• Filesystem
  • Motivation
  • Why the namespace GHC?
  • Platforms
  • Tests
  • Usage
    • Downloads
    • Using it as Single-File-Header
    • Using it as Forwarding-/Implementation-Header
    • Git Submodule and CMake
    • Versioning
  • Documentation
    • ghc::filesystem::ifstream, ghc::filesystem::ofstream, ghc::filesystem::fstream
    • ghc::filesystem::u8arguments
  • Differences
    • LWG Defects
    • Not Implemented on C++ before C++17
    • Differences in API
      • Differences of Specific Interfaces
    • Differences in Behavior
      • fs.path (ref)
  • Open Issues
    • Windows
      • Symbolic Links on Windows
      • Permissions
  • Release Notes

This is a header-only single-file std::filesystem compatible helper library, based on the C++17 and C++20 specs, but implemented for C++11, C++14, C++17 or C++20 (tightly following the C++17 standard with very few documented exceptions). It is currently tested on macOS 10.12/10.14/10.15/11.6, Windows 10, Ubuntu 18.04, Ubuntu 20.04, CentOS 7, CentOS 8, FreeBSD 12, Alpine ARM/ARM64 Linux and Solaris 10 but should work on other systems too, as long as you have at least a C++11 compatible compiler. It should work with Android NDK, Emscripten and I even had reports of it being used on iOS (within sandboxing constraints) and with v1.5.6 there is experimental support for QNX. The support of Android NDK, Emscripten, QNX, and since 1.5.14 GNU/Hurd and Haiku is not backed up by automated testing but PRs and bug reports are welcome for those too and they are reported to work. It is of course in its own namespace ghc::filesystem to not interfere with a regular std::filesystem should you use it in a mixed C++17 environment (which is possible).

Test coverage is well above 90%, and starting with v1.3.6 and in v1.5.0 more time was invested in benchmarking and optimizing parts of the library. I'll try to continue to optimize some parts and refactor others, striving to improve it as long as it doesn't introduce additional C++17/C++20 compatibility issues. Feedback is always welcome. Simply open an issue if you see something missing or wrong or not behaving as expected and I'll comment.

I'm often in need of filesystem functionality, mostly fs::path, but directory access too, and when beginning to use C++11, I used that language update to try to reduce my third-party dependencies. I could drop most of what I used, but still missed some stuff that I started implementing for the fun of it. Originally I based these helpers on my own coding- and naming conventions. When C++17 was finalized, I wanted to use that interface, but it took a while, to push myself to convert my classes.

The implementation is closely based on chapter 30.10 from the C++17 standard and a draft close to that version is Working Draft N4687. It is from after the standardization of C++17 but it contains the latest filesystem interface changes compared to the Working Draft N4659. Staring with v1.4.0, when compiled using C++20, it adapts to the changes according to path sorting order and std::u8string handling from Working Draft N4860.

I want to thank the people working on improving C++, I really liked how the language evolved with C++11 and the following standards. Keep on the good work!

If you ask yourself, what ghc is standing for, it is simply gulraks helper classes, yeah, I know, not very imaginative, but I wanted a short namespace and I use it in some of my private classes (so it has nothing to do with Haskell, sorry for the name clash).

ghc::filesystem is developed on macOS but CI tested on macOS, Windows, various Linux Distributions, FreeBSD and starting with v1.5.12 on Solaris. It should work on any of these with a C++11-capable compiler. Also there are some checks to hopefully better work on Android, but as I currently don't test with the Android NDK, I wouldn't call it a supported platform yet, same is valid for using it with Emscripten. It is now part of the detected platforms, I fixed the obvious issues and ran some tests with it, so it should be fine. All in all, I don't see it replacing std::filesystem where full C++17 or C++20 is available, it doesn't try to be a "better" std::filesystem, just an almost drop-in if you can't use it (with the exception of the UTF-8 preference).

ℹ️ Important: This implementation is following the "UTF-8 Everywhere" philosophy in that all std::string instances will be interpreted the same as std::u8string encoding wise and as being in UTF-8. The std::u16string will be seen as UTF-16. See Differences in API for more information.

Unit tests are currently run with:

• macOS 10.12: Xcode 9.2 (clang-900.0.39.2), GCC 9.2, Clang 9.0, macOS 10.13: Xcode 10.1, macOS 10.14: Xcode 11.2, macOS 10.15: Xcode 11.6, Xcode 12.4
• Windows: Visual Studio 2017, Visual Studio 2015, Visual Studio 2019, MinGW GCC 6.3 (Win32), GCC 7.2 (Win64), Cygwin GCC 10.2 (no CI yet)
• Linux (Ubuntu): GCC (5.5, 6.5, 7.4, 8.3, 9.2), Clang (5.0, 6.0, 7.1, 8.0, 9.0)
• Linux (Alpine ARM/ARM64): GCC 9.2.0 (The Drone build scripts stopped working, as they where a contribution, I couldn't figure out what went wrong, any help appreciated.)
• FreeBSD: Clang 8.0
• Solaris: GCC 5.5

The header comes with a set of unit-tests and uses CMake as a build tool and Catch2 as test framework. All tests are registered with in CMake, so the ctest commando can be used to run the tests.

All tests against this implementation should succeed, depending on your environment it might be that there are some warnings, e.g. if you have no rights to create Symlinks on Windows or at least the test thinks so, but these are just informative.

To build the tests from inside the project directory under macOS or Linux just:

mkdir build
cd build
cmake -DCMAKE_BUILD_TYPE=Debug ..
make
ctest

This generates the test binaries that run the tests and the last command executes them.

If the default compiler is a GCC 8 or newer, or Clang 7 or newer, it additionally tries to build a version of the test binary compiled against GCCs/Clangs std::filesystem implementation, named std_filesystem_test as an additional test of conformance. Ideally all tests should compile and succeed with all filesystem implementations, but in reality, there are some differences in behavior, sometimes due to room for interpretation in in the standard, and there might be issues in these implementations too.

The latest release version is v1.5.14 and source archives can be found here.

The latest pre-native-backend version is v1.4.0 and source archives can be found here.

The latest pre-C++20-support release version is v1.3.10 and source archives can be found here.

Currently only the latest minor release version receives bugfixes, so if possible, you should use the latest release.

As ghc::filesystem is at first a header-only library, it should be enough to copy the header or the include/ghc directory into your project folder or point your include path to this place and simply include the filesystem.hpp header (or ghc/filesystem.hpp if you use the subdirectory).

Everything is in the namespace ghc::filesystem, so one way to use it only as a fallback could be:

#if _MSVC_LANG >= 201703L || __cplusplus >= 201703L && defined(__has_include)
    // ^ Supports MSVC prior to 15.7 without setting /Zc:__cplusplus to fix __cplusplus
    // _MSVC_LANG works regardless. But without the switch, the compiler always reported 199711L: https://blogs.msdn.microsoft.com/vcblog/2018/04/09/msvc-now-correctly-reports-__cplusplus/
    #if __has_include(<filesystem>) // Two stage __has_include needed for MSVC 2015 and per https://gcc.gnu.org/onlinedocs/cpp/_005f_005fhas_005finclude.html
        #define GHC_USE_STD_FS

        // Old Apple OSs don't support std::filesystem, though the header is available at compile
        // time. In particular, std::filesystem is unavailable before macOS 10.15, iOS/tvOS 13.0,
        // and watchOS 6.0.
        #ifdef __APPLE__
            #include <Availability.h>
            // Note: This intentionally uses std::filesystem on any new Apple OS, like visionOS
            // released after std::filesystem, where std::filesystem is always available.
            // (All other __<platform>_VERSION_MIN_REQUIREDs will be undefined and thus 0.)
            #if __MAC_OS_X_VERSION_MIN_REQUIRED && __MAC_OS_X_VERSION_MIN_REQUIRED < 101500 \
             || __IPHONE_OS_VERSION_MIN_REQUIRED && __IPHONE_OS_VERSION_MIN_REQUIRED < 130000 \
             || __TV_OS_VERSION_MIN_REQUIRED && __TV_OS_VERSION_MIN_REQUIRED < 130000 \
             || __WATCH_OS_VERSION_MAX_ALLOWED && __WATCH_OS_VERSION_MAX_ALLOWED < 60000
                #undef GHC_USE_STD_FS
            #endif  
        #endif
    #endif
#endif

#ifdef GHC_USE_STD_FS
    #include <filesystem>
    namespace fs = std::filesystem;
#else
    #include "filesystem.hpp"
    namespace fs = ghc::filesystem;
#endif

If you want to also use the fstream wrapper with path support as fallback, you might use:

#if _MSVC_LANG >= 201703L || __cplusplus >= 201703L && defined(__has_include)
    // ^ Supports MSVC prior to 15.7 without setting /Zc:__cplusplus to fix __cplusplus
    // _MSVC_LANG works regardless. But without the switch, the compiler always reported 199711L: https://blogs.msdn.microsoft.com/vcblog/2018/04/09/msvc-now-correctly-reports-__cplusplus/
    #if __has_include(<filesystem>) // Two stage __has_include needed for MSVC 2015 and per https://gcc.gnu.org/onlinedocs/cpp/_005f_005fhas_005finclude.html
        #define GHC_USE_STD_FS

        // Old Apple OSs don't support std::filesystem, though the header is available at compile
        // time. In particular, std::filesystem is unavailable before macOS 10.15, iOS/tvOS 13.0,
        // and watchOS 6.0.
        #ifdef __APPLE__
            #include <Availability.h>
            // Note: This intentionally uses std::filesystem on any new Apple OS, like visionOS
            // released after std::filesystem, where std::filesystem is always available.
            // (All other __<platform>_VERSION_MIN_REQUIREDs will be undefined and thus 0.)
            #if __MAC_OS_X_VERSION_MIN_REQUIRED && __MAC_OS_X_VERSION_MIN_REQUIRED < 101500 \
             || __IPHONE_OS_VERSION_MIN_REQUIRED && __IPHONE_OS_VERSION_MIN_REQUIRED < 130000 \
             || __TV_OS_VERSION_MIN_REQUIRED && __TV_OS_VERSION_MIN_REQUIRED < 130000 \
             || __WATCH_OS_VERSION_MAX_ALLOWED && __WATCH_OS_VERSION_MAX_ALLOWED < 60000
                #undef GHC_USE_STD_FS
            #endif  
        #endif
    #endif
#endif

#ifdef GHC_USE_STD_FS
    #include <filesystem>
    namespace fs {
        using namespace std::filesystem;
        using ifstream = std::ifstream;
        using ofstream = std::ofstream;
        using fstream = std::fstream;
    }
#else
    #include "filesystem.hpp"
    namespace fs {
        using namespace ghc::filesystem;
        using ifstream = ghc::filesystem::ifstream;
        using ofstream = ghc::filesystem::ofstream;
        using fstream = ghc::filesystem::fstream;
    }
#endif

Now you have e.g. fs::ofstream out(somePath); and it is either the wrapper or the C++17 std::ofstream.

ℹ️ Be aware, as a header-only library, it is not hiding the fact, that it uses system includes, so they "pollute" your global namespace. Use the forwarding-/implementation-header based approach (see below) to avoid this. For Windows it needs Windows.h and it might be a good idea to define WIN32_LEAN_AND_MEAN or NOMINMAX prior to including filesystem.hpp or fs_std.hpp headers to reduce pollution of your global namespace and compile time. They are not defined by ghc::filesystem to allow combination with contexts where the full Windows.his needed, e.g. for UI elements.

ℹ️ Hint: There is an additional header named ghc/fs_std.hpp that implements this dynamic selection of a filesystem implementation, that you can include instead of ghc/filesystem.hpp when you want std::filesystem where available and ghc::filesystem where not.

Alternatively, starting from v1.1.0 ghc::filesystem can also be used by including one of two additional wrapper headers. These allow to include a forwarded version in most places (ghc/fs_fwd.hpp) while hiding the implementation details in a single cpp file that includes ghc/fs_impl.hpp to implement the needed code. Using ghc::filesystem this way makes sure system includes are only visible from inside the cpp file, all other places are clean.

Be aware, that it is currently not supported to hide the implementation into a Windows-DLL, as a DLL interface with C++ standard templates in interfaces is a different beast. If someone is willing to give it a try, I might integrate a PR but currently working on that myself is not a priority.

If you use the forwarding/implementation approach, you can still use the dynamic switching like this:

#if _MSVC_LANG >= 201703L || __cplusplus >= 201703L && defined(__has_include)
    // ^ Supports MSVC prior to 15.7 without setting /Zc:__cplusplus to fix __cplusplus
    // _MSVC_LANG works regardless. But without the switch, the compiler always reported 199711L: https://blogs.msdn.microsoft.com/vcblog/2018/04/09/msvc-now-correctly-reports-__cplusplus/
    #if __has_include(<filesystem>) // Two stage __has_include needed for MSVC 2015 and per https://gcc.gnu.org/onlinedocs/cpp/_005f_005fhas_005finclude.html
        #define GHC_USE_STD_FS

        // Old Apple OSs don't support std::filesystem, though the header is available at compile
        // time. In particular, std::filesystem is unavailable before macOS 10.15, iOS/tvOS 13.0,
        // and watchOS 6.0.
        #ifdef __APPLE__
            #include <Availability.h>
            // Note: This intentionally uses std::filesystem on any new Apple OS, like visionOS
            // released after std::filesystem, where std::filesystem is always available.
            // (All other __<platform>_VERSION_MIN_REQUIREDs will be undefined and thus 0.)
            #if __MAC_OS_X_VERSION_MIN_REQUIRED && __MAC_OS_X_VERSION_MIN_REQUIRED < 101500 \
             || __IPHONE_OS_VERSION_MIN_REQUIRED && __IPHONE_OS_VERSION_MIN_REQUIRED < 130000 \
             || __TV_OS_VERSION_MIN_REQUIRED && __TV_OS_VERSION_MIN_REQUIRED < 130000 \
             || __WATCH_OS_VERSION_MAX_ALLOWED && __WATCH_OS_VERSION_MAX_ALLOWED < 60000
                #undef GHC_USE_STD_FS
            #endif  
        #endif
    #endif
#endif

#ifdef GHC_USE_STD_FS
    #include <filesystem>
    namespace fs {
        using namespace std::filesystem;
        using ifstream = std::ifstream;
        using ofstream = std::ofstream;
        using fstream = std::fstream;
    }
#else
    #include "fs_fwd.hpp"
    namespace fs {
        using namespace ghc::filesystem;
        using ifstream = ghc::filesystem::ifstream;
        using ofstream = ghc::filesystem::ofstream;
        using fstream = ghc::filesystem::fstream;
    }
#endif

and in the implementation hiding cpp, you might use (before any include that includes ghc/fs_fwd.hpp to take precedence:

#if _MSVC_LANG >= 201703L || __cplusplus >= 201703L && defined(__has_include)
    // ^ Supports MSVC prior to 15.7 without setting /Zc:__cplusplus to fix __cplusplus
    // _MSVC_LANG works regardless. But without the switch, the compiler always reported 199711L: https://blogs.msdn.microsoft.com/vcblog/2018/04/09/msvc-now-correctly-reports-__cplusplus/
    #if __has_include(<filesystem>) // Two stage __has_include needed for MSVC 2015 and per https://gcc.gnu.org/onlinedocs/cpp/_005f_005fhas_005finclude.html
        #define GHC_USE_STD_FS

        // Old Apple OSs don't support std::filesystem, though the header is available at compile
        // time. In particular, std::filesystem is unavailable before macOS 10.15, iOS/tvOS 13.0,
        // and watchOS 6.0.
        #ifdef __APPLE__
            #include <Availability.h>
            // Note: This intentionally uses std::filesystem on any new Apple OS, like visionOS
            // released after std::filesystem, where std::filesystem is always available.
            // (All other __<platform>_VERSION_MIN_REQUIREDs will be undefined and thus 0.)
            #if __MAC_OS_X_VERSION_MIN_REQUIRED && __MAC_OS_X_VERSION_MIN_REQUIRED < 101500 \
             || __IPHONE_OS_VERSION_MIN_REQUIRED && __IPHONE_OS_VERSION_MIN_REQUIRED < 130000 \
             || __TV_OS_VERSION_MIN_REQUIRED && __TV_OS_VERSION_MIN_REQUIRED < 130000 \
             || __WATCH_OS_VERSION_MAX_ALLOWED && __WATCH_OS_VERSION_MAX_ALLOWED < 60000
                #undef GHC_USE_STD_FS
            #endif  
        #endif
    #endif
#endif

#ifndef GHC_USE_STD_FS
    #include "fs_impl.hpp"
#endif

ℹ️ Hint: There are additional helper headers, named ghc/fs_std_fwd.hpp and ghc/fs_std_impl.hpp that use this technique, so you can simply include them if you want to dynamically select the filesystem implementation.

Starting from v1.1.0, it is possible to add ghc::filesystem as a git submodule, add the directory to your CMakeLists.txt with add_subdirectory() and then simply use target_link_libraries(your-target ghc_filesystem) to ensure correct include path that allow #include <ghc/filesystem.hpp> to work.

The CMakeLists.txt offers a few options to customize its behavior:

• GHC_FILESYSTEM_BUILD_TESTING - Compile tests, default is OFF when used as a submodule, else ON.
• GHC_FILESYSTEM_BUILD_EXAMPLES - Compile the examples, default is OFF when used as a submodule, else ON.
• GHC_FILESYSTEM_WITH_INSTALL - Add install target to build, default is OFF when used as a submodule, else ON.
• GHC_FILESYSTEM_BUILD_STD_TESTING - Compile std_filesystem_test, the variant of the test suite running against std::filesystem, defaulting to GHC_FILESYSTEM_BUILD_TESTING. This is only done if the compiler is detected as being able to do it.
• GHC_FILESYSTEM_TEST_COMPILE_FEATURES can be set to a list of features to override CMAKE_CXX_COMPILE_FEATURES when the detection of C++17 or C++20 for additional tests is not working (e.g. cxx_std_20 to enforce building a filesystem_test_cpp20 with C++20).

Please use hedronvision/bazel-cc-filesystem-backport, which will automatically set everything up for you.

There is a version macro GHC_FILESYSTEM_VERSION defined in case future changes might make it needed to react on the version, but I don't plan to break anything. It's the version as decimal number (major * 10000 + minor * 100 + patch).

ℹ️ Note: Only even patch versions will be used for releases and odd patch version will only be used for in between commits while working on the next version.

There is almost no documentation in this release, as any std::filesystem documentation would work, besides the few differences explained in the next section. So you might head over to https://en.cppreference.com/w/cpp/filesystem for a description of the components of this library.

When compiling with C++11, C++14 or C++17, the API is following the C++17 standard, where possible, with the exception that std::string_view parameters are only supported on C++17. When Compiling with C++20, ghc::filesysytem defaults to the C++20 API, with the char8_t and std::u8string interfaces and the deprecated fs::u8path factory method.

ℹ️ Note: If the C++17 API should be enforced even in C++20 mode, use the define GHC_FILESYSTEM_ENFORCE_CPP17_API. Even then it is possible to create fws::path from std::u8string but fs::path::u8string() and fs::path::generic_u8string() return normal UTF-8 encoded std::string instances, so code written for C++17 could still work with ghc::filesystem when compiled with C++20.

The only additions to the standard are documented here:

These are simple wrappers around std::ifstream, std::ofstream and std::fstream. They simply add an open() method and a constructor with an ghc::filesystem::path argument as the fstream variants in C++17 have them.

This is a helper class that currently checks for UTF-8 encoding on non-Windows platforms but on Windows it fetches the command line arguments as Unicode strings from the OS with

::CommandLineToArgvW(::GetCommandLineW(), &argc)

and then converts them to UTF-8, and replaces argc and argv. It is a guard-like class that reverts its changes when going out of scope.

So basic usage is:

namespace fs = ghc::filesystem;

int main(int argc, char* argv[])
{
    fs::u8arguments u8guard(argc, argv);
    if(!u8guard.valid()) {
        std::cerr << "Bad encoding, needs UTF-8." << std::endl;
        exit(EXIT_FAILURE);
    }

    // now use argc/argv as usual, they have utf-8 encoding on windows
    // ...

    return 0;
}

That way argv is UTF-8 encoded as long as the scope from main is valid.

Note: On macOS, while debugging under Xcode the code currently will return false as Xcode starts the application with US-ASCII as encoding, no matter what encoding is actually used and even setting LC_ALL in the product scheme doesn't change anything. I still need to investigate this.

As this implementation is based on existing code from my private helper classes, it derived some constraints of it. Starting from v1.5.0 most of the differences between this and the standard C++17/C++20 API where removed.

This implementation has switchable behavior for the LWG defects #2682, #2935, #2936 and #2937. The currently selected behavior (starting from v1.4.0) is following #2682, #2936, #2937 but not following #2935, as I feel it is a bug to report no error on a create_directory() or create_directories() where a regular file of the same name prohibits the creation of a directory and forces the user of those functions to double-check via fs::is_directory if it really worked. The more intuitive approach to directory creation of treating a file with that name as an error is also advocated by the newer paper WG21 P1164R0, the revision P1161R1 was agreed upon on Kona 2019 meeting see merge and GCC by now switched to following its proposal (GCC #86910).

// methods in ghc::filesystem::path:
path& operator+=(basic_string_view<value_type> x);
int compare(basic_string_view<value_type> s) const;

These are not implemented under C++11 and C++14, as there is no std::basic_string_view available and I did want to keep this implementation self-contained and not write a full C++17-upgrade for C++11/14. Starting with v1.1.0 these are supported when compiling ghc::filesystem under C++17 of C++20.

Starting with v1.5.2 ghc::filesystem will try to allow the use of std::experimental::basic_string_view where it detects is availability. Additionally if you have a basic_string_view compatible c++11 implementation it can be used instead of std::basic_string_view by defining GHC_HAS_CUSTOM_STRING_VIEW and importing the implementation into the ghc::filesystem namespace with:

namespace ghc {
    namespace filesystem {
        using my::basic_string_view;
    }
}

before including the filesystem header.

To not depend on any external third party libraries and still stay portable and compact, this implementation is following the "UTF-8 Everywhere" philosophy in that all std::string instances will be interpreted the same as std::u8string encoding wise and as being in UTF-8. The std::u16string will be seen as UTF-16 and std::u32string will be seen as Unicode codepoints. Depending on the size of std::wstring characters, it will handle std::wstring as being UTF-16 (e.g. Windows) or char32_t Unicode codepoints (currently all other platforms).

Starting with v1.5.0 ghc::filesystem is following the C++17 standard in using wchar_t and std::wstring on Windows as the types internally used for path representation. It is still possible to get the old behavior by defining GHC_WIN_DISABLE_WSTRING_STORAGE_TYPE and get filesystem::path::string_type as std::string and filesystem::path::value_type as wchar_t.

If you need to call some Windows API, with v1.5.0 and above, simply use the W-variant of the Windows-API call (e.g. GetFileAttributesW(p.c_str())).

ℹ️ Note: When using the old behavior by defining GHC_WIN_DISABLE_WSTRING_STORAGE_TYPE, use the path::wstring() member (e.g. GetFileAttributesW(p.wstring().c_str())). This gives you the Unicode variant independent of the UNICODE macro and makes sharing code between Windows, Linux and macOS easier and works with std::filesystem and ghc::filesystem.

std::string path::u8string() const;
std::string path::generic_u8string() const;
vs.
std::u8string path::u8string() const;
std::u8string path::generic_u8string() const;

The return type of these two methods is depending on the used C++ standard and if GHC_FILESYSTEM_ENFORCE_CPP17_API is defined. On C++11, C++14 and C++17 or when GHC_FILESYSTEM_ENFORCE_CPP17_API is defined, the return type is std::string, and on C++20 without the define it is std::u8string.

I created a wiki entry about quite a lot of behavioral differences between different std::filesystem implementations that could result in a mention here, but this readme only tries to address the design choice differences between ghc::filesystem and those. I try to update the wiki page from time to time.

Any additional observations are welcome!

Since v1.5.0 the complete inner mechanics of this implementations fs::path where changed to the native format as the internal representation. Creating any mixed slash fs::path object under Windows (e.g. with "C:\foo/bar") will lead clean path with "C:\foo\bar" via native() and "C:/foo/bar" via generic_string() API. On all platforms redundant additional separators are removed, even if this is not enforced by the standard and other implementations mostly not do this.

Additionally this implementation follows the standards suggestion to handle posix paths of the form "//host/path" and USC path on windows also as having a root-name (e.g. "//host"). The GCC implementation didn't choose to do that while testing on Ubuntu 18.04 and macOS with GCC 8.1.0 or Clang 7.0.0. This difference will show as warnings under std::filesystem. This leads to a change in the algorithm described in the standard for operator/=(path& p) where any path p with p.is_absolute() will degrade to an assignment, while this implementation has the exception where *this == *this.root_name() and p == preferred_separator a normal append will be done, to allow:

fs::path p1 = "//host/foo/bar/file.txt";
fs::path p2;
for (auto p : p1) p2 /= p;
ASSERT(p1 == p2);

For all non-host-leading paths the behavior will match the one described by the standard.

As symbolic links on Windows, while being supported more or less since Windows Vista (with some strict security constraints) and fully since some earlier build of Windows 10, when "Developer Mode" is activated, are at time of writing (2018) rarely used, still they are supported wiit th this implementation.

The Windows ACL permission feature translates badly to the POSIX permission bit mask used in the interface of C++17 filesystem. The permissions returned in the file_status are therefore currently synthesized for the user-level and copied to the group- and other-level. There is still some potential for more interaction with the Windows permission system, but currently setting or reading permissions with this implementation will most certainly not lead to the expected behavior.

• Fix for #166, extension() did return non empty result for the directory name ".."

• Pull request #163, build support for Haiku (also fixes #159)
• Pull request #162, fix for directory iterator treating all files subsequent to a symlink as symlink on Windows
• Pull request #161, the CMake alias ghcFilesystem::ghc_filesystem is now set unconditionally
• Fix for #160, the cmake config now only sets install targets by default if the project is no subproject, as documented
• Fix for #157, suppress C4191 warning on MSVC for GetProcAddress casts
• Fix for #156, on POSIX stem(), filename() and extension() of fs::path would return wrong result if a colon was in the filename
• Pull request #154, build support for GNU/Hurd
• Pull request #153, fixed fs::last_write_time(path, time, ec) setter on iOS, tvOS and watchOS
• Fix for #151, fs::directory_entry::refresh() now, consistently with status() will not throw on symlinks to non-existing targets, but make the entry have file_type::not_found as the type
• Pull request #149, add version to CMake project and export it
• Fix for #146, handle EINTR on POSIX directory iteration and file copy to avoid errors on network filesystems
• Pull request #145, fix for Y2038 bug in timeToFILETIME on Windows
• Pull request #144, fs::copy_file() now also copies the permissions
• Pull request #143, fix for fs::copy_file() ignoring the skip_existing option.

• Fix for #142, removed need for GHC_NO_DIRENT_D_TYPE on systems that don't support dirent::d_type and fixed build configuration and tests to support Solaris as new platform.
• Pull request #138, if the platform uses the POSIX backend and has no PATH_MAX, one is defined.
• Pull request #137, update of Catch2 to version v2.13.7
• Added macOS 11 to the automatically tested platforms.

• Pull request #136, the Windows implementation used some unnecessary expensive shared pointer for resource management and these where replaced by a dedicated code.
• Fix for #132, pull request #135, fs::remove_all now just deletes symbolic links instead of following them.
• Pull request #133, fix for fs::space where a numerical overflow could happen in a multiplication.
• Replaced travis-ci.org with GitHub Workflow for the configurations: Ubuntu 20.04: GCC 9.3, Ubuntu 18.04: GCC 7.5, GCC 8.4, macOS 10.15: Xcode 12.4, Windows 10: Visual Studio 2019

• Fix for #125, where fs::create_directories on Windows no longer breaks on long filenames.

• Fix for #124, ghc::filesystem treated mounted folder/volumes erroneously as symlinks, leading fs::canonical to fail on paths containing those.
• Fix for #122, incrementing the recursive_directory_iterator will not try to enter dead symlinks.
• Fix for #121, on Windows backend the fs::remove failed when the path pointed to a read-only entry, see also (microsoft/STL#1511) for the corresponding issue in std::fs on windows.
• Fix for #119, added missing support for char16_t and char32_t and on C++20 char8_t literals.
• Pull request #118, when running tests as root, disable tests that would not work.
• Pull request #117, added checks to tests to detect the clang/libstdc++ combination.
• Fix for #116, internal macro GHC_NO_DIRENT_D_TYPE allows os detection to support systems without the dirent.d_type member, experimental first QNX compile support as initial use case, fixed issue with filesystems returning DT_UNKNOWN (e.g. reiserfs).
• Pull request #115, added string_view support when clang with libstdc++ is detected.
• Fix for #114, for macOS the pre-Catalina deployment target detection worked only if <Availability.h> was included before <ghc/fs_std.hpp> or <ghc/fs_std_fwd.hpp>/<ghc/fs_std_impl.hpp>.
• Fix for #113, the use of standard chapter numbers was misleading since C++17 and C++20 std::filesystem features are supported, and was replaced by the tag-like chapter names that stay (mostly) consistent over the versions.

• Pull request #112, lots of cleanup work on the readme, thanks!
• Enhancement for #111, further optimization of directory iteration, performance for recursive_directory_iterator over large trees now somewhere between libc++ and libstdc++.
• Enhancement for #110, ghc::filesystem now has preliminary support for Cygwin. Changes where made to allow the tests to compile and run successfully (tested with GCC 10.2.0), feedback and additional PRs welcome as it is currently not part of the CI configuration.
• Pull request #109, various spelling errors in error messages and comments fixed.
• Pull request #108, old style casts removed.
• Fix for #107, the error handling for status calls was suppressing errors on symlink targets.
• Pull request #106, fixed detection of AppleClang for compile options.
• Pull request #105, added option GHC_FILESYSTEM_BUILD_STD_TESTING to override additional build of std::filesystem versions of the tests for comparison and the possibility to use GHC_FILESYSTEM_TEST_COMPILE_FEATURES to prefill the used compile features defaulting to CMAKE_CXX_COMPILE_FEATURES when not given.

• Enhancement #104, on POSIX backend: optimized reuse of status information and reduced directory_entry creation leads to about 20%-25% in tests with recursive_directory_iterator over a larger directory tree.
• Pull request #103, wchar_t was not in the list of supported char types on non-Windows backends.
• Pull request #102, improved string_view support makes use of <string_view> or <experimental/string_view> when available, and allows use of custom basic_string_view implementation when defining GHC_HAS_CUSTOM_STRING_VIEW and importing the string view into the ghc::filesystem namespace before including filesystem header.
• Pull request #101, fix for #100, append and concat type of operations on path called redundant conversions.
• Pull request #98, on older linux variants (GCC 7/8), the comparison std::filesystem tests now link with -lrt to avoid issues.
• Fix for #97, on BTRFS the test case for fs::hard_link_count failed due to the filesystems behavior, the test case was adapted to take that into account.
• Pull request #96, the export attribute defines GHC_FS_API and GHC_FS_API_CLASS are now honored when when set from outside to allow override of behavior.
• Fix for #95, the syntax for disabling the deprecated warning in tests in MSVC was wrong.
• Pull request #93, now the CMake configuration file is configured and part of the make install files.

• Fix for #91, the way the CMake build options GHC_FILESYSTEM_BUILD_TESTING, GHC_FILESYSTEM_BUILD_EXAMPLES and GHC_FILESYSTEM_WITH_INSTALL where implemented, prohibited setting them from a parent project when using this via add_subdirectory, this fix allows to set them again.
• Major refactoring for #90, the way, the Windows version of fs::path was originally created from the POSIX based implementation was, by adaption of the incoming and outgoing strings. This resulted in a mutable cache inside fs::pathon Windows, that was inherently not thread-safe, even for const methods. To not add additional patches to a suboptimal solution, this time I reworked the path code to now store native path-representation. This changed a lot of code, but when combined with wchar_t as value_type helped to avoid lots of conversion for calls to Win-API.
  As interfaces where changed, it had to be released in a new minor version. The set of refactorings resulted in the following changes:
  • fs::path::native() and fs::path::c_str() can now be noexcept as the standard mandates
  • On Windows wchar_t is now the default for fs::path::value_type and std::wstring is the default for fs::path::string_type.
  • This allows the implementation to call Win-API without allocating conversions
  • Thread-safety on const methods of fs::path is no longer an issue
  • Some code could be simplified during this refactoring
  • Automatic prefixing of long path on Windows can now be disabled with defining GHC_WIN_DISABLE_AUTO_PREFIXES, for all other types of prefixes or namespaces the behavior follows that of MSVC std::filesystem::path
  • In case the old char/std::string based approach for Windows is still needed, it can be activated with GHC_WIN_DISABLE_WSTRING_STORAGE_TYPE
• Enhancement for #89, fs::file_status now supports operator== introduced in std::filesystem with C++20.
• Refactoring for #88, fs::path::parent_path() had a performance issue, as it was still using a loop based approach to recreate the parent from elements. This created lots of temporaries and was too slow especially on long paths.

• Enhancements for #71, when compiled with C++20:
  • char8_t and std::u8string are supported where Source is the parameter type
  • fs::path::u8string() and fs::path::generic_u8string() now return a std::u8string
  • The spaceship operator <=> is now supported for fs::path
  • With the define GHC_FILESYSTEM_ENFORCE_CPP17_API ghc::filesystem will fall back to the old fs::path::u8string() and fs::path::generic_u8string() API if preferred
• Bugfix for fs::proximate(p, ec) where the internal call to fs::current_path() was not using the error_code variant, throwing possible exceptions instead of setting ec.
• Enhancement LWG_2936_BEHAVIOUR is now on by default.
• Some cleanup work to reduce preprocessor directives for better readability and remove unneeded template specializations.

• Fix for #81, fixed issues with handling Source parameters that are string views.
• Fix for #79, the bit operations for filesystem bitmasks that should be are now constexpr.

• Refactoring for #78, the dynamic switching helper includes are now using __MAC_OS_X_VERSION_MIN_REQUIRED to ensure that std::filesystem is only selected on macOS if the deployment target is at least Catalina.
• Bugfix for #77, the directory_iterator and the recursive_directory_iterator had an issue with the skip_permission_denied option, that leads to the inability to skip SIP protected folders on macOS.
• Enhancement for #76, _MSVC_LANG is now used when available, additionally to __cplusplus, in the helping headers to allow them to work even when /Zc:__cplusplus is not used.
• Bugfix for #75, NTFS reparse points to mapped volumes where handled incorrect, leading to false on fs::exists or not-found-errors on fs::status. Namespaced paths are not filtered anymore.

• Pull request #74, on Windows symlink evaluation used the wrong reparse struct information and was not handling the case of relative paths well, thanks for the contribution.
• Refactoring for #73, enhanced performance in path handling. the changes lead to much fewer path/string creations or copies, speeding up large directory iteration or operations on many path instances.
• Bugfix for #72, the TestAllocator in filesystem_test.cpp was completed to fulfill the requirements to build on CentOS 7 with devtoolset-9. CentOS 7 and CentOS 8 are now part of the CI builds.
• Bugfix for #70, root names are now case insensitive on Windows. This fix also adds the new behavior switch LWG_2936_BEHAVIOUR that allows to enable post C++17 fs::path::compare behavior, where the comparison is as if it was an element wise path comparison as described in LWG 2936 and C++20 [fs.path.compare]. It is default off in v1.3.6 and will be default starting from v1.4.0 as it changes ordering.

• Pull request #69, use wchar_t versions of std::fstream from ghc::filesystem::fstream wrappers on Windows if using GCC with libc++.
• Bugfix for #68, better handling of permission issues for directory iterators when using fs::directory_options::skip_permission_denied and initial support for compilation with emscripten.
• Refactoring for #66, unneeded shared_ptr guards where removed and the file handles closed where needed to avoid unnecessary allocations.
• Bugfix for #63, fixed issues on Windows with clang++ and C++17.
• Pull request #62, various fixes for better Android support, thanks for the PR
• Pull request #61, ghc::filesystem now supports use in projects with disabled exceptions. API signatures using exceptions for error handling are not available in this mode, thanks for the PR (this resolves #60 and #43)

• Bugfix for #58, on MinGW the compilation could fail with an error about an undefined ERROR_FILE_TOO_LARGE constant.
• Bugfix for #56, fs::lexically_relative didn't ignore trailing slash on the base parameter, thanks for PR #57.
• Bugfix for #55, fs::create_directories returned true when nothing needed to be created, because the directory already existed.
• Bugfix for #54, error_code was not reset, if cached result was returned.
• Pull request #53, fix for wrong handling of leading whitespace when reading fs::path from a stream.
• Pull request #52, an ARM Linux target is now part of the CI infrastructure with the service of Drone CI.
• Pull request #51, FreeBSD is now part of the CI infrastructure with the service of Cirrus CI.
• Pull request #50, adaptive cast to timespec fields to avoid warnings.

• Important: ghc::filesystem is re-licensed from BSD-3-Clause to MIT license. (see #47)
• Pull request #46, suppresses unused parameter warning on Android.
• Bugfix for #44, fixes for warnings from newer Xcode versions.

• The Visual Studio 2019 compiler, GCC 9.2 and Clang 9.0 where added to the CI configuration.
• Bugfix for #41, fs::rename on Windows didn't replace an existing regular file as required by the standard, but gave an error. New tests and a fix as provided in the issue was implemented.
• Bugfix for #39, for the forwarding use via fs_fwd.hpp or fs_std_fwd.hpp there was a use of DWORD in the forwarding part leading to an error if Windows.h was not included before the header. The tests were changed to give an error in that case too and the useage of DWORD was removed.
• Bugfix for #38, casting the return value of GetProcAddress gave a warning with -Wcast-function-type on MSYS2 and MinGW GCC 9 builds.

• Pull request #30, the CMakeLists.txt will automatically exclude building examples and tests when used as submodule, the configuration options now use a prefixed name to reduce risk of conflicts.
• Pull request #24, install target now creates a ghcFilesystemConfig.cmake in ${CMAKE_INSTALL_LIBDIR}/cmake/ghcFilesystem for find_package that exports a target as ghcFilesystem::ghc_filesystem.
• Pull request #31, fixes error: redundant redeclaration of 'constexpr' static data member deprecation warning in C++17 mode.
• Pull request #32, fixes old-style-cast warnings.
• Pull request #34, fixes TOCTOU situation on fs::create_directories, thanks for the PR!
• Feature #35, new CMake option to add an install target GHC_FILESYSTEM_WITH_INSTALL that is defaulted to OFF if ghc::filesystem is used via add_subdirectory.
• Bugfix for #33, fixes an issue with fs::path::lexically_normal() that leaves a trailing separator in case of a resulting path ending with .. as last element.
• Bugfix for #36, warnings on Xcode 11.2 due to unhelpful references in path element iteration.

• Pull request #23, tests and examples can now be disabled in CMake via setting BUILD_TESTING and BUILD_EXAMPLES to NO, OFF or FALSE.
• Pull request #25, missing specialization for construction from std::string_view when available was added.
• Additional test case when std::string_view is available.
• Bugfix for #27, the fs::path::preferred_separator declaration was not compiling on pre C++17 compilers and no test accessed it, to show the problem. Fixed it to an construction C++11 compiler should accept and added a test that is successful on all combinations tested.
• Bugfix for #29, stricter warning settings where chosen and resulting warnings where fixed.

• Enabled stronger warning switches and resulting fixed issues on GCC and MinGW
• Bugfix for #22, the fs::copy_options where not forwarded from fs::copy to fs::copy_file in one of the cases.

• Fix for (#21), when compiling on Alpine Linux with musl instead of glibc, the wrong strerror_r signature was expected. The complex preprocessor define mix was dropped in favor of the usual dispatch by overloading a unifying wrapper.

• Added MinGW 32/64 and Visual Studio 2015 builds to the CI configuration.
• Fixed additional compilation issues on MinGW.
• Pull request (#13), set minimum required CMake version to 3.7.2 (as in Debian 8).
• Pull request (#14), added support for a make install target.
• Bugfix for (#15), the forward/impl way of using ghc::filesystem missed a <vector> include in the windows case.
• Bugfix for (#16), VS2019 didn't like the old size dispatching in the utf8 decoder, so it was changed to a sfinae based approach.
• New feature (#17), optional support for standard conforming wchar_t/std::wstring interface when compiling on Windows with defined GHC_WIN_WSTRING_STRING_TYPE, this is default when using the ghc/fs_std*.hpp header, to enhance compatibility.
• New feature (#18), optional filesystem exceptions/errors on Unicode errors with defined GHC_RAISE_UNICODE_ERRORS (instead of replacing invalid code points or UTF-8 encoding errors with the replacement character U+FFFD).
• Pull request (#20), fix for file handle leak in fs::copy_file.
• Coverage now checked in CI (~95% line coverage).

• Additional Bugfix for (#12), error in old unified readdir/readdir_r code of fs::directory_iterator; as readdir_r is now deprecated, I decided to drop it and the resulting code is much easier, shorter and due to more refactoring faster
• Fix for crashing unit tests against MSVC C++17 std::filesystem
• Travis-CI now additionally test with Xcode 10.2 on macOS
• Some minor refactorings

• Bugfix for (#11), fs::path::lexically_normal() had some issues with ".."-sequences.
• Bugfix for (#12), fs::recursive_directory_iterator could run into endless loops, the methods depth() and pop() had issues and the copy behavior and input_iterator_tag conformance was broken, added tests
• Restructured some CMake code into a macro to ease the support for C++17 std::filesystem builds of tests and examples for interoperability checks.
• Some fixes on Windows tests to ease interoperability test runs.
• Reduced noise on fs::weakly_canonical() tests against std::fs
• Added simple du example showing the recursive_directory_iterator used to add the sizes of files in a directory tree.
• Added error checking in fs::file_time_type test helpers
• fs::copy() now conforms LWG #2682, disallowing the use of `copy_option::create_symlinks' to be used on directories

• Restructuring of the project directory. The header files are now using hpp as extension to be marked as c++ and they where moved to include/ghc/ to be able to include by <ghc/filesystem.hpp> as the former include name might have been to generic and conflict with other files.
• Better CMake support: ghc::filesystem now can be used as a submodul and added with add_subdirectory and will export itself as ghc_filesystem target. To use it, only target_link_libraries(your-target ghc_filesystem) is needed and the include directories will be set so #include <ghc/filesystem.hpp> will be a valid directive. Still you can simply only add the header file to you project and include it from there.
• Enhancement (#10), support for separation of implementation and forwarded api: Two additional simple includes are added, that can be used to forward ghc::filesystem declarations (fs_fwd.hpp) and to wrap the implementation into a single cpp (fs_impl.hpp)
• The std::basic_string_view variants of the fs::path api are now supported when compiling with C++17.
• Added CI integration for Travis-CI and Appveyor.
• Fixed MinGW compilation issues.
• Added long filename support for Windows.

• Bugfix for (#9), added missing return statement to ghc::filesystem::path::generic_string()
• Added checks to hopefully better compile against Android NDK. There where no tests run yet, so feedback is needed to actually call this supported.
• filesystem.h was renamed filesystem.hpp to better reflect that it is a c++ language header.

• Bugfix for (#6), where ghc::filesystem::remove() and ghc::filesystem::remove_all() both are now able to remove a single file and both will not raise an error if the path doesn't exist.
• Merged pull request (#7), a typo leading to setting error code instead of comparing it in ghc::filesystem::remove() under Windows.
• Bugfix for ((#8), the Windows version of ghc::filesystem::directory_iterator now releases resources when reaching end() like the POSIX one does.

• Bugfix for (#4), missing error_code propagation in ghc::filesystem::copy() and ghc::filesystem::remove_all fixed.
• Bugfix for (#5), added missing std namespace in ghc::filesystem::recursive_directory_iterator::difference_type.

• Bugfix for (#3), fixed missing inlines and added test to ensure including into multiple implementation files works as expected.
• Building tests with -Wall -Wextra -Werror and fixed resulting issues.

• Updated catch2 to v2.4.0.
• Refactored fs.op.permissions test to work with all tested std::filesystem implementations (gcc, clang, msvc++).
• Added helper class ghc::filesystem::u8arguments as argv converter, to help follow the UTF-8 path on windows. Simply instantiate it with argc and argv and it will fetch the Unicode version of the command line and convert it to UTF-8. The destructor reverts the change.
• Added examples folder with hopefully some usefull example usage. Examples are tested (and build) with ghc::filesystem and C++17 std::filesystem when available.
• Starting with this version, only even patch level versions will be tagged and odd patch levels mark in-between non-stable wip states.
• Tests can now also be run against MS version of std::filesystem for comparison.
• Added missing fstream include.
• Removed non-conforming C99 timespec/timeval usage.
• Fixed some integer type mismatches that could lead to warnings.
• Fixed chrono conversion issues in test and example on clang 7.0.0.

• Bugfix: ghc::filesystem::canonical now sees empty path as non-existant and reports an error. Due to this ghc::filesystem::weakly_canonical now returns relative paths for non-existant argument paths. (#1)
• Bugfix: ghc::filesystem::remove_all now also counts directories removed (#2)
• Bugfix: recursive_directory_iterator tests didn't respect equality domain issues and dereferencapable constraints, leading to fails on std::filesystem tests.
• Bugfix: Some noexcept tagged methods and functions could indirectly throw exceptions due to UFT-8 decoding issues.
• std_filesystem_test is now also generated if LLVM/clang 7.0.0 is found.

This was the first public release version. It implements the full range of C++17 std::filesystem, as far as possible without other C++17 dependencies.