build helper. #
  • Feature: generates a file with the generator to be used in the command line later, so it is not necessary to duplicate logic, and is explicit what generator should be used. # . Docs here
  • Feature: learned to build with different toolsets, down to the minor compiler version, for the compiler. # . Docs here
  • Feature: Validate checksum and retry download for corrupted downloaded cache files. #
  • Feature: defining CMAKE_GENERATOR_TOOLSET for msvc version different than the default. #
  • Feature: Implement in recipes, so build_requires can be tested as such. # . Docs here
  • Feature: Make available the full recipe and package reference to consumers via . #
  • Feature: New customization and extensibility mechanism with blocks of components instead of inheritance. # . Docs here
  • Feature: Add set_property and get_property to set properties and access them in generators. Can be set only for a specific generator or as a default value for all of them. # . Docs here
  • Feature: Use set_property and get_property to support custom defined content in pkg_config generator. # . Docs here
  • Feature: Add new property names: cmake_target_name, cmake_file_name, pkg_config_name and cmake_build_modules that can be used for multiple generators of the same type allowing also an easier migration of names, filenames and build_modules properties to this model. # . Docs here
  • Feature: Skip package when building all package from sources at once using –build=!<package> syntax. # . Docs here
  • Feature: will generate for Ninja builds for . #
  • Fix: Remove , superseded by . # . Docs here
  • Fix: Allow spaces in the path for new environment files and Visual toolchain file. #
  • Fix: Return deprecated attribute in conan inspect command. #
  • Fix: Check if Artifactory url for publishing the build_info has artifactory string as the service context and remove from the API url if it doesn’t. #
  • Fix: Recognize as a CMake multi-configuration generator. #
  • Fix: using CMAKE_CURRENT_LIST_DIR in CMakeToolchain to locate CMakeDeps config files. #
  • Fix: config_install_interval no longer enter in loop when invalid. # . Docs here
  • Fix: Remove multi-config support for generator. # . Docs here
  • Fix: Accept relative profile path when folder is on same tree level. # . Docs here
  • Bugfix: Fixed test_package/sprers.eu using for a package belonging to a lockfile. #

(Apr)¶

  • Bugfix: Revert regression that replaces first by in in _CMake_ generators because it can break passing objects and other paths that start with . #

(Apr)¶

  • Fix: Avoid breaking users calling forbidden private api . #
  • Bugfix: Fix opensuse SystemPackageTools incorrectly using apt-get when zypper-aptitude. #
  • Bugfix: Fix linker flags in cmake (find_package based) generators. #
  • Bugfix: Fixed bug in transitive build_requires of MSBuildDeps. #

(Mar)¶

  • Feature: generator uses new visitor model and handles conditional requirements correctly. # . Docs here
  • Feature: CMake toolchain supports include_guard() feature #
  • Feature: New command. # . Docs here
  • Feature: Use file for Meson toolchain #
  • Feature: Allow arbitrary defines in conan new templates. # . Docs here
  • Feature: Automatically handle CONAN_RUN_TESTS to avoid extra boilerplate. # . Docs here
  • Feature: More fine-grained control (using [conf]) for build parallelization. # . Docs here
  • Feature: Add support for testing with different tools versions. #
  • Feature: Add different CMake versions for testing. #
  • Feature: Move the definition of CMakeDeps variables to its own file # . Docs here
  • Feature: Added sprers.eu to apply a single patch (new interface for legacy sprers.eu function. # . Docs here
  • Feature: Added sprers.eu_conandata_patches to apply patches defined in sprers.eu. # . Docs here
  • Feature: Allow integers as in . #
  • Feature: New model for recipes and profiles # . Docs here
  • Feature: Do not remove sh from the path in the new CMake helper. # . Docs here
  • Feature: Allow definition of custom Visual Studio version for msvc compiler in MSBuild helpers. # . Docs here
  • Feature: MSBuildToolchain creates sprers.eu containing vcvars command for command line building. # . Docs here
  • Feature: Set CMAKE_FIND_PACKAGE_PREFER_CONFIG=ON. #
  • Feature: Include the recipe name when constrained settings prevent install. # . Docs here
  • Feature: Create new sprers.eu for #
  • Feature: New AutotoolsDeps, AutotoolsToolchain helpers in sprers.eu # . Docs here
  • Feature: Experimental that can install a lockfile in the cache, all the binaries or only the recipes with , intended for CI flows. # . Docs here
  • Fix: Fix incorrect output of and in . #
  • Fix: remotes not being loaded for the conan alias command, which was preventing conan alias from working if python_requires is used. #
  • Fix: Improve error message for providing a path instead of full path with filename. #
  • Fix: Rename sprers.euoft:msbuild_verbosity to sprers.eud:verbosity# . Docs here
  • Fix: Simplifications to generator to remove legacy code. #
  • Fix: Add dirty management in download cache, so interrupted downloads doesn’t need a manual cleaning of such download cache. #
  • Fix: Build helper qbs install now installs directly into package_folder. #
  • Fix: Allow arbitrary template structure. #
  • Fix: Restoring the behavior that exports and exports_sources were case sensitive by default. #
  • Fix: Remove default dummy value for iOS XCode signature. #
  • Fix: Do not order Settings lists, so error messages are in declared order. #
  • BugFix: Command conan new accepts short reference with address sign. #
  • Bugfix: Fix profile definitions of env-vars per-package using patterns, not only the package name. #
  • Bugfix: Preserve the explicit value None for SCM attributes if the default is a different value. #
  • Bugfix: Properly detect Amazon Linux 2 distro. #
  • Bugfix: Fix config install not working when .git* folder is in the path. #
  • Bugfix: Fix: Transitive python requires not working with the new syntax. #

(Mar)¶

  • Fix: Allow and to be aliases for and to allow easy migration. #
  • Bugfix: Restoring the behavior that exports and exports_sources were case sensitive by default. #
  • BugFix: Solved issues with already existing packages appearing in . #

(Feb)¶

  • Feature: Add path and repository properties to conan_build_info v2. #
  • Feature: Setting _conan_ as name for buildAgent in conan_build_info. #
  • Feature: Using actual conan version in version for buildAgent in conan_build_info instead of 1.X. #
  • Feature: Add type _conan_ to Conan build info modules. #
  • Feature: Add output in conan info command. #
  • Feature: Forked into , to allow evolution without breaking. #
  • Feature: Use built-in retries in requests lib to retry http requests with _5xx_ response code. #
  • Feature: New lockfile “bundle” feature that can integrate different lockfiles for different configurations and different graphs into a single lockfile bundle that can be used to vastly optimize CI (specially for multiple products), implementing bundle build-order and bundle update operations. # . Docs here
  • Fix: Renamed generator QbsToolchain to QbsProfile. # . Docs here
  • Fix: Renamed default filename of _QbsProfile_ generated file to _conan_toolchain_profile_.qbs. # . Docs here
  • Fix: Renamed Qbs attribute use_toolchain_profile to profile. # . Docs here
  • Fix: Remove extra spaces in flags and colons in path variables. #
  • Fix: conan_v2_error if scm_to_conandata is not enabled. #
  • Fix: CONAN_V2_MODE env-var does not longer alter behavior, only raises errors for Conan incompatibilities # . Docs here
  • Fix: meson : Add target and jobs arguments. # . Docs here
  • Fix: Set sprers.euPlatform with qbs toolchain. #
  • Fix: Remove warnings for old toolchains imports and calls (use new imports and calls. #
  • BugFix: Improve sprers.eu_path for Cygwin. #
  • BugFix: Allow run_in_windows_bash in MSYS/Cygwin. #
  • BugFix: Add some sanity check to avoid a vague error for custom architectures. #
  • BugFix: Fix Apple M1 detection. #
  • Bugfix: Fix repeated , including conflicting versions in profile composition or inclusion that repeats values. #
  • Bugfix: Fixing a CMakeDeps bug with components, not finding the _conan_sprers.eu_ file. #
  • Bugfix: Fix exit code for conan_build_info. #

(Feb)¶

  • Fix: Rename _sprers.eu_ to _sprers.eu_. # . Docs here
  • Fix: Make CMakeDeps generator available in declarative mode #
  • Fix: Make the new Macos subsystem Catalyst lowercase to be consistent with existing subsystems. # . Docs here
  • BugFix: Fix Apple Catalyst flags. # . Docs here

(Jan)¶

  • Feature: Introducing a new section in profiles that allows a more systematic configuration management for recipes and helpers (build helpers, toolchains). Introducing a new cache configuration file that contains configuration definition with the same syntax as in profiles. # . Docs here
  • Feature: Add Apple Catalyst support (as new sprers.eutem) # . Docs here
  • Feature: Add sprers.eu sub-settings for Apple # . Docs here
  • Feature: Provide support for compiler in tools # . Docs here
  • Feature: Specify build modules by the generator in cpp_info. Added backwards compatibility for *.cmake build modules added at global scope, but not for other file extensions. # . Docs here
  • Feature: The sprers.eu, sprers.eu and sprers.euz now accept a new argument strip_root=True to unzip moving all the files to the parent folder when all of them belongs to a single folder. # . Docs here
  • Feature: Add new compiler setting and preliminary support in generator and toolchain. # . Docs here
  • Feature: now takes values for configurations from . # . Docs here
  • Feature: Implement in generator. #
  • Feature: Add _Meson_ build helper. # . Docs here
  • Feature: Add QbsToolchain and a Qbs build helper class (currently working for Mcus, not for Android or iOS). # . Docs here
  • Feature: Add e2k (elbrus) architectures and mcst-lcc compiler # . Docs here
  • Feature: New generator (at the moment is the , that allows custom configurations, like . # . Docs here
  • Feature: Allow custom configurations in method. # . Docs here
  • Fix: Fixed help message in command conan remove –outdated with reference or pattern # . Docs here
  • Fix: Do not define CMAKE_GENERATOR_PLATFORM and CMAKE_GENERATOR_TOOLSET in the file unless the CMake generator is “Visual Studio”. Fix sprers.eu#
  • Fix: Remove spurious ‘-find’ argument to XCodes xcrun tool. #
  • Fix: Update pylint plugin, some fields are now available in the base ConanFile. #
  • Fix: Remove PyJWT deprecation warning by adding explicitly argument. #
  • Fix: CMake’s generator name for Visual Studio compiler uses only the major version. #
  • Fix: Remove nosetests support, now using pytest for the test suite. #
  • Fix: Remove CMAKE_PROJECT_INCLUDE in CMakeToolchain, no longer necessary as the MSVC runtime can be defined with a generator expression in the toolchain. #
  • Fix: Temporarily allow generator for , to allow start using the toolchain for users that depend on that generator. #
  • Fix: Let CMake generator generate code for checking against “ClangCL” msvc toolset. #
  • Fix: Include in the global file generated by MSBuildDeps. #
  • Fix: Change MSBuildDeps file to to avoid collision with a package named “deps”. #
  • Fix: Throw error when the recipe description is not a string. #
  • Fix: Inject build modules after CMake targets are created # . Docs here
  • Fix: Define in the folder (defaulting to “package”), so the test recipe can execute in its method. #
  • Fix: Remove the downloaded file if it doesn’t satisfy provided checksums (modifies sprers.euad). # . Docs here
  • Bugfix: Solved assertion that happened when using that also exist in , and using and #
  • BugFix: Fix SCM user and password by making them url-encoded #
  • Bugfix: Fix bug in definition of variables in . #
  • BugFix: fix -j being passed to _NMake_ in AutotoolsBuildEnvironment. #
  • BugFix: Fix per-package settings exact match for packages without user/channel. #
  • BugFix: Fix detected_architecture() for Apple M1, mapping to (from returned ). #
  • Bugfix: Make prompt names unique when using multiple virtualenv scripts in Powershell. #
  • Bugfix: Fix when _sprers.eu_ is empty and scm_to_conandata is enabled #
  • Bugfix: Removed a reference to deprecated FlagsForFile function in place of current Settings function. #
  • Bugfix: Add test for AutoToolsBuildEnvironment on Apple platforms #
  • Bugfix: Change the rpath_flags flag to always use the comma separator instead of “=”, because the current behaviour causes linker error messages when attempting to cross-compile to Mac OS, and the comma separator is accepted everywhere. #

(Dec)¶

  • Bugfix: Avoid conflict of user custom generators names with new generators. #
  • Bugfix: Fix errors when using and in Windows, due to creation of empty folders in the short-paths storage. #
  • Bugfix: wrong include when not using (using the hardcoded git repo). #
  • Bugfix: Fix pattern case-insensitive in non Windows platforms. #
  • Bugfix: Enabling set_name, set_version for lockfile roo not location. #

(Dec)¶

  • Feature: Generate <pkgname>sprers.eu files for lowercase packages to improve case compatibility. # . Docs here
  • Feature: Add meson cross-build toolchain. #
  • Feature: Temporary acquire write permissions in replace_in_file. #
  • Feature: Update conan new to latest guidelines. #
  • Feature: Deprecate experimental in favor of more generic method. Deprecate toolchains to new method. # . Docs here
  • Feature: Move the and new experimental helpers to the new import. # . Docs here
  • Feature: Move the and new experimental helpers to the new import. # . Docs here
  • Feature: Move the experimental helper to the new import. # . Docs here
  • Feature: Add to list recipes without a remote defined. # . Docs here
  • Feature: Add to list packages without a remote defined. # . Docs here
  • Feature: Add argument to so it can target different packages with same reference (different binary, this can happen with private requirements). # . Docs here
  • Feature: Proof that can be used (as a workaround) to affect the of consumers of that otherwise will not be rebuilt based on changes. # . Docs here
  • Feature: Introduce configuration to allow packaging of Python .pyc files. # . Docs here
  • Feature: Tests for toolchains and Intel compiler. #
  • Feature: Add recipe and package revision to show a complete Conan reference when generating the build_info –v2 id fields. #
  • Feature: Introduce a new BINARY_INVALID mode for more flexible definition and management of invalid configurations. # . Docs here
  • Feature: Add headers with settings and options to HTTP GET requests when searching for packages. #
  • Feature: Preliminary experimental support for toolchains with CMake + Visual + Ninja. #
  • Feature: Allow (experimental) custom configuration of the generator. #
  • Feature: Rename generator to and use the new method. #
  • Feature: Make the to provide variable filenames and messages that include the package name and version, instead of a hardcoded “hello” one. #
  • Feature: Tagged tests and created a sprers.eu to run the tests with pytest skipping the tests using not available tools (cmake, visual studio…). #
  • Feature: Provide correct –pure_c implementation to conan new. #
  • Feature: System package tools can install a list of different packages. # . Docs here
  • Feature: meson toolchain # . Docs here
  • Feature: Add Conan package name and version to Visual Studio generator properties file. #
  • Fix: Remove in the root of the repo, which was useless, without a purpose, but caused issues with other projects importing Conan Python code. #
  • Fix: Make variables defined in cache variables, so they can define directly values defined in . #
  • Fix: Remove cryptography, pyopenssl and idna from OSX requirements in Python. #
  • Fix: Rename the generated file of to so it doesn’t collide with a potential package name and the generator. # . Docs here
  • Fix: Avoid warning in generator importing multiple times the same .props file due to transitive dependencies. #
  • Fix: Set username or password individually in git SCM with ssh. #
  • Fix: When using lockfiles, allow and to compute different options as long as the final evaluated values match the locked ones. #
  • Fix: Make the pure C template to remove both and settings, as described in the docs. #
  • BugFix: Fix linkage to a same global target of different package components in cmake_find_package/_multi generators. # . Docs here
  • Bugfix: Solve crash when using with a short path storage located in another Windows drive unit. #
  • BugFix: Allow lockfiles to be relaxed with the –build argument. # . Docs here
  • Bugfix: Append existing in generator. #
  • Bugfix: Remove correctly short-paths folders in Windows. #

(Nov)¶

  • Feature: Add new CONAN_CMAKE_SYSROOT environment variable to enable the definition of sysroot from environment, without abusing CONAN_CMAKE_FIND_ROOT_PATH. # . Docs here
  • Bugfix: remove definition of sysroot from CONAN_CMAKE_FIND_ROOT_PATH. # . Docs here
  • Bugfix: Bugfix: Solve sprers.eu crash when using short_paths with a short path storage located in another Windows drive unit. Ported from: #

(Nov)¶

  • Bugfix: Fix addition of CMAKE_SYSTEM_NAME for SunOS and AIX >32 bits builds #

(Nov)¶

  • Bugfix: Recent library updates is breaking the constraints in Conan as has a limitation for . This PR constrains version to be less than , so it does not break with requests #

(Nov)¶

  • Fix: Bump _cryptography_ dependency in MacOS to equal or later than #
  • Bugfix: Fix a problem with the function not being called when the recipe loader uses some cached data, which can happen when using lockfiles and with . #
  • Bugfix: Fixed incorrectly handling . #
  • Bugfix: Fixed wrong default in section of sprers.eu. #
  • Bugfix: Do not try to encrypt a None value when using CONAN_LOGIN_ENCRYPTION_KEY environment variable. #

(Oct)¶

  • Feature: Add argument conanfile to pre_download_package and post_download_package hook functions. # . Docs here
  • Feature: Add CONAN_LOGIN_ENCRYPTION_KEY environment variable to obfuscate stored auth token. # . Docs here
  • Feature: Use profile to filter results in the conan search HTML output. #
  • Feature: Changed recommended way to launch test suite, with pytest over nosetests. #
  • Feature: Provide a helper class that encapsulates calling MSBuild. # . Docs here
  • Feature: Download and keep the and in the cache, so they are not affected by different Operating Systems compression and de-compression and uploading is way more efficient. #
  • Feature: Add provides and deprecated fields to conan info output #
  • Feature: Including package revision information in output from conan info (when revisions are enabled). #
  • Feature: Download and keep the conan_sprers.eu in the cache, so they are not affected by different Operating Systems compression and de-compression and uploading is way more efficient. #
  • Feature: Add POC on a toolchain for iOS (using CMake XCode generator). # . Docs here
  • Feature: Add POC on a toolchain for Android (using CMake provided modules). # . Docs here
  • Feature: Allow of a single file # . Docs here
  • Feature: Use Python loggers for Conan output in cli #
  • Fix: Improve permission error message when migrating cache folder. #
  • Fix: Make per-package settings definition complete the existing settings values, not requiring a complete redefinition. #
  • Fix: Avoid unnecessary extra loading of sprers.eu file in the version migrations check. #
  • Fix: Simplified MakeToolchain to remove things that were not checked by tests or unused. #
  • Fix: displayed message when settings of the recipe are constrained. # . Docs here
  • Fix: Set CMAKE_SYSTEM_NAME set to iOS, tvOS or watchOS or Darwin depending on the CMake version. #
  • Fix: Remove duplicate entries while modifying PATH-like environment variables internally. Especially important for Windows where system PATH size is limited by charachers (when using sprers.eu). #
  • Fix: Make default behaviour explicit in search help output. # . Docs here
  • Fix: Automatically add OSX deployment flags in with the value of , unless the values are already defined in environment variables CFLAGS or CXXFLAGS. #
  • Fix: Remove toolset variability from the generator and . #
  • Fix: Component requirement checking now properly handles private and override requirements. #
  • Bugfix: Set default storage_folder to .conan/data in case if storage_path entry fails to be defined by sprers.eu #
  • Bugfix: Fix regression in sprers.eu(output=xxxx) that have a write() method but do not wrap a stream. #
  • Bugfix: Fix local flow (conan install + build) support for and . #
  • Bugfix: Fix forcing to retrieve the remote for evaluation, overwriting what is in the local cache. #
  • Bugfix: Copy symbolic links to directory with deploy generator. # . Docs here

(Oct)¶

  • Fix: Use quotes around the install path, it can contain spaces. #
  • Fix: prefix intel functions with because they are now exposed via tools. Fixes sprers.eu # . Docs here
  • Bugfix: Fix regression introduced in (sprers.eu), with incorrect matches of user/channel for version ranges. #
  • Bugfix: Fix with multiple variables definitions. #
  • Bugfix: Check comparing the host and the build architecture to decide if cross building and set CMAKE_SYSTEM_NAME in the build helper. #

(Oct)¶

  • Feature: Implement real detection of value for compiler. Only enabled in CONAN_V2_MODE, otherwise it would be breaking. #
  • Feature: Added [Depends](sprers.eu) items for every public dependency of conanfiles requires/dependencies. # . Docs here
  • Feature: Instructions on how to run conan tests against a real Artifactory server. #
  • Feature: List cpp_sprers.eu and cpp_sprers.eumes in JSON and Markdown generator. # . Docs here
  • Feature: Add information about components to markdown generator. #
  • Feature: New experimental to generate files (it is multi-config, will generate one specific toolchain file per configuration) for more transparent integration and better developer experience with Visual Studio. # . Docs here
  • Feature: Allow packages that do not declare components to depend on other packages components and manage transitivity correctly, with the new attribute. # . Docs here
  • Feature: Add MacOS 11 (“Big Sur”) support. # . Docs here
  • Feature: Expose intel_installation_path, compilervars, compilervars_dict, and compilervars_command under tools module in order to support usage of the intel compiler. # . Docs here
  • Feature: Allow user-defined generators to be installed and used from the Conan cache. # . Docs here
  • Feature: Add conan remote proposal for cli #
  • Fix: Allow usage of MD5 checksums in FIPS systems that would raise error otherwise. #
  • Fix: Fix capture output when running tests that call the ConanRunner in the conanfile. #
  • Fix: Consider absolute paths when parsing sprers.eu#
  • Fix: Update parallel uploads help message. # . Docs here
  • Fix: Removed check in lockfiles computed from other lockfile that it should be part of it. Users can check the resulting lockfile themselves if they want to. # . Docs here
  • Fix: Extend help message indicating how to run conan export without user/channel. # . Docs here
  • Fix: Conan copy shows better description when using full reference for destination. #
  • Fix: Do not capture output for normal conan run (no logging or testing) when launching processes via ConanRunner
set)_object` to the way of extending `zend_object` (more stable and efficient) (#) (@twose)
* Support Github Actions (@twose) (@matyhtf) 4.4.13RC2 New API

+ Built-in Connection Pool ([examples](sprers.eu)) ([doc](sprers.eu)) (@twose)

Enhancement

+ Support websocket compression on Coroutine\Http\Server (#) (@twose)
+ Support `CURLOPT_PUT`, `CURLOPT_INFILE`, `CURLOPT_INFILESIZE` (#) (@twose)
+ Add error info for bind error (#) (@LeiZhang-Hunter) (@twose)
+ Update error info if socket call failed (it will not warn anymore if not necessary) (1e9b8cbb) (@twose)
+ Move the library to [a separate repo](sprers.eu) (@twose)
+ Support `CURLOPT_IPRESOLVE` (libraryffb4d6) (@twose)
+ Support `CURLOPT_PORT`, `CURLOPT_PROXYPORT`, `CURLOPT_DNS_USE_GLOBAL_CACHE` (library-5a) (@twose)
+ Better `CURLOPT_ENCODING` support (library-7b) (@twose)
+ Support `curl_reset`, `CURLOPT_PROXYTYPE`, `CURLOPT_PROXYUSERNAME`, `CURLOPT_PROXYPASSWORD`, `CURLOPT_PROXYUSERPWD` (libraryce0a04) (@twose)

Fixed

* Fixed gdbinit in docker (missing symbol) (d46dc2d7 (@twose)
* Fixed compression in POST request (#) (@twose)
* Fixed compatibility of `Websocket->push` method (ac7dbcbc) (@twose)
* Fixed `$output` result inconsistency (@Yurunsoft) (#)
* Fixed `Coroutine::fread` (#) (@matyhtf)
* Fixed `Coroutine\Server` can not exit normally # (#) (@twose)
* Fixed memory leak in HTTP2 Server (#) (@twose)
* Fixed `unsafe_event` (#) (@matyhtf)
* Fixed illegal call after server start (#) (@twose)
* Fixed `WebSocket\Frame` can not enable mask (#) (#) (@twose)
* Fixed memory leak of `Process` callback (a7) (@twose)
* Fixed case where minfd is never updated (#)
* Fixed HTTP2 module is always enable (e) (@twose)
* Fixed HTTP2 context double free (#) (f4cdff1b) (@twose)
* Fixed string comparison (#) (97c6c) (9e5eae63) (@twose)
* Fixed HTTP2 pipeline (#) (@twose)
* Fixed coroutine context has been destroyed (#) (@twose)
* Fixed illegal usage of `Process::daemon` (#) (@twose)
* Fixed compilation without zlib (#) (@twose)

Kernel

* Refactor `swoole_(get Overview of

Release date: 24 Apr

For the highlights of this release, see the release notes.

The revision number links will take you to the revision's page on Launchpad. On Launchpad you can view more details of the revision and view diffs of the code modified in that revision.

  • Revision #Tue UTC
    • Merge base ->
  • Revision #Sun
  • Revision #Sun +
    • Fixing build-bot compilation failure on SolarisX86, who has two different incompatible copies of zlib installed: (in /usr and in /usr/local). cmake errorneously chose *.h file from /usr/local/include, while zlib.a/sprers.eu from /usr/lib/. Compilation failed because of a missing symbol gzopen
  • Revision #Sun +
    • Packaging clean-ups for the cassandra and connect engines.
  • Revision #Sun
    • fixes for buildbot. increase the version.
  • Revision #Sun +
    • Fixing compilation failure in Solaris/OpenSolaris error: too many initializers for `tm'
  • Revision #Sun +
    • Removing -Wfatal-errors, as it does not present in the older gcc versions. (e.g. sol in build bot).
  • Revision # [merge] Sat
    • Merge base ->
    • Revision #Fri +
      • Fix missing reset of debug_sync, which could cause subsequent test to fail.
    • Revision #Fri +
      • MDEV - remove incorrect fix, replace with correct one - change default to OFF for innodb_use_fallocate
  • Revision #Sat +
    • Recoding new results (the format changed during Sergei's merge of the connect engine and discovery)
  • Revision # [merge] Sat +
    • Merging from mariaconnect (DEB packaging related fixes)
    • Revision #Sat +
      • Adding mariadb-connect-engine deb package for Debian and Ubuntu
    • Revision #Fri +
      • Removing the connect engine from mariadb-server package, as it needs unixODBC and libxml2 as dependencies
  • Revision #Sat +
    • Fixing a few problems found by Build Bot after merging the CONNECT engine.
  • Revision #Sat +
    • After-fix for Sergei's merge of the connect engine. odbc_sqlite3 failed. Fixed.
  • Revision # [merge] Fri +
    • merge with mariaconnect
    • Revision #Fri +
      • Fix to handle CMake component names with dash in the name, when producing MSI package. Rename connect_engine component back to connect-engine.
    • Revision #Fri +
      • Adding command in the queries that use $PORT
    • Revision #Fri +
      • - Added and
    • Revision # [merge] Fri +
      • - Committing merged changes
      • Revision #Fri +
        • "" did not work. Applying a patch from SerG fixing this problem.
    • Revision #Fri +
      • - Fix problem of mysql connection in TBL tables By adding the option_list='port=$PORT' in the table create statement.
    • Revision #Mon +
      • - Bug: When trace is ON the variable c was used uninitialized in CSORT::Istc
  • Revision #Fri +
    • CONNECT engine fixes after merge. Adaptation to new API, small simplifications and bug fixes
  • Revision #Fri +
    • fix plugin locking/unlocking when assisted discovery fails
  • Revision #Fri +
    • String::append_for_single_quote() should signal OOM condition, just like other String::append() methods do
  • Revision #Fri +
    • bugfix: CREATE .. SELECT should not pass engine defined field options from SELECT clause to CREATE
  • Revision # [merge] Fri +
    • merge
    • Revision #Fri +
      • MDEV - Do not try to unlock an already unlocked mutex.
    • Revision # [merge] Thu +
      • base merge
      • Revision #Thu +
      • Revision # [merge] Wed
        • Merge.
        • Revision # [merge] Wed
          • Merge base -> mwl
        • Revision #Wed
          • Fixed a typo/bug that could lead to wrong selectivity numbers for tables retrieved by range scans.
        • Revision # [merge] Tue
          • Merge base -> mwl
      • Revision #Wed +
        • Changed the client library to only mark memory as THREAD_SPECIFIC if one has called mysql_options() with MYSQL_OPT_USE_THREAD_SPECIFIC_MEMORY
      • Revision #Wed +
        • Fixed compiler warnings and a not critical memory leak
    • Revision # [merge] Wed
      • Merge base ->
      • Revision #Wed +
        • Remove forgotten debug printout in test, was a bit too quick there :-(
      • Revision #Wed +
        • Fix that multi_source tests did not reset @@sprers.eu_pos between tests.
      • Revision # [merge] Wed +
      • Revision #Tue +
        • Fixes for stuff seen in buildbot:
      • Revision #Tue +
        • Makeing rpl_filter for each Master_info.
      • Revision # [merge] Mon
        • Merge mwl -> base
        • Revision #Mon
          • Added comments. Renamed the virtual method middle_point_pos for the class Field to pos_in_interval.
        • Revision # [merge] Mon
          • Merge base -> mwl
        • Revision # [merge] Mon
        • Revision #Sat +
          • Temporarily disable show_sprers.eu
        • Revision # [merge] Fri
          • Merge base->mwl
        • Revision #Fri
          • Fixed bug MDEV The selectivity of a range degenerated into a point never should be set to 0.
        • Revision #Tue
          • Fixed valgrind complain on usage of uninitialized data.
        • Revision #Mon
          • Fixed MDEV Uninitialized field next_equal_field of the Field objects created for the fields of a temporary table could hang the server.
        • Revision #Mon
          • Fixed MDEV Uninitialized field cond_selectivity of the Field objects created for the fields of a temporary table could cause an assertion abort.
        • Revision #Sat
          • Fixed bug MDEV When calculating the selectivity of a range in the function get_column_range_cardinality a check whether NULL values are included into into the range must be done.
        • Revision #Sat
          • Fixed bug MDEV The function was adjusted to be able to aggregate the counters of the merged elements. Before this change it was not possible to guarantee the correctness of the counters passed to the the call-back parameter walk_action. As a result, when some elements of a Unique object were flushed into disk the function passed to merge_walk() as the call-back parameter could return wrong counters of elements. This could lead to building wrong histograms.
        • Revision #Sat
          • Fixed bug MDEV Range analysis of non-indexed columns should handle properly range trees with type == SEL_TREE::MAYBE.
        • Revision #Sat
          • Fixed bug MDEV Valgrind complained on usage of uninitialized memory.
        • Revision #Fri
          • Fixed bugs MDEV and MDEV The values of the column HIST_TYPE from the statistical table sprers.eu_stats were stored in the table and read from the table incorrectly.
        • Revision #Fri
          • Fixed bug MDEV Avoid possible division by 0.
        • Revision #Fri
          • Fixed bug MDEV Don't try to a histogram if it is not read into the cache for statistical data. It may happen so if optimizer_use_condition_selectivity is set to 3. This setting orders the optimizer not use histograms to calculate selectivity.
        • Revision #Thu
          • Fixed bug MDEV When performing the range analysis for a conjunction the function calculate_cond_selectivity_for_table should take in to account that the analysis of some conjuncts may return SEL_ARG::IMPOSSIBLE.
        • Revision #Wed
          • Fixed bug MDEV When calculating selectivity of conditions one should take into account the cases when some tables to be joined are empty.
        • Revision #Wed
          • Fixed bug MDEV Wrong formulas used by the function Histogram::point_selectivity() could result in a negative value of selectivity returned by the function.
        • Revision # [merge] Wed
          • Merge base->mariamwl
        • Revision #Wed
          • Fixed bug MDEV Range analysis of the condition for a non-indexed column may return an impossible range. This must be taken into account.
        • Revision #Mon
          • Fixed bug MDEV The bug was caused a wrong casting.
        • Revision # [merge] Mon
        • Revision #Mon
          • Fixed a valgrind complain on usage of an uninitialized value. It popped up because the latest fix to handle properly null ranges was not complete.
        • Revision #Sun
          • Take into account the number of null values in any used column when calculating selectivity of conditions.
        • Revision #Sun
          • Fixed a valgrind complain on usage of an uninitialized value.
        • Revision # [merge] Sun
          • Merge base -> mwl
        • Revision # [merge] Sat
          • Merge
          • Revision #Sat
            • Added the type of histogram for mwl # Introduced double precision height-balanced histograms.
        • Revision #Sat
          • Fixed several bugs for mwl #
        • Revision # [merge] Tue
          • Merge
          • Revision #Wed
            • Correction for the previous fix.
          • Revision #Wed
            • Fixed some bugs in the function that calculated the selectivity of the table conditions.
          • Revision #Sun
            • Fixed a typo that caused a wrong calculation of the selectivity for pushed down condtions.
        • Revision #Mon
          • Added histogams for table columns.
        • Revision #Mon
    • Revision # [merge] Wed +
    • Revision #Tue +
      • don't compare table names to detect temporary tables (also fixes a bug with unaccounted table names like #sql, see drop and alter_table tests)
    • Revision #Tue +
      • remove a sleep from the test
    • Revision #Mon +
      • remove numerous #ifdef HAVE_PSI_TABLE_INTERFACE simplify ha_table_share_psi()
    • Revision # [merge] Mon +
      • base merge
      • Revision # [merge] Sun +
        • merge
        • Revision #Sun +
        • Revision #Fri +
          • Increase default value of max_binlog_cache_size and max_binlog_stmt_cache_size to ulonglong_max. This fixes that by default LOAD DATA INFILE will not generate the error: "Multi-statement transaction required more than 'max_binlog_cache_size' bytes of storage"
        • Revision #Fri +
          • complier warnings. hide the redundant condition under #ifdef (because only there it makes any sense)
        • Revision # [merge] Fri +
          • merge
          • Revision # [merge] Thu +
          • Revision #Sat +
            • MDEV [PATCH] Buffer overruns and use-after-free errors
          • Revision #Sat +
            • MDEV MariaDB server crash with signal 11
          • Revision #Mon +
            • If a range tree has a branch that is an expensive constant, currently get_mm_tree skipped the evaluation of this constant and icorrectly proceeded. The correct behavior is to return a NULL subtree, according to the IF branch being fixed - when it evaluates the constant it returns a value, and doesn't continue further.
          • Revision #Thu +
            • Update tests results, mysql-test/r/sprers.eu
        • Revision #Sun +
          • MDEV : MariaDB does not start if bind-address gets resolved to more than single IP address.
        • Revision #Sat +
          • MDEV - Support FusionIO/directFS atomic writes
        • Revision #Sat +
          • MDEV - Support FusionIO/directFS atomic writes
        • Revision #Thu +
        • Revision #Thu +
          • fix have_debug_sprers.eu to be more robust (debug_sync value can have single quotes)
        • Revision #Thu +
          • MDEV Assertion `status_sprers.eu_used == 0' fails in virtual THD::THD()
        • Revision #Thu +
          • MDEV Warnings/errors while compiling with clang
        • Revision # [merge] Wed +
          • Merge ->
          • Revision #Mon +
            • MDEV mariadb using more memory than MySQL for an inefficient query - Let index_merge allocate table handlers on quick select's MEM_ROOT, not on statement's MEM_ROOT. This is crucial for big "range checked for each record" queries, where index_merge can be created and deleted many times during query exection. We should not make O(#rows) allocations on statement's MEM_ROOT.
          • Revision #Fri +
            • MDEV Unexpected results when selecting on information_schema - When converting a subquery to a semi-join, propagate OPTION_SCHEMA_TABLE.
      • Revision # [merge] Sat +
      • Revision #Sat +
        • MDEV IF (NOT) EXIST clauses for ALTER TABLE (MWL #). Syntax modified to allow statements: ALTER TABLE ADD/DROP COLUMN ALTER TABLE ADD/DROP INDEX ALTER TABLE ADD/DROP FOREIGN KEY ALTER TABLE ADD/DROP PARTITION ALTER TABLE CHANGE COLUMN ALTER TABLE MODIFY COLUMN DROP INDEX to have IF (NOT) EXISTS options. Appropriate implementations added to mysql_alter_table().
      • Revision #Fri +
        • MDEV multiple use locks (GET_LOCK) in one connection. The patch contributed by Konstantin Osipov applied. Native comments: Implement multiple user-level locks per connection.
      • Revision #Wed +
        • portability fixes for mysql-test
      • Revision #Wed +
        • Linking problem on Windows
      • Revision #Tue +
        • MDEV Semisync plugins to link statically into MariaDB
      • Revision #Tue +
      • Revision #Tue +
        • remove old workaround for replicating from old MySQL and alpha trees.
      • Revision #Tue +
        • prefer static inline functions to macros. avoid unnecessary strlen()'s
      • Revision #Tue +
        • put status variables in the proper pluginname_ scope (but support the scopeless mysql style too). always output status/system variables in the correct lettercase
      • Revision #Tue +
        • MDEV show plugins soname 'xxx' and INFORMATION_sprers.eu_PLUGINS table with condition pushdown for I_sprers.eu_PLUGINS and a new status variable to cound successful dlopen's
      • Revision #Tue +
        • MDEV table attributes with sysvar as a default value
      • Revision #Tue +
        • create sys_var::val_str(), sys_var::val_int(), sys_var::val_real(). Change Item_func_get_system_var::val_xxx functions to use that.
      • Revision #Tue +
      • Revision #Tue +
        • error messages: name the storage engine explicitly, instead of "used storage engine" and similar changes.
      • Revision #Tue +
        • post-review comments and other minor edits
      • Revision #Tue +
        • add sequence and sql_discovery suites to the default list. implement
      • Revision #Tue +
        • optimize discovery for cases when the storage engine is known in advance
      • Revision #Tue +
      • Revision #Tue +
        • assisted discovery in federatedx
      • Revision #Tue +
      • Revision #Tue +
        • fix internal plugin names
      • Revision #Tue +
        • dead code, remove unused argument
      • Revision #Tue +
        • remove HA_CREATE_INFO::frm_only - it's internal server flag, not part of the SE API, and, again, mutually exclusive with C_ORDINARY_CREATE and C_CREATE_SELECT.
      • Revision #Tue +
      • Revision #Tue +
        • TABLE_SHARE::free_frm_image() method to free the memory allocated by the same allocator as in TABLE_SHARE::read_frm_image()
      • Revision #Tue +
        • limit frm size, when reading it in memory
      • Revision #Tue +
        • cleanup: merge two mutually dependent function arguments into one, eliminating reduncancy and a possibility of setting them to a pair of invalid values.
      • Revision #Tue +
      • Revision #Tue +
        • * frm extra2 segment. * persistent table versions in the extra2 * ha_archive::frm_compare using TABLE_SHARE::tabledef_version * distinguish between "important" and "optional" extra2 frm values * write engine-defined attributes (aka "table options") to extra2, not to extra, but still read from the old location, if they're found there.
      • Revision #Tue +
      • Revision #Tue +
      • Revision #Tue +
      • Revision #Tue +
      • Revision #Tue +
        • test_sql_discovery storage engine
      • Revision #Tue +
        • discovery using sql CREATE TABLE statement
      • Revision #Tue +
        • discover of table non-existance on drop
      • Revision #Tue +
        • CREATE TABLE and frm-less discovering engines. Now CREATE TABLE does not write the frm file on disk, if the engine can discover it
      • Revision #Tue +
        • split mysql_create_frm() in create_frm_image() and writefrm()
      • Revision #Tue +
        • write frm pieces contiguously, don't align them on a 4K boundary (in partition_sprers.eu frm size goes down 16 times :)
      • Revision #Tue +
        • Instead of creating and writing frm into a file peacewise (allocating and freeing buffers on the way), allocate one frm buffer, prepare the frm image completely in memory, and then write it down.
      • Revision #Tue +
        • cleanup frm creation: * comments * cosmetic changes, *(ptr+5) -> ptr[5] * a couple of trivial functions -> inline * remove unused argument from pack_header() * create_frm() no longer creates frm file (the function used to prepare and fill a memory buffer and call my_create at the end. Now it only prepares a memory buffer). Renamed accordingly. * don't call pack_screen twice, go for a smaller screen area in the first attempt * remove useless calls to check_duplicate_warning() * don't write unireg screens to .frm files * remove make_new_entry(), it's basically dead code, always calculating and writing into frm the same string value. replace the function call with the constant string.
      • Revision #Tue +
        • if discovery in recover_from_failed_open fails, don't bother to reopen
      • Revision #Tue +
        • HA_ERR_TABLE_DEF_CHANGED support in ha_archive
      • Revision #Tue +
        • remove dd_frm_type(), dd_frm_storage_engine(), dd_check_storage_engine_flag() from everywhere - now RENAME, SHOW FULL TABLES, and TRUNCATE work with discovery. improve error messages in truncate
      • Revision #Tue +
        • open_table_def() no longer probes for the old pre table file names. SELECT * FROM `t-1` will no longer read "tfrm" file, use `#mysql50#t-1` for this.
      • Revision #Tue +
      • Revision #Tue +
        • ha_create_table: remove unused argument
      • Revision #Tue +
        • init_from_binary_frm_image: verify that we don't read beyond the image buffer
      • Revision #Tue +
        • move writing of the frm into init_from_binary_frm_image()
      • Revision #Tue +
        • rename a handler method to more precisely reflect what kind of a hack it does
      • Revision #Tue +
        • handlerton::discover_table_names() can not discover only "unknown" tables. Duplicates are possible - deal with them.
      • Revision #Tue +
        • fix mysql_rm_table_no_locks() not to use dd_frm_type, because the frm file may not exist (the table exists only in the engine).
      • Revision #Tue +
        • single table discovery: handlerton::discover_table() method. fixes for need_full_discover_for_existence mode
      • Revision #Tue +
        • remove ha_create_table_from_engine() replace enum read_frm_op with a bitmap flags. remove always-unused 'error' argument of get_table_share
      • Revision #Tue +
        • handlerton::discover_table_existence() method
      • Revision #Tue +
        • convenience helpers for get_table_share() and tdc_open_view(). Pass db and table_name into a function instead of the table_list, when only db and table name are needed.
      • Revision #Tue +
        • * remove ha_check_if_table_exists() and get_table_share_with_discover(). * rename check_if_table_exists() -> table_exists() and remove unneeded arguments
      • Revision #Tue +
        • revert " revision-id: [email protected] committer: [email protected] branch nick: work-mariaCREATE-merge timestamp: Wed + Bugfix: New table creation/renaming block added if old encoded table present " the old behavior was less inconsistent than the new one. In the new one the error message was sometimes different (under LOCK TABLES e.g.), and there were race conditions (if this CREATE happened when a concurrent ALTER has renamed the old table away but haven't put the new table in place)
      • Revision #Tue +
        • rename open_binary_frm() to TABLE_SHARE::init_from_binary_frm_image() simplify open_table_def()
      • Revision #Tue +
        • Don't hold LOCK_open over open_table_def/open_binary_frm call
      • Revision #Tue +
        • moving LOCK_open into get_table_share()
      • Revision #Tue +
        • * don't use numbers for open_table_error codes, use an enum. * print "table doesn't exist in engine" when a table doesn't exist in the engine, instead of "file not found" (if no file was involved) * print a complete filename that cannot be found ('sprers.eu', not 't1') * it's not an error for a DROP if a table doesn't exist in the engine (or some table files cannot be found) - if the DROP succeeded regardless
      • Revision #Tue +
        • don't use I_S constants for open_table_def and get_table_share, have a specially defined enum with clearly named values
      • Revision #Tue +
        • remove dead HAVE_CRYPTED_FRM code and now-unused read_string() function
      • Revision #Tue +
        • make the open_table_def to read the whole frm in memory and let open_binary_frm to parse it from the buffer, not a file. this avoids jumping back in forth in the frm file, and doing intermediate buffer mallocs.
      • Revision #Tue +
        • file-based table discovery for discovering storage engines
      • Revision #Sun +
        • find_files(): don't sort files in my_dir(), sort table names after all engines have discovered their tables
      • Revision #Sun +
        • * use Dynamic_array<> instead of List<> for the list of found tables in sql_sprers.eu * make find_files static
      • Revision #Sun +
        • remove #if MYSQL_VERSION_ID there is no "upstream" OQGraph version that needs to stay compatible with different MariaDB releases.
      • Revision #Sun +
        • * move bas_ext from the handler to the handlerton * provide a default bas_ext value of the empty list
      • Revision #Sun +
        • remove handlerton::find_files (new implementation is coming), and handlerton::table_exists_in_engine (new implementation is coming), and handlerton::license (redundant)
      • Revision #Sun +
        • small cleanup - remove prehistoric DB_TYPE_xxx values
      • Revision #Sun +
        • remove long time obsolete xyz_ci() copies of xyz() macros
      • Revision #Sun +
        • simpler fix for mysql bug #, no need to write an error interceptor, when the function has a dedicated flag for this very use case.
      • Revision #Sun +
      • Revision #Sun +
        • s/st_ha_create_information/HA_CREATE_INFO/
      • Revision #Sun +
      • Revision #Sun +
        • bugfix: MYSQL_THDVAR_STR plugins with PLUGIN_VAR_MEMALLOC didn't work
      • Revision #Sun +
        • macro do {} while(0) safety
      • Revision #Sun +
        • my_dir() cleanup * replace pointer acrobatics with a struct * make sorting explicit: MY_DONT_SORT -> MY_WANT_SORT (if you want something to be done - say it. fixes all places where my_dir() was used without thinking) * typo s/number_off_files/number_of_files/ * directory_file_name() doesn't need to be extern * remove #ifdef BORLANDC * ignore '.' and '..' entries
      • Revision #Sun +
      • Revision #Sun +
        • mtr bug: files outside of both the suite dir and the overlay dir, were treated as coming from the overlay.
      • Revision #Sun +
        • remove the service for installing the closed-source mysql thread pool plugin
      • Revision #Sun +
        • consistency in declaring service symbols
      • Revision #Sun +
      • Revision #Sun +
      • Revision #Sun +
        • split THD::make_lex_string() in two
      • Revision #Sun +
      • Revision #Sun +
        • remove ULL() and LL(), because they're totally unnecessary and sometimes harmful (used with expressions)
      • Revision #Tue +
        • Fix buildbot failure in show_sprers.eu: mysqltest: At line query 'show explain for $thr2' failed: Target is not running an EXPLAINable command
      • Revision # [merge] Sun
        • Merge >base
        • Revision #Fri +
        • Revision #Fri +
        • Revision # [merge] Thu
          • Merge >
          • Revision # [merge] Wed
            • Merge.
            • Revision #Fri
              • Fixed bug MDEV In some cases, when using views the optimizer incorrectly determined possible join orders for queries with nested outer and inner joins. This could lead to invalid execution plans for such queries.
        • Revision # [merge] Wed
          • Merge
          • Revision #Wed
            • Fixed bug MDEV (bug #). This bug was introduced by the patch for WL# If the memory allocated for the tree to store unique elements to be counted is not big enough to include all of them then an external file is used to store the elements. The unique elements are guaranteed not to be nulls. So, when reading them from the file we don't have to care about the null flags of the read values. However, we should remove the flag at the very beginning of the process. If we don't do it and if the last value written into the record buffer for the field whose distinct values needs to be counted happens to be null, then all values read from the file are considered to be nulls and are not counted in. The fix does not remove a possible null flag for the read values. Rather it just counts the values in the same way it was done before WL #
      • Revision # [merge] Wed +
        • merge
        • Revision # [merge] Wed +
          • merge
          • Revision # [merge] Tue +
            • merge
            • Revision # [merge] Tue +
              • merge
              • Revision #Wed +
                • add 'plugins' suite - empty, but the line will work on all branches.
              • Revision #Tue +
                • MDEV Server crashes in get_point on a query with Area, AsBinary, MultiPoint. Need to check if the number of points is 0 for the polygon.
              • Revision #Tue +
                • MDEV Assertion `n_linear_rings > 0' fails in Gis_polygon::centroid_xy. Forgotten DBUG_ASSERT should be replaced with the 'return error'.
              • Revision #Mon +
                • MDEV fix. Item_default_value inherited form Item_field so should create temporary table field similary.
              • Revision #Mon +
                • MDEV geometry query crashes server. Additional fixes for possible overflows in length-related calculations in 'spatial' implementations. Checks added to the ::get_data_size() methods. max_n_points decreased to occupy less 2G size. An object of that size is practically inoperable anyway.
              • Revision #Mon +
                • MDEV Assertion `0' fails in make_sortkey with GROUP_CONCAT, MAKE_SET, GROUP BY
              • Revision #Sun +
                • MDEV geometry query crashes server. The bug was found by Alyssa Milburn. If the number of points of a geometry feature read from binary representation is greater than 0x, then the (uint32) (num_points * 16) will cut the higher byte, which leads to various errors. Fixed by additional check if (num_points > max_n_points).
          • Revision #Tue +
            • GEOMETRYCOLLECTION EMPTY handling fixed. The get_mbr() method shouldn't return the error, rather an invalid MBR in this case.
          • Revision #Tue +
          • Revision #Fri +
            • MDEV geometry function equals hangs forever. The Geometry::get_mbr() function can return an error on a bad data. We have to check for that and act respectively.
          • Revision # [merge] Thu +
            • Merge
            • Revision #Thu +
              • MDEV Crash inside mi_killed_in_mariadb() with myisammrg - Set MI_INFO::external_ref for MyISAM tables that are parts of myisamMRG table.
          • Revision #Wed +
            • MDEV Valgrind warnings (Conditional jump or move depends on uninitialised value) in remove_eq_conds on time functions with NULL argument
          • Revision #Mon +
            • MDEV Assertion `scale <= precision' fails in strings/decimal.c
          • Revision #Sun +
            • MDEV Server crashes in Protocol_text::store, stack smashing detected
          • Revision #Sun +
            • MDEV Assertion `maybe_null && item->null_value' fails in make_sortkey on CASE with different return types, GROUP_CONCAT, GROUP BY
        • Revision #Tue +
          • MDEV Support at least 48 utf8 characters in username in server and PAM
        • Revision #Tue +
          • fix @@external_user variable
        • Revision #Mon +
        • Revision #Wed +
        • Revision #Tue +
          • extend check_global_access() to avoid my_error when it's not needed (in INFORMATION_SCHEMA).
        • Revision #Tue +
        • Revision #Tue +
          • MDEV - get_tty_password() does not work if input redirection is used.
        • Revision #Mon +
          • Patch by Ian Good for MDEV mysqlbinlog output ambiguous escaping
        • Revision #Sun +
          • MDEV Assertion `cmp_items[(uint)cmp_type]' fails in sql/item_sprers.eu
        • Revision #Thu +
        • Revision #Thu +
          • OPTION is now a valid identifier (not a reserved word)
        • Revision #Thu +
          • MDEV : sprers.euion_rename_longfilename fails on eCryptFS Adding an include file which checks whether long names are supported
        • Revision #Wed +
          • MDEV is slower than because of many str_to_datetime calls
        • Revision #Mon +
          • fix innodb failures on solaris
        • Revision #Tue +
          • Fix clang warning (suggest parentheses)
        • Revision #Tue +
          • MDEV : do not copy sql_sprers.eu and sql_yacc.h from unpacked source tarball into build directory, if usable bison is installed on the build machine.
        • Revision #Tue +
          • MDEV : func_math test fails, when clang compiler is used.
        • Revision #Wed +
          • MDEV : when autodetecting default client charset on Windows, fallback to GetACP() whenever GetConsoleCP() returns 0 (i.e appkication does not have a console , which is the case for GUI apps, Windows services etc)
        • Revision #Mon +
          • The i specific code improving character set conversion on the ASCII range was not enabled on x86_64 machines. Enabling it. Gives up to 18 times conversion performance improvement.
        • Revision # [merge] Sun +
        • Revision #Fri +
          • MDEV Test case sprers.eupack fails on ppc32 (only)
        • Revision #Fri +
          • MDEV auth_socket to build on OpenBSD / Bitrig
        • Revision # [merge] Fri +
          • merge with XtraDB as of Percona-Serverrel
          • Revision #Fri +
            • sprers.eu
        • Revision #Wed +
          • hack in dependencies to imitate mysql-*.rpm even better
        • Revision #Wed +
          • MDEV rpm scriptlet chown command dangerous
        • Revision #Tue +
          • MDEV semisync_master + temporary tables causes memory leaks
        • Revision #Wed +
          • TODO geometry query crashes server. The bug was found by Alyssa Milburn. If the number of points of a geometry feature read from binary representation is greater than 0x, then the (uint32) (num_points * 16) will cut the higher byte, which leads to various errors. Fixed by additional check if (num_points > max_n_points).
        • Revision #Tue +
          • Fix for assert found by mysql-test-run
        • Revision #Tue +
          • Fixed issue with LOCK TABLE + ALTER TABLE ENABLE KEYS + SHOW commands.
        • Revision #Mon +
          • Fix wrong install location for DEB supportfiles.
        • Revision # [merge] Sat
          • Merge
          • Revision #Sat
            • Fixed bug MDEV This bug is a regression bug. The regression was introduced by the patch for MDEV, that tried to weaken the condition when a ref access with an extended key can be converted to an eq_ref access. The patch incorrectly formed this condition. As a result, while improving performance for some queries, the patch caused worse performance for another queries.
        • Revision #Fri +
          • MDEV : export additional functions mysql_get_timeout_value(),mysql_get_timeout_value_ms(), mysql_get_socket() from shared client library. They are documented as part of async API.
        • Revision #Fri
        • Revision # [merge] Fri +
          • Automatic merge
          • Revision #Fri +
            • Fixed bug MPDEV / Bug # - Querying myisam table metadata may corrupt the table.
          • Revision #Thu +
            • Added test case for bug in replace with replication that existed in MySQL Replace with an auto_increment primary key and another unique key didn't replicate correctly with REPLACE
          • Revision #Thu +
            • Added support for in mysqld_safe. Trivial cleanup
        • Revision #Fri +
          • Fix compile error when building with DBUG, but without DEBUG_SYNC.
        • Revision # [merge] Fri +
          • Merge >
          • Revision #Fri +
            • Fix compile error on windows in fix for MDEV
          • Revision # [merge] Thu
            • Merge
            • Revision #Thu
              • Fixed bug MDEV Do not include BLOB fields into the key to access the temporary table created for a materialized view/derived table. BLOB components are not allowed in keys.
        • Revision # [merge] Thu +
          • merge with XtraDB as of Percona-Serverrel
          • Revision #Thu +
            • sprers.eu
        • Revision # [merge] Thu +
          • > merge
          • Revision # [merge] Thu +
          • Revision #Thu
            • Fixed a compile error for some platform.
          • Revision #Sun
            • Fixed bug MDEV The function remove_eq_cond removes the parts of a disjunction for which it has been proved that they are always true. In the result of this removal the disjunction may be converted into a formula without OR that must be merged into the the AND formula that contains the disjunction. The merging of two AND conditions must take into account the multiple equalities that may be part of each of them. These multiple equality must be merged and become part of the and object built as the result of the merge of the AND conditions. Erroneously the function remove_eq_cond lacked the code that would merge multiple equalities of the merged AND conditions. This could lead to confusing situations when at the same AND level there were two multiple equalities with common members and the list of equal items contained only some of these multiple equalities. This, in its turn, could lead to an incorrect work of the function substitute_for_best_equal_field when it tried to optimize ref accesses. This resulted in forming invalid TABLE_REF objects that were used to build look-up keys when materialized subqueries were exploited.
          • Revision #Thu
            • Fixed bug MDEV This bug in the legacy code could manifest itself in queries with semi-join materialized subqueries. When a subquery is materialized all conditions that are imposed only on the columns belonging to the tables from the subquery are taken into sprers.eu code responsible for subquery optimizations that employes subquery materialization makes sure to remove these conditions from the WHERE conditions of the query obtained after it has transformed the original query into a query with a semi-join. If the condition to be removed is an equality condition it could be added to ON expressions and/or conditions from disjunctive branches (parts of OR conditions) in an attempt to generate better access keys to the tables of the query. Such equalities are supposed to be removed later from all the formulas where they have been added to. However, erroneously, this was not done in some cases when an ON expression and/or a disjunctive part of the OR condition could be converted into one multiple equality. As a result some equality predicates over columns belonging to the tables of the materialized subquery remained in the ON condition and/or the a disjunctive part of the OR condition, and the excuter later, when trying to evaluate them, returned wrong answers as the values of the fields from these equalities were not valid. This happened because any standalone multiple equality (a multiple equality that are not ANDed with any other predicates) lacked the information about equality predicates inherited from upper levels (in particular, inherited from the WHERE condition). The fix adds a reference to such information to any standalone multiple equality.
          • Revision # [merge] Wed
            • Merge.
            • Revision #Wed
              • Fixed bug MDEV The wrong result set returned by the left join query from the bug test case happened due to several inconsistencies and bugs of the legacy mysql code.
          • Revision #Wed +
          • Revision # [merge] Tue
            • Merge.
            • Revision #Thu
              • Fixed bug MDEV This bug happened because the executor tried to use a wrong TABLE REF object when building access keys. It constructed keys from fields of a materialized table from a ref object created to construct keys from the fields of the underlying base table. This could happen only when materialized table was created for a non-correlated IN subquery and only when the materialized table used for lookups. In this case we are guaranteed to be able to construct the keys from the fields of tables that would be outer tables for the tables of the IN subquery. The patch makes sure that no ref objects constructed from fields of materialized lookup tables are to be used.
          • Revision #Mon +
          • Revision #Mon +
        • Revision #Thu +
        • Revision # [merge] Thu +
          • merge with mysql minus few incorrect or not applicable changesets
          • Revision #Mon +
            • Approved by Jimmy and Inaam. rb#
        • Revision #Wed +
          • MDEV Test sprers.eu_rotate_purge_deadlock has incorrect preamble
        • Revision #Sun +
          • Compilation : fix oqgraph's system check, in case where boost header aren't in standard include directory.
        • Revision #Thu +
        • Revision #Thu +
          • MDEV : Enable Ctrl-C handler when reading password, on Windows.
        • Revision #Wed +
          • MDEV : ensure mysql client's beep works on all Windows systems. Use MessageBeep, which employs sound card, rather than system speaker. The secondary benefit is that one can use volume control for this sound (see MySQL's Bug #)
        • Revision #Thu +
          • MDEV missing constraints for spatial column types. Checks added to return and error when inappropriate geometry type is stored in a field.
        • Revision #Tue +
          • MDEV - Use kqueue for threadpool implementation on more BSD variants than just FreeBSD or OSX - i.e NetBSD, OpenBSD, DragonFly, etc.
        • Revision #Mon +
        • Revision #Mon +
          • MDEV Export additional symbols from RPMs , compatibly to distribution RPMs. -Ensure that symbols listed in CLIENT_API_EXTRA are not thrown away by the linker. -Add THR_KEY_mysys to this list, because Fedora18 exports it.
        • Revision #Fri +
          • MDEV Test cases query_cache and query_cache_size_basic fail on 32 bit ppc and s
        • Revision #Fri +
          • make rpm packages to respect CMAKE_INSTALL_PREFIX
        • Revision #Mon +
          • MDEV : Export additional symbols when building RPM, to enable both recompiling mysqli or odbc from sources in addition to drop-in replacement functionality.
        • Revision #Sun +
          • MDEV - Converted rdiff files to uniform MDEV - Modifed tests and result files to use explicit column lists in INSERT and SELECT statements
        • Revision #Wed +
          • MDEV Assertion (group->connection_count > 0) fails with Percona server in replication test.
      • Revision #Tue +
        • MDEV : Fix maintainer compilation flags.
      • Revision # [merge] Mon +
        • Automatic merge
        • Revision # [merge] Mon +
          • Automatic merge
          • Revision #Mon +
            • Speed up connection time: -Change my_rnd() slightly to make it safer if two threads use it at the same time. -Avoid some sprintf and strmov in vio. -Changed thread_count to be automaticly incremented (instead of under LOCK_thread_count). -Thread cache now uses LOCK_thread_cache instead of LOCK_thread_count. -Moved delete thd out from LOCK_thread_count. -Save some mysql_cond_broadcast(&COND_thread_count) calls. -Removed call to getsockname() during connect. -Initialize random generator without locks.
        • Revision #Wed +
          • Speed up connection time: -Change my_rnd() slightly to make it safer if two threads use it at the same time. -Avoid some sprintf and strmov in vio. -Changed thread_count to be automically incremented (instead of under LOCK_thread_count). -Thread cache now uses LOCK_thread_cache instead of LOCK_thread_count. -Moved delete thd out from LOCK_thread_count. -Save some mysql_cond_broadcast(&COND_thread_count) calls. -Removed call to getsockname() during connect. -Initialize random generator without locks.
      • Revision #Mon +
        • MDEV Race in binlog checkpointing during server shutdown.
      • Revision #Sun +
        • Patch by Hartmut Holzgraefe
      • Revision #Sun +
        • Don't show sphinx_error% as this may be a different set of variables in different SPHINX releases
      • Revision #Tue +
        • Performance improvements in "from latin1" and "to utf8" conversion. Mini-benchmarking demonstrates up to 10% improvement in latin1->utf8 conversion.
      • Revision #Thu +
        • Fix buildbot failure: Backport the code that runs sprers.eu only when Cassandra is present and running from to base.
      • Revision # [merge] Wed +
        • Automatic merge
        • Revision #Fri +
          • Fixed BUG# Can't delete rows from MEMORY table with HASH key
      • Revision #Tue +
        • [NOT] EXISTS to IN transformation.
    • Revision #Thu +
      • add P_S instrumentation to file operations in mf_iocache2.c
    • Revision #Thu +
      • MDEV [PATCH] "Unused variable" warnings in the tarball
    • Revision #Wed +
      • MDEV MYSQL_VERSION_sprers.eu_VERSION_MINOR not replaced
    • Revision #Thu +
      • MDEV Warnings/errors while compiling with clang
  • Revision #Fri +
    • small mtr cleanup and move the connect suite into storage/connect/
  • Revision #Thu +
    • Fixing check_access() not to be inlined in case of embedded server, to make connect engine work with embedded server dynamically.
  • Revision #Wed +
    • Dashed name did not work on Windows. Renaming dashed name to underscored name:
  • Revision #Wed
    • Fix MSI package creation for connect engine. Also, do not use /MP option when compiling CONNECT (incompatible with COM #import directive) Also, fix ODBC_LIBRARY to be a list on Windows, not string with spaces inside.
  • Revision #Tue +
    • Adding RPM related definitions to create a separate RPM package (e.g. MariaDBfcx86_sprers.eu) with automatic installation of libxml2 and unixODBC as dependencies.
  • Revision #Sat +
    • - Fix use of s->db_plugin in GetSubTable for release versions
  • Revision #Fri +
    • - Extend the TBL type to support sub-tables of any engines. Not CONNECT sub-tables are accessed via the MySQL API like the MYSQL CONNECT tables.
  • Revision #Wed +
    • - Add support of NULLs for file table columns. Update CONNECT version number and date.
  • Revision #Wed +
    • - Add routine to test index equality
  • Revision #Tue +
    • - All the processing of creating, dropping, modifying indexes was redesigned. The code was a legacy from the first versions of the XDB engine dating and was not working anymore with recent versions of MySQL/MariaDB. A patch in create had been added but is was unsatisfying, recreating all indexes on any alter statement and sometimes doing nothing when it should have. This is a major update to be tested for stability. It was in most important cases et all current tests pass with this new version
  • Revision #Sun +
    • - Prepar using indexes in MAP mode (not used yet)
  • Revision #Sat +
    • - Bug fix: Variable rc could used uninitialized when tracing is on.
  • Revision #Fri +
    • - Do not check columns name length of dbf catalog tables. They return info on a dbf table but are not dbf tables.
  • Revision #Thu +
    • - Update some DBUG_PRINT to avois warning on Linux
  • Revision #Thu +
    • - Do not accept creating XML2 tables when libxml2 is not available
  • Revision #Thu +
    • Adding security tests for "ALTER TABLE t1 FILE_NAME='xxx'"
  • Revision #Thu +
    • - Commit added test on TBL tables + update sprers.eu
  • Revision #Wed +
    • - Block creating tables with auto_incremented colummns (not supported) + allow nullable columns for TBL tables
  • Revision #Wed
    • Fixed a problem in sprers.eu failures when run in a non-Paris time zone.
  • Revision #Tue +
    • Adjusting test according to the recent change: Fixing file extension from upper case to lower case.
  • Revision #Tue +
    • - Fix making default file name with lower case type
  • Revision #Tue +
    • - Update test results impacted by using type in lower case
  • Revision #Tue +
    • - Fold type name to lowercase when used for default file type
  • Revision # [merge] Tue +
    • - Commit pulled changes
    • Revision #Tue +
      • Respect the server options when dealing with tables having FILE_NAME.
    • Revision #Tue +
      • Respect the server option when working with tables having the FILE_NAME='xxx' table option.
  • Revision #Tue +
    • - Comment out the last commited change
  • Revision # [merge] Tue +
    • - Commit merged changes
    • Revision #Mon +
      • Adding FILE privilege tests for TABLE_TYPE=INI
    • Revision #Mon +
      • Adding file privilege tests for TABLE_TYPE=VEC
    • Revision #Mon +
      • Fixing a crash in the latest push from Olivier.
  • Revision #Tue +
    • - Commit change pulled from Linux virtual machine (fixing typo in Win32 part)
  • Revision #Tue +
    • - Fix huge vec tables bug. This was not yet completely implemented for Unix.
  • Revision #Mon +
    • - Fix bugs with VEC tables header and empty file making for HUGE tables. Adding tests of MAPPED and HUGE tables in sprers.eu
  • Revision #Sat +
    • - Add some warnings to Create Table process
  • Revision #Fri +
    • Adding tests for FILE privilege for TABLE_TYPE=DIR
  • Revision #Fri +
    • Adding secutiry tests for TABLE_TYPE=MySQL
  • Revision #Fri +
    • Adding GRANT tests for ODBC
  • Revision #Fri +
  • Revision #Fri +
    • Adding tests for SQLite3 ODBC Driver
  • Revision #Fri +
    • Adding grant tests for TABLE_TYPE=xml
  • Revision # [merge] Fri +
    • Merge from the latest mariadb
    • Revision #Thu +
      • MDEV Invalid code in fts_savepoint_release() in InnoDB
    • Revision #Thu +
      • Fixing test failure in the previous commit (utf16le merge from MySQL). Recording new correct results in mysql-test/r/func_encrypt_sprers.eu
    • Revision #Thu +
      • Merging utf16le from MySQL
    • Revision #Tue +
      • MDEV Threadpool - add thd_wait_begin/thd_wait_end to the network IO functions
    • Revision # [merge] Fri +
      • Automatic merge
      • Revision #Fri +
        • Fix for MDEV, backport from committer: Sergei Golubchik <[email protected]> fix for a valgrind builds. my_alloca() cannot have MY_THREAD_SPECIFIC, because can be used outside of the THD context.
      • Revision #Thu +
        • MDEV Make multi-column non-top level subqueries to be executed via index (index/unique subquery) instead of single_select_engine
      • Revision #Tue +
        • set THR_THD key and current_thd=NULL as early as possible, otherwise safemalloc and my_malloc_size_cb_func will use current_thd before it's defined
  • Revision #Fri +
    • Adding tests for the bug "TABLE_TYPE=ini does not clear memory between CREATE TABLEs" fixed by Olivier.
  • Revision #Fri +
    • - Fix problems related to table file names when not specified: Split unspecified VEC tables are no more allowed. Empty XML files are now accepted. Separate index files are now depending upon the SEPINDEX option and not allowed when file name is not specified. DROP now can erase table and index file.
  • Revision #Thu +
    • - Commit revision pushed from Linux
  • Revision #Thu +
    • - Fixing Linux index compile errors
  • Revision #Thu +
    • - Miscelleanous fix for compiling on Linux
  • Revision #Thu +
    • - Implemented: not split indexes (all indexes in one file)
  • Revision #Tue +
    • - Adding the sprers.eu file used in sprers.eu
  • Revision #Tue +
    • - PROFILE_Close cannot be static
  • Revision #Tue +
    • - On Linux, closing an INI table removes it from the inihandl cache
  • Revision # [merge] Mon +
    • - Commit merged files
    • Revision #Mon +
      • Adding FILE privilege tests to table types CSV, DBF, FIX
    • Revision #Mon +
      • Adding "echo" in the end of sprers.eu, to read *.result files easier.
  • Revision # [merge] Mon +
    • - Commit merged files
    • Revision #Mon +
      • Adding privilege tests for ALTER.
    • Revision #Mon +
      • Fixing tests results according to the previous changeset by Olivier.
    • Revision #Mon +
      • Fixing tests results according to the previous changeset by Olivier.
  • Revision #Mon +
    • - Add a new table option SEPINDEX (not used yet) and remove an unused parameter to all catalog info functions.
  • Revision #Sat +
    • - Upated columns must be allocated before opening the table
  • Revision #Sat +
    • - Update test results and add a new one (UPD)
  • Revision #Sat +
    • - Wrong FLAG values transmitted to created table by the AS SELECT table: It is not enough to ignore the flags while populating the table. They have to be removed from the definition in pre_create. The issue is to pass the info from the selected table handler to the created table handler. It is done via the only common item between them: the GLOBAL structure.
  • Revision #Sat +
    • - Fix a typo error: in AllocateValue(2) Tiny values were given the TYPE_SHORT type.
  • Revision #Sat +
    • - Fix bug: When a table is created AS SELECT the offsets (FLAG value) of the source table columns must be ignored by the created table.
  • Revision #Fri +
    • - Fix check_if_incompatible_data to correctly get thd.
  • Revision #Fri +
    • - Fix bug to allow creating views.
  • Revision #Fri +
    • Fixing compilation failure in Windows: unknown symbol "any_db".
  • Revision #Fri +
    • Require FILE privilege to DROP a table with FILE_NAME.
  • Revision #Fri +
    • - Require FILE privilege for the file based TABLE_TYPEs when FILE_NAME is specified
  • Revision #Fri +
    • Skipping MySQL test when no MySQL support is compiled.
  • Revision # [merge] Thu +
    • - Commit merged items
    • Revision #Thu +
      • Removing executable flag.
    • Revision #Thu +
      • Skip xml test if no XML support is compiled.
  • Revision #Thu +
    • - Fix GetTypeID (to take care of supported features) and IsFileType functions (INI was missing)
  • Revision #Wed +
    • - A specified table type not supported is now flagged as an error instead of being replaced by the default type DOS.
  • Revision #Wed +
    • - Add a test on null columns
  • Revision #Wed +
    • - Changing CRLF line endings to LF only.
  • Revision #Wed +
    • - Fix a bug causing the index file not being made or erased on statements such as CREATE INDEX, DROP INDEX, ALTER TABLE ADD INDEX etc.
  • Revision #Tue +
    • - Make INI and XML tables not writing null column values
  • Revision #Tue +
    • - Use all columns in case of INSERT so default values are generated for columns not specified in the statemant.
  • Revision #Sun +
    • - Update tests that failed because of NULL columns
  • Revision #Sun +
    • - Add in create a check for nullable columns not supported by some table types.
  • Revision #Fri +
    • - A quick and dirty fix for erased or not erased files when executing an Alter table on an auto-generated CONNECT table.
  • Revision #Wed +
    • - Fix tests after last modification, in particular support of TINY
  • Revision #Wed +
    • - Fix bug on ALTER TABLE t1 MODIFY a VARCHAR(10) NOT NULL;
  • Revision #Wed +
    • - Make Tiny compatible with DBF tables.
  • Revision #Wed +
    • - Commit changes from ubuntu
  • Revision #Wed +
    • - Fix a GCC compile error (crosses initialization of pos) an change _O_CREAT to O_CREAT
  • Revision #Wed +
    • - Handle delete_table and rename_table for CONNECT tables whose files are auto generated and must be erased or renamed on Drop or Rename Table. TODO: Take care of eventual index files.
  • Revision #Tue +
    • - New handling of default file name: Not added as an option but handled when the table is used. An empty file is created in the database directory if not exists.
  • Revision #Mon +
    • - somethin about property
  • Revision #Mon +
    • - Resetting chmod to non executable
  • Revision #Mon +
    • - Fix a bug on DBF file name such as 'x/sprers.eu' leading to a crash on Insert.
  • Revision #Mon +
    • - Add tiny integer as a supported type by CONNECT
  • Revision #Mon +
    • - Enable type conversion on Insert and Update
  • Revision #Sun +
    • - Make indexes to be rebuilt on Update only when an indexed column was updated.
  • Revision #Sun +
    • - XML and INI tables now return NULL when a node does not exist in a row (XML) or if the key is missing in a section (INI)
  • Revision #Fri +
    • - Delete ALL or truncate must not be done using file mapping.
  • Revision #Fri +
    • - Commit changes on memory mapping that hopefully should now work on Linux.
  • Revision #Fri +
    • - Results updated to reflect new features
  • Revision #Fri +
    • - Until the problem is fixed, temporarily not use mapping on Linux for Update.
  • Revision # [merge] Fri +
    • - Working on the problem of writing to mapped files on Linux
    • Revision #Thu +
      • - Fix a bug causing the server to crash when writing on a memory mapped file on Linux.
  • Revision #Thu +
    • - Fix bug causing a crash when writing to a memory mapped file on Linux.
  • Revision #Thu +
    • - Updated to accept calculated default field length for date columns.
  • Revision #Thu +
    • - Set file name when unspecified as sprers.euype.
  • Revision #Thu +
    • - Calculate default date field length from the specified format.
  • Revision #Wed +
    • - Change line ending to LF only
  • Revision #Wed +
  • Revision #Wed +
    • - In sprers.eu t1 was back to T1! fixed
  • Revision #Wed +
  • Revision #Tue +
    • - Update the sprers.eu, which now works (at least on Windows)
  • Revision #Tue +
    • - Fix wrong format used in ScanRecord for MYSQL_TYPE_TIME - Change field option BUF_LENGTH to FIELD_LENGTH. It now used to specify the field length of DATE columns (MySQL doesn't allow to give a length)
  • Revision #Mon +
    • - Add domdoc.h in source list
  • Revision #Mon +
    • - Adding sprers.eu (with some statements temporarily skipped) - Updating sprers.eu
  • Revision #Mon +
    • - FIX a bug causing a crash when deleting a mapped table
  • Revision #Sun +
    • - FIX a BUG: error (wrong value set) in: update xempl set ddentree = adddate(ddentree, interval 16 year); The same value sdval was used to convert MySQL dates to CONNECT date value and CONNECT dates to MySQL date. This was wrong in update because the second time the wrong value was used converting to incoherent values. There are now 2 separate values used: sdvalin and sdvalout.
  • Revision #Sat +
    • - Some end of lines changed from CRLF to LF
  • Revision #Sat +
    • - Fix conversion bug for MS-DOM XML tables. The node content was written and read as if the table DATA_CHARSET was ANSI instead of UTF Warning are now provided when the read content of a node is truncated.
  • Revision # [merge] Sat +
    • - Commit merged file
    • Revision #Fri +
      • Adding forgotten semicolon. ODBC check did not work without it.
  • Revision #Sat +
    • - MySQL connection was not close if mysql_real_connect failed.
  • Revision #Sat +
    • - MySQL connection was not closed in case of error
  • Revision #Fri +
    • - Rewrite some VALBLK classes as templates - Correct typo initializing datm in DTVAL::MakeDate as {0,0,0,2,0,70,0,0,0} instead of {0,0,0,1,0,70,0,0,0}
  • Revision #Fri +
    • - Begin fixing memory leaks
  • Revision #Fri +
    • - Fix warning on missing initial values on Linux
  • Revision #Fri +
  • Revision #Thu +
    • - Second version of template value classes The first one did not compile with GCC on Linux
  • Revision #Wed +
    • - Modify the way value.h and valblk.h are included to try fixing the gcc error: invalid use of incomplete type
  • Revision #Wed +
    • - Change Subclass name from TYPE to T
  • Revision #Wed +
    • - Fix a miscalculation of column length in ODBCTables - Clean code in Value.h
  • Revision #Wed +
    • - Rewriting the VALUE classes as template classes
  • Revision #Wed +
    • - Result changed since nulls are supported
  • Revision #Wed +
    • Removing the comment about NULLs
  • Revision #Wed +
    • Updating sprers.eu, it now supports NULL columns.
  • Revision #Mon +
    • - Catalog table: Use XFLD as Flag value instead of column index. - Use the COLDEF flag to initialize column nullable value. - Fix a bug on inserting null values in MYSQL tables.
  • Revision #Sun +
    • - Add support to NULL values. This concern the MYSQL and ODBC table types. Not supported yet for indexes.
  • Revision #Fri +
    • - Add header files to source files
  • Revision # [merge] Fri +
    • - Commiting merges Simplify update in pre_create
    • Revision #Fri +
    • Revision #Fri +
      • Adding a test for BIGINT.
    • Revision #Fri +
      • - Adding tests for DBNAME=<unknown database name>. - Fixing a bug that DBNAME value was forgotten when followed by OPTION_LIST.
    • Revision #Fri +
      • Skip test XML if no XML support is compiled.
  • Revision #Fri +
    • - Suppress warning on non virtual descriptor
  • Revision #Fri +
    • - Fix connect string wrongly used when zero length
  • Revision #Fri +
    • - Add the support of URL connection string fo MYSQL tables Federated servers are not yet supported.
  • Revision #Fri +
    • Skip ODBC test if no ODBC is compiled.
  • Revision # [merge] Thu +
    • - Oops! db was no more initialized in pre-create. Fixed
    • Revision #Thu +
      • Suppress additional debug info printed in _DEBUG build
    • Revision #Thu +
      • Adding forgotten for port number
  • Revision #Thu +
    • - Change DB_NAME option name to DBNAME. Temporarily, "database" is still accepted in OPTION_LIST for compatibity but DB_NAME is no more recognized.
  • Revision #Thu +
    • - Fix bad conversion from "bigint" to TYPE_FLOAT.
  • Revision #Thu +
    • Adding tests for TABLE_TYPE=MySQL
  • Revision #Thu +
    • Adding a test for entities (special characters and characters outside of the file ENCODING)
  • Revision #Wed +
    • - Re-install blank trimming to have the xml test pass. Note that the problem if far more complex. To be revisited.
  • Revision #Wed +
    • - Fix the elimination of control characters from node content - Take care of XML special chars (<>& etc.) - Remove Encode, Decode
  • Revision #Wed +
    • - Update the MYSQL table handling to use only client API functions. It is no more necessary to be liked to sprers.eu nor sprers.eu
  • Revision #Wed +
    • Suppress a warning on unused variable.
  • Revision #Wed +
    • Fixing test failure due to not strict order of the files. Adding ORDER BY clause.
  • Revision #Wed +
    • Removing the MODULE_ONLY option, to be able to run with valgrind. (with MODULE_ONLY valgrind does not display file name and line number if a problem happens inside a dlopen-ed plugin)
  • Revision #Wed +
    • Fixing problems with running tests caused by the fact that "" conflicts with "".
  • Revision #Wed +
    • Fixing valgrind warning: uninitialized memory read.
  • Revision # [merge] Wed +
    • Revision #Tue +
      • Adding tests for DATA_CHARSET and TABLE_TYPE=CSV.
  • Revision #Wed +
    • - Fix crash on making an XML table with encoding=XXX - Set parameters so libxml2 does not anymore add extra characters when retrieving several subnodes of a node. - Make a CONNECT file header (was PlugDB)
  • Revision # [merge] Tue +
    • Commit merged changes.
    • Revision #Tue +
      • Removing unused code. modified: storage/connect/ha_sprers.eu
    • Revision #Tue +
    • Revision #Tue +
      • Adding tests to check that TABLE_TYPE=XML creates the XML file according to the ENCODING option.
  • Revision # [merge] Mon +
    • Don't why bazaar asked me to commit what was merged
    • Revision #Mon +
      • Adding DATA_CHARSET table option.
  • Revision #Mon +
    • Moved PushWarning out of HAVE_PSI_INTERFACE brackets
  • Revision #Mon +
    • Modification of the ODBConn class to change m_henv member from static to dynamic. Apparently this caused errors with a obscure message saying "Invalid handle vale" (probably m_henv).
  • Revision #Sun +
    • Modification of the DBX class to have more information in case of error while processing ODBC type tables.
  • Revision #Fri +
    • Put trace in ODBC source files
  • Revision #Fri +
    • Updating tests according to the latest Olivier's change removing extra space characters in TABLE_TYPE=XML with libxml2
  • Revision #Fri +
    • Suppress multiple blanks and control characters from the node text retrieved by libxml2.
  • Revision #Fri +
    • Adding a new test: creating a TABLE_TYPE=CSV table from a MyISAM table using
  • Revision #Fri +
    • Recording error messages according to the last change from Olivier.
  • Revision #Fri +
    • Fixed a bug in ha_connect::GetListOption causing a crash when passed a NULL oplist argument.
  • Revision #Fri +
    • Implementing pre_create option test and setting of default values. Currently, only TABLE_TYPE is tested, and if wrong or unspecified, is replaced by the default value DOS.
  • Revision #Thu +
    • Code reorganization in the initialization and clean-up code.
  • Revision #Thu +
    • Adding forgotten "", so it passed idependently from the source tree location.
  • Revision #Thu +
  • Revision # [merge] Thu +
    • Warning message changed in ha_connect::check_if_incompatible_data.
    • Revision #Wed +
      • Adding a test for unknown TABLE_TYPE
    • Revision #Wed +
  • Revision #Thu +
    • Warning message changed in ha_connect::check_if_incompatible_data.
  • Revision #Thu +
    • Update ha_connect::pre_create to restore the code translating of SQL types previously done in the removed function MyODBCCols.
  • Revision #Thu +
    • Fix a typo that caused CONNECT to crash on ODBC catalog tables in ODBCDEF::DefineAM line
  • Revision #Wed +
    • Fix bug on readonly table option not recognized in TABDEF::Define line
  • Revision #Tue +
    • Add a test on table types in ha_connect::create