vcpkg/toolsrc/include/vcpkg-test/util.h

102 lines
4.0 KiB
C
Raw Normal View History

#include <catch2/catch.hpp>
[vcpkg] Make Filesystem::remove_all faster #7570 I added benchmarks to measure how fast the parallel remove_all code was -- it turns out, about 3x slower than stdfs::remove_all. Since this was the case, I removed all of the parallelism and rewrote it serially, and ended up about 30% faster than stdfs::remove_all (in addition to supporting symlinks). In addition, I did the following three orthogonal changes: - simplified the work queue, basing it on Billy O'Neal's idea - Fix warnings on older versions of compilers in tests, by splitting the pragmas out of pch.h. - Ran clang-format on some files In fixing up remove_all, the following changes were made: - On Windows, regular symlinks and directory symlinks are distinct; as an example, to remove directory symlinks (and junctions, for that matter), one must use RemoveDirectory. Only on Windows, I added new `file_type` and `file_status` types, with `file_type` including a new `directory_symlink` enumerator, and `file_status` being exactly the same as the old one except using the new `file_type`. On Unix, I didn't make that change since they don't make a distinction. - I added new `symlink_status` and `status` functions which use the new `file_status` on Windows. - I made `Filesystem::exists` call `fs::exists(status(p))`, as opposed to the old version which called `stdfs::exists` directly. - Added benchmarks to `vcpkg-test/files.cpp`. They test the performance of `remove_all` on small directories (~20 files), with symlinks and without, and on large directories (~2000 files), with symlinks and without.
2019-08-03 07:49:45 +08:00
#include <vcpkg/pragmas.h>
2019-07-20 03:56:24 +08:00
#include <vcpkg/base/files.h>
#include <vcpkg/statusparagraph.h>
#include <memory>
[vcpkg] Make Filesystem::remove_all faster #7570 I added benchmarks to measure how fast the parallel remove_all code was -- it turns out, about 3x slower than stdfs::remove_all. Since this was the case, I removed all of the parallelism and rewrote it serially, and ended up about 30% faster than stdfs::remove_all (in addition to supporting symlinks). In addition, I did the following three orthogonal changes: - simplified the work queue, basing it on Billy O'Neal's idea - Fix warnings on older versions of compilers in tests, by splitting the pragmas out of pch.h. - Ran clang-format on some files In fixing up remove_all, the following changes were made: - On Windows, regular symlinks and directory symlinks are distinct; as an example, to remove directory symlinks (and junctions, for that matter), one must use RemoveDirectory. Only on Windows, I added new `file_type` and `file_status` types, with `file_type` including a new `directory_symlink` enumerator, and `file_status` being exactly the same as the old one except using the new `file_type`. On Unix, I didn't make that change since they don't make a distinction. - I added new `symlink_status` and `status` functions which use the new `file_status` on Windows. - I made `Filesystem::exists` call `fs::exists(status(p))`, as opposed to the old version which called `stdfs::exists` directly. - Added benchmarks to `vcpkg-test/files.cpp`. They test the performance of `remove_all` on small directories (~20 files), with symlinks and without, and on large directories (~2000 files), with symlinks and without.
2019-08-03 07:49:45 +08:00
#define CHECK_EC(ec) \
do \
{ \
if (ec) \
{ \
FAIL(ec.message()); \
} \
} while (0)
[vcpkg] Make Filesystem::remove_all faster #7570 I added benchmarks to measure how fast the parallel remove_all code was -- it turns out, about 3x slower than stdfs::remove_all. Since this was the case, I removed all of the parallelism and rewrote it serially, and ended up about 30% faster than stdfs::remove_all (in addition to supporting symlinks). In addition, I did the following three orthogonal changes: - simplified the work queue, basing it on Billy O'Neal's idea - Fix warnings on older versions of compilers in tests, by splitting the pragmas out of pch.h. - Ran clang-format on some files In fixing up remove_all, the following changes were made: - On Windows, regular symlinks and directory symlinks are distinct; as an example, to remove directory symlinks (and junctions, for that matter), one must use RemoveDirectory. Only on Windows, I added new `file_type` and `file_status` types, with `file_type` including a new `directory_symlink` enumerator, and `file_status` being exactly the same as the old one except using the new `file_type`. On Unix, I didn't make that change since they don't make a distinction. - I added new `symlink_status` and `status` functions which use the new `file_status` on Windows. - I made `Filesystem::exists` call `fs::exists(status(p))`, as opposed to the old version which called `stdfs::exists` directly. - Added benchmarks to `vcpkg-test/files.cpp`. They test the performance of `remove_all` on small directories (~20 files), with symlinks and without, and on large directories (~2000 files), with symlinks and without.
2019-08-03 07:49:45 +08:00
namespace vcpkg::Test
{
[vcpkg] Add Supports: field. Use contents of triplets instead of names for dependency resolution. (#8601) * remove unfinished "supports" tag * extract "supports" from control files But do nothing with the value * Start `Supports` documentation * Use Supports in a bunch of control files I only tried matching the already existing logic in the portfile.cmake. * Cmake var provider (#8) * Cmake var provider (#9) * fix windows build (#10) * Add missing files to build * Fix test (#11) * adding hooks for cmake variables in expressions * Adding hooks for 'supports' in CI test * Fix test (#12) * Add overrides to evaluation environment * use "supported" tag in CI testing * cleanup comment * Fix issues with PR * [var_provider] Get library linkage variables from triplet * Fix compilation errors in tests * Add unimplemented functions * Fix unit tests part 1 * Fix issue when buildtrees dir does not exist * Change binary output hash * Fix handling of * feature * Add core feature when using * * Do not add Default-Features when installing 'core' * [vcpkg] WIP. 6 failing tests. * [vcpkg] WIP. 1 failing tests. * [vcpkg] WIP. 0 failing tests. * [vcpkg] Removed 'remove_graph'. 0 failing tests. * [vcpkg] Removed 'install_graph'. 0 failing tests. * [vcpkg] Remove AnyAction; replace with ActionPlan * [vcpkg] Minor cleanup. * [vcpkg][z3][qt5-connectivity][qt5-purchasing] Improve error messages while parsing. Fix a few trivial port issues. * [vcpkg] Work around ICE with MSVC v140 * [vcpkg] Add purge on fail to decompress for CI * [vcpkg] Fix parsing of nested parentheses in qualifiers * [vcpkg] Fix Linux builds (explicit qualification in declaration) * [vcpkg] Fix Build-Depends implying default features. Fix qualified dependencies regression. * [mmx] Add to skip list and full rebuild -- mmx causes problems by installing 'sched.h' * [libpqxx][mqtt-cpp] Prevent installing include/CMakeLists.txt * [cppitertools] Fix installed include namespace (should be include/cppitertools) * [libsoundio] Move headers into soundio/ subdirectory as per original cmake * [ci.baseline] Temporarily skip charls due to conflict with dcmtk * [vcpkg] Add restricted include files post build check -- bump global abi version * [libsoundio] Hotfix stray line in portfile * [vcpkg] Fix regression: CMake information was not being displayed for build-and-install actions * [jsonnet] Fix installation of internal headers; use system nlohmann-json * [grpc][upb] Teach grpc to use packaged upb. Add find_package(upb). Remove inappropriate upb features. * [zfp] Move problematic 'include/bitstream.h' to 'include/zfp/bitstream.h' * [x265] Bump control version to trigger rebuild after zfp conflict * [akali] Disable parallel configure * [dirent][dlfcn-win32][getopt-win32][pthreads] Grandfather into VCPKG_POLICY_ALLOW_RESTRICTED_HEADERS * [ci.baseline] Update baseline for improved upb support * [tgui] Disable parallel configure * [libiconv] Enable VCPKG_POLICY_ALLOW_RESTRICTED_HEADERS * [aws-sdk-cpp] Disable parallel configure * [vcpkg] Implement policy VCPKG_POLICY_ALLOW_RESTRICTED_HEADERS * [aws-sdk-cpp] Fix amount of escaping semicolons -- Note: I do not know the root cause requiring this change * [libodb-sqlite] Fix configuring into source directory * [gettext] Grandfather into VCPKG_POLICY_ALLOW_RESTRICTED_HEADERS * [libodb] DISABLE_PARALLEL_CONFIGURE * [vcpkg] Add 'config.h' and 'local.h' to restricted header list * [mcpp] Remove unused and problematic include 'config.h' from installed files * [teemo] Move installed headers into subdirectory to prevent conflicts with x265 * [ci.baseline] Update current OSX. Skip libmesh on all platforms due to heavy conflicts. * [vcpkg] Add 'slice.h' as a restricted header * [osg] Improve accuracy of dependencies (disable some, add some to Depends) * [vcpkg] Skip invoking a subprocess for 0 specs in load_tag_vars * [ci.baseline] Skip mongo-c-driver on osx due to flakiness * [teemo] Fix incorrect include file read * [osg] Fix dependency typo: glut -> freeglut * [vcpkg] Recover some lost performance with the addition of vcpkg_get_tags. A huge performance cost was loading the triplet files over and over; instead, we splice the sources into a macro and load it once, then just call that macro for each port. Remove use of hashing because we aren't cross-process-safe anyway (global static will do instead). * [vcpkg] Change Supports atom 'windows' to include UWP. Improve Supports field documentation. * [vcpkg] Add docs for VCPKG_ENV_PASSTHROUGH and VCPKG_DEP_INFO_OVERRIDE_VARS * Fix typo Co-authored-by: Curtis J Bezault <curtbezault@gmail.com> Co-authored-by: Victor Romero <romerosanchezv@gmail.com> Co-authored-by: Robert Schumacher <roschuma@microsoft.com>
2020-02-04 06:22:52 +08:00
std::unique_ptr<SourceControlFile> make_control_file(
const char* name,
const char* depends,
const std::vector<std::pair<const char*, const char*>>& features = {},
const std::vector<const char*>& default_features = {});
inline auto test_parse_control_file(const std::vector<std::unordered_map<std::string, std::string>>& v)
{
std::vector<vcpkg::Parse::Paragraph> pghs;
for (auto&& p : v)
{
pghs.emplace_back();
for (auto&& kv : p)
pghs.back().emplace(kv.first, std::make_pair(kv.second, vcpkg::Parse::TextRowCol{}));
}
return vcpkg::SourceControlFile::parse_control_file("", std::move(pghs));
}
[vcpkg] Make Filesystem::remove_all faster #7570 I added benchmarks to measure how fast the parallel remove_all code was -- it turns out, about 3x slower than stdfs::remove_all. Since this was the case, I removed all of the parallelism and rewrote it serially, and ended up about 30% faster than stdfs::remove_all (in addition to supporting symlinks). In addition, I did the following three orthogonal changes: - simplified the work queue, basing it on Billy O'Neal's idea - Fix warnings on older versions of compilers in tests, by splitting the pragmas out of pch.h. - Ran clang-format on some files In fixing up remove_all, the following changes were made: - On Windows, regular symlinks and directory symlinks are distinct; as an example, to remove directory symlinks (and junctions, for that matter), one must use RemoveDirectory. Only on Windows, I added new `file_type` and `file_status` types, with `file_type` including a new `directory_symlink` enumerator, and `file_status` being exactly the same as the old one except using the new `file_type`. On Unix, I didn't make that change since they don't make a distinction. - I added new `symlink_status` and `status` functions which use the new `file_status` on Windows. - I made `Filesystem::exists` call `fs::exists(status(p))`, as opposed to the old version which called `stdfs::exists` directly. - Added benchmarks to `vcpkg-test/files.cpp`. They test the performance of `remove_all` on small directories (~20 files), with symlinks and without, and on large directories (~2000 files), with symlinks and without.
2019-08-03 07:49:45 +08:00
std::unique_ptr<vcpkg::StatusParagraph> make_status_pgh(const char* name,
const char* depends = "",
const char* default_features = "",
const char* triplet = "x86-windows");
[vcpkg] Make Filesystem::remove_all faster #7570 I added benchmarks to measure how fast the parallel remove_all code was -- it turns out, about 3x slower than stdfs::remove_all. Since this was the case, I removed all of the parallelism and rewrote it serially, and ended up about 30% faster than stdfs::remove_all (in addition to supporting symlinks). In addition, I did the following three orthogonal changes: - simplified the work queue, basing it on Billy O'Neal's idea - Fix warnings on older versions of compilers in tests, by splitting the pragmas out of pch.h. - Ran clang-format on some files In fixing up remove_all, the following changes were made: - On Windows, regular symlinks and directory symlinks are distinct; as an example, to remove directory symlinks (and junctions, for that matter), one must use RemoveDirectory. Only on Windows, I added new `file_type` and `file_status` types, with `file_type` including a new `directory_symlink` enumerator, and `file_status` being exactly the same as the old one except using the new `file_type`. On Unix, I didn't make that change since they don't make a distinction. - I added new `symlink_status` and `status` functions which use the new `file_status` on Windows. - I made `Filesystem::exists` call `fs::exists(status(p))`, as opposed to the old version which called `stdfs::exists` directly. - Added benchmarks to `vcpkg-test/files.cpp`. They test the performance of `remove_all` on small directories (~20 files), with symlinks and without, and on large directories (~2000 files), with symlinks and without.
2019-08-03 07:49:45 +08:00
std::unique_ptr<vcpkg::StatusParagraph> make_status_feature_pgh(const char* name,
const char* feature,
const char* depends = "",
const char* triplet = "x86-windows");
[vcpkg] Add Supports: field. Use contents of triplets instead of names for dependency resolution. (#8601) * remove unfinished "supports" tag * extract "supports" from control files But do nothing with the value * Start `Supports` documentation * Use Supports in a bunch of control files I only tried matching the already existing logic in the portfile.cmake. * Cmake var provider (#8) * Cmake var provider (#9) * fix windows build (#10) * Add missing files to build * Fix test (#11) * adding hooks for cmake variables in expressions * Adding hooks for 'supports' in CI test * Fix test (#12) * Add overrides to evaluation environment * use "supported" tag in CI testing * cleanup comment * Fix issues with PR * [var_provider] Get library linkage variables from triplet * Fix compilation errors in tests * Add unimplemented functions * Fix unit tests part 1 * Fix issue when buildtrees dir does not exist * Change binary output hash * Fix handling of * feature * Add core feature when using * * Do not add Default-Features when installing 'core' * [vcpkg] WIP. 6 failing tests. * [vcpkg] WIP. 1 failing tests. * [vcpkg] WIP. 0 failing tests. * [vcpkg] Removed 'remove_graph'. 0 failing tests. * [vcpkg] Removed 'install_graph'. 0 failing tests. * [vcpkg] Remove AnyAction; replace with ActionPlan * [vcpkg] Minor cleanup. * [vcpkg][z3][qt5-connectivity][qt5-purchasing] Improve error messages while parsing. Fix a few trivial port issues. * [vcpkg] Work around ICE with MSVC v140 * [vcpkg] Add purge on fail to decompress for CI * [vcpkg] Fix parsing of nested parentheses in qualifiers * [vcpkg] Fix Linux builds (explicit qualification in declaration) * [vcpkg] Fix Build-Depends implying default features. Fix qualified dependencies regression. * [mmx] Add to skip list and full rebuild -- mmx causes problems by installing 'sched.h' * [libpqxx][mqtt-cpp] Prevent installing include/CMakeLists.txt * [cppitertools] Fix installed include namespace (should be include/cppitertools) * [libsoundio] Move headers into soundio/ subdirectory as per original cmake * [ci.baseline] Temporarily skip charls due to conflict with dcmtk * [vcpkg] Add restricted include files post build check -- bump global abi version * [libsoundio] Hotfix stray line in portfile * [vcpkg] Fix regression: CMake information was not being displayed for build-and-install actions * [jsonnet] Fix installation of internal headers; use system nlohmann-json * [grpc][upb] Teach grpc to use packaged upb. Add find_package(upb). Remove inappropriate upb features. * [zfp] Move problematic 'include/bitstream.h' to 'include/zfp/bitstream.h' * [x265] Bump control version to trigger rebuild after zfp conflict * [akali] Disable parallel configure * [dirent][dlfcn-win32][getopt-win32][pthreads] Grandfather into VCPKG_POLICY_ALLOW_RESTRICTED_HEADERS * [ci.baseline] Update baseline for improved upb support * [tgui] Disable parallel configure * [libiconv] Enable VCPKG_POLICY_ALLOW_RESTRICTED_HEADERS * [aws-sdk-cpp] Disable parallel configure * [vcpkg] Implement policy VCPKG_POLICY_ALLOW_RESTRICTED_HEADERS * [aws-sdk-cpp] Fix amount of escaping semicolons -- Note: I do not know the root cause requiring this change * [libodb-sqlite] Fix configuring into source directory * [gettext] Grandfather into VCPKG_POLICY_ALLOW_RESTRICTED_HEADERS * [libodb] DISABLE_PARALLEL_CONFIGURE * [vcpkg] Add 'config.h' and 'local.h' to restricted header list * [mcpp] Remove unused and problematic include 'config.h' from installed files * [teemo] Move installed headers into subdirectory to prevent conflicts with x265 * [ci.baseline] Update current OSX. Skip libmesh on all platforms due to heavy conflicts. * [vcpkg] Add 'slice.h' as a restricted header * [osg] Improve accuracy of dependencies (disable some, add some to Depends) * [vcpkg] Skip invoking a subprocess for 0 specs in load_tag_vars * [ci.baseline] Skip mongo-c-driver on osx due to flakiness * [teemo] Fix incorrect include file read * [osg] Fix dependency typo: glut -> freeglut * [vcpkg] Recover some lost performance with the addition of vcpkg_get_tags. A huge performance cost was loading the triplet files over and over; instead, we splice the sources into a macro and load it once, then just call that macro for each port. Remove use of hashing because we aren't cross-process-safe anyway (global static will do instead). * [vcpkg] Change Supports atom 'windows' to include UWP. Improve Supports field documentation. * [vcpkg] Add docs for VCPKG_ENV_PASSTHROUGH and VCPKG_DEP_INFO_OVERRIDE_VARS * Fix typo Co-authored-by: Curtis J Bezault <curtbezault@gmail.com> Co-authored-by: Victor Romero <romerosanchezv@gmail.com> Co-authored-by: Robert Schumacher <roschuma@microsoft.com>
2020-02-04 06:22:52 +08:00
/// <summary>
/// Map of source control files by their package name.
/// </summary>
struct PackageSpecMap
{
std::unordered_map<std::string, SourceControlFileLocation> map;
Triplet triplet;
PackageSpecMap(Triplet t = Triplet::X86_WINDOWS) noexcept : triplet(t) {}
[vcpkg] Add Supports: field. Use contents of triplets instead of names for dependency resolution. (#8601) * remove unfinished "supports" tag * extract "supports" from control files But do nothing with the value * Start `Supports` documentation * Use Supports in a bunch of control files I only tried matching the already existing logic in the portfile.cmake. * Cmake var provider (#8) * Cmake var provider (#9) * fix windows build (#10) * Add missing files to build * Fix test (#11) * adding hooks for cmake variables in expressions * Adding hooks for 'supports' in CI test * Fix test (#12) * Add overrides to evaluation environment * use "supported" tag in CI testing * cleanup comment * Fix issues with PR * [var_provider] Get library linkage variables from triplet * Fix compilation errors in tests * Add unimplemented functions * Fix unit tests part 1 * Fix issue when buildtrees dir does not exist * Change binary output hash * Fix handling of * feature * Add core feature when using * * Do not add Default-Features when installing 'core' * [vcpkg] WIP. 6 failing tests. * [vcpkg] WIP. 1 failing tests. * [vcpkg] WIP. 0 failing tests. * [vcpkg] Removed 'remove_graph'. 0 failing tests. * [vcpkg] Removed 'install_graph'. 0 failing tests. * [vcpkg] Remove AnyAction; replace with ActionPlan * [vcpkg] Minor cleanup. * [vcpkg][z3][qt5-connectivity][qt5-purchasing] Improve error messages while parsing. Fix a few trivial port issues. * [vcpkg] Work around ICE with MSVC v140 * [vcpkg] Add purge on fail to decompress for CI * [vcpkg] Fix parsing of nested parentheses in qualifiers * [vcpkg] Fix Linux builds (explicit qualification in declaration) * [vcpkg] Fix Build-Depends implying default features. Fix qualified dependencies regression. * [mmx] Add to skip list and full rebuild -- mmx causes problems by installing 'sched.h' * [libpqxx][mqtt-cpp] Prevent installing include/CMakeLists.txt * [cppitertools] Fix installed include namespace (should be include/cppitertools) * [libsoundio] Move headers into soundio/ subdirectory as per original cmake * [ci.baseline] Temporarily skip charls due to conflict with dcmtk * [vcpkg] Add restricted include files post build check -- bump global abi version * [libsoundio] Hotfix stray line in portfile * [vcpkg] Fix regression: CMake information was not being displayed for build-and-install actions * [jsonnet] Fix installation of internal headers; use system nlohmann-json * [grpc][upb] Teach grpc to use packaged upb. Add find_package(upb). Remove inappropriate upb features. * [zfp] Move problematic 'include/bitstream.h' to 'include/zfp/bitstream.h' * [x265] Bump control version to trigger rebuild after zfp conflict * [akali] Disable parallel configure * [dirent][dlfcn-win32][getopt-win32][pthreads] Grandfather into VCPKG_POLICY_ALLOW_RESTRICTED_HEADERS * [ci.baseline] Update baseline for improved upb support * [tgui] Disable parallel configure * [libiconv] Enable VCPKG_POLICY_ALLOW_RESTRICTED_HEADERS * [aws-sdk-cpp] Disable parallel configure * [vcpkg] Implement policy VCPKG_POLICY_ALLOW_RESTRICTED_HEADERS * [aws-sdk-cpp] Fix amount of escaping semicolons -- Note: I do not know the root cause requiring this change * [libodb-sqlite] Fix configuring into source directory * [gettext] Grandfather into VCPKG_POLICY_ALLOW_RESTRICTED_HEADERS * [libodb] DISABLE_PARALLEL_CONFIGURE * [vcpkg] Add 'config.h' and 'local.h' to restricted header list * [mcpp] Remove unused and problematic include 'config.h' from installed files * [teemo] Move installed headers into subdirectory to prevent conflicts with x265 * [ci.baseline] Update current OSX. Skip libmesh on all platforms due to heavy conflicts. * [vcpkg] Add 'slice.h' as a restricted header * [osg] Improve accuracy of dependencies (disable some, add some to Depends) * [vcpkg] Skip invoking a subprocess for 0 specs in load_tag_vars * [ci.baseline] Skip mongo-c-driver on osx due to flakiness * [teemo] Fix incorrect include file read * [osg] Fix dependency typo: glut -> freeglut * [vcpkg] Recover some lost performance with the addition of vcpkg_get_tags. A huge performance cost was loading the triplet files over and over; instead, we splice the sources into a macro and load it once, then just call that macro for each port. Remove use of hashing because we aren't cross-process-safe anyway (global static will do instead). * [vcpkg] Change Supports atom 'windows' to include UWP. Improve Supports field documentation. * [vcpkg] Add docs for VCPKG_ENV_PASSTHROUGH and VCPKG_DEP_INFO_OVERRIDE_VARS * Fix typo Co-authored-by: Curtis J Bezault <curtbezault@gmail.com> Co-authored-by: Victor Romero <romerosanchezv@gmail.com> Co-authored-by: Robert Schumacher <roschuma@microsoft.com>
2020-02-04 06:22:52 +08:00
PackageSpec emplace(const char* name,
const char* depends = "",
const std::vector<std::pair<const char*, const char*>>& features = {},
const std::vector<const char*>& default_features = {});
PackageSpec emplace(vcpkg::SourceControlFileLocation&& scfl);
};
[vcpkg] Make Filesystem::remove_all faster #7570 I added benchmarks to measure how fast the parallel remove_all code was -- it turns out, about 3x slower than stdfs::remove_all. Since this was the case, I removed all of the parallelism and rewrote it serially, and ended up about 30% faster than stdfs::remove_all (in addition to supporting symlinks). In addition, I did the following three orthogonal changes: - simplified the work queue, basing it on Billy O'Neal's idea - Fix warnings on older versions of compilers in tests, by splitting the pragmas out of pch.h. - Ran clang-format on some files In fixing up remove_all, the following changes were made: - On Windows, regular symlinks and directory symlinks are distinct; as an example, to remove directory symlinks (and junctions, for that matter), one must use RemoveDirectory. Only on Windows, I added new `file_type` and `file_status` types, with `file_type` including a new `directory_symlink` enumerator, and `file_status` being exactly the same as the old one except using the new `file_type`. On Unix, I didn't make that change since they don't make a distinction. - I added new `symlink_status` and `status` functions which use the new `file_status` on Windows. - I made `Filesystem::exists` call `fs::exists(status(p))`, as opposed to the old version which called `stdfs::exists` directly. - Added benchmarks to `vcpkg-test/files.cpp`. They test the performance of `remove_all` on small directories (~20 files), with symlinks and without, and on large directories (~2000 files), with symlinks and without.
2019-08-03 07:49:45 +08:00
template<class T, class S>
T&& unwrap(vcpkg::ExpectedT<T, S>&& p)
{
REQUIRE(p.has_value());
return std::move(*p.get());
}
[vcpkg] Make Filesystem::remove_all faster #7570 I added benchmarks to measure how fast the parallel remove_all code was -- it turns out, about 3x slower than stdfs::remove_all. Since this was the case, I removed all of the parallelism and rewrote it serially, and ended up about 30% faster than stdfs::remove_all (in addition to supporting symlinks). In addition, I did the following three orthogonal changes: - simplified the work queue, basing it on Billy O'Neal's idea - Fix warnings on older versions of compilers in tests, by splitting the pragmas out of pch.h. - Ran clang-format on some files In fixing up remove_all, the following changes were made: - On Windows, regular symlinks and directory symlinks are distinct; as an example, to remove directory symlinks (and junctions, for that matter), one must use RemoveDirectory. Only on Windows, I added new `file_type` and `file_status` types, with `file_type` including a new `directory_symlink` enumerator, and `file_status` being exactly the same as the old one except using the new `file_type`. On Unix, I didn't make that change since they don't make a distinction. - I added new `symlink_status` and `status` functions which use the new `file_status` on Windows. - I made `Filesystem::exists` call `fs::exists(status(p))`, as opposed to the old version which called `stdfs::exists` directly. - Added benchmarks to `vcpkg-test/files.cpp`. They test the performance of `remove_all` on small directories (~20 files), with symlinks and without, and on large directories (~2000 files), with symlinks and without.
2019-08-03 07:49:45 +08:00
template<class T>
T&& unwrap(vcpkg::Optional<T>&& opt)
{
REQUIRE(opt.has_value());
return std::move(*opt.get());
}
struct AllowSymlinks
{
enum Tag : bool
{
No = false,
Yes = true,
} tag;
constexpr AllowSymlinks(Tag tag) noexcept : tag(tag) {}
constexpr explicit AllowSymlinks(bool b) noexcept : tag(b ? Yes : No) {}
[vcpkg] Make Filesystem::remove_all faster #7570 I added benchmarks to measure how fast the parallel remove_all code was -- it turns out, about 3x slower than stdfs::remove_all. Since this was the case, I removed all of the parallelism and rewrote it serially, and ended up about 30% faster than stdfs::remove_all (in addition to supporting symlinks). In addition, I did the following three orthogonal changes: - simplified the work queue, basing it on Billy O'Neal's idea - Fix warnings on older versions of compilers in tests, by splitting the pragmas out of pch.h. - Ran clang-format on some files In fixing up remove_all, the following changes were made: - On Windows, regular symlinks and directory symlinks are distinct; as an example, to remove directory symlinks (and junctions, for that matter), one must use RemoveDirectory. Only on Windows, I added new `file_type` and `file_status` types, with `file_type` including a new `directory_symlink` enumerator, and `file_status` being exactly the same as the old one except using the new `file_type`. On Unix, I didn't make that change since they don't make a distinction. - I added new `symlink_status` and `status` functions which use the new `file_status` on Windows. - I made `Filesystem::exists` call `fs::exists(status(p))`, as opposed to the old version which called `stdfs::exists` directly. - Added benchmarks to `vcpkg-test/files.cpp`. They test the performance of `remove_all` on small directories (~20 files), with symlinks and without, and on large directories (~2000 files), with symlinks and without.
2019-08-03 07:49:45 +08:00
constexpr operator bool() const noexcept { return tag == Yes; }
};
2019-07-20 03:56:24 +08:00
[vcpkg] Make Filesystem::remove_all faster #7570 I added benchmarks to measure how fast the parallel remove_all code was -- it turns out, about 3x slower than stdfs::remove_all. Since this was the case, I removed all of the parallelism and rewrote it serially, and ended up about 30% faster than stdfs::remove_all (in addition to supporting symlinks). In addition, I did the following three orthogonal changes: - simplified the work queue, basing it on Billy O'Neal's idea - Fix warnings on older versions of compilers in tests, by splitting the pragmas out of pch.h. - Ran clang-format on some files In fixing up remove_all, the following changes were made: - On Windows, regular symlinks and directory symlinks are distinct; as an example, to remove directory symlinks (and junctions, for that matter), one must use RemoveDirectory. Only on Windows, I added new `file_type` and `file_status` types, with `file_type` including a new `directory_symlink` enumerator, and `file_status` being exactly the same as the old one except using the new `file_type`. On Unix, I didn't make that change since they don't make a distinction. - I added new `symlink_status` and `status` functions which use the new `file_status` on Windows. - I made `Filesystem::exists` call `fs::exists(status(p))`, as opposed to the old version which called `stdfs::exists` directly. - Added benchmarks to `vcpkg-test/files.cpp`. They test the performance of `remove_all` on small directories (~20 files), with symlinks and without, and on large directories (~2000 files), with symlinks and without.
2019-08-03 07:49:45 +08:00
AllowSymlinks can_create_symlinks() noexcept;
2019-07-20 03:56:24 +08:00
[vcpkg] Make Filesystem::remove_all faster #7570 I added benchmarks to measure how fast the parallel remove_all code was -- it turns out, about 3x slower than stdfs::remove_all. Since this was the case, I removed all of the parallelism and rewrote it serially, and ended up about 30% faster than stdfs::remove_all (in addition to supporting symlinks). In addition, I did the following three orthogonal changes: - simplified the work queue, basing it on Billy O'Neal's idea - Fix warnings on older versions of compilers in tests, by splitting the pragmas out of pch.h. - Ran clang-format on some files In fixing up remove_all, the following changes were made: - On Windows, regular symlinks and directory symlinks are distinct; as an example, to remove directory symlinks (and junctions, for that matter), one must use RemoveDirectory. Only on Windows, I added new `file_type` and `file_status` types, with `file_type` including a new `directory_symlink` enumerator, and `file_status` being exactly the same as the old one except using the new `file_type`. On Unix, I didn't make that change since they don't make a distinction. - I added new `symlink_status` and `status` functions which use the new `file_status` on Windows. - I made `Filesystem::exists` call `fs::exists(status(p))`, as opposed to the old version which called `stdfs::exists` directly. - Added benchmarks to `vcpkg-test/files.cpp`. They test the performance of `remove_all` on small directories (~20 files), with symlinks and without, and on large directories (~2000 files), with symlinks and without.
2019-08-03 07:49:45 +08:00
const fs::path& base_temporary_directory() noexcept;
[vcpkg] Make Filesystem::remove_all faster #7570 I added benchmarks to measure how fast the parallel remove_all code was -- it turns out, about 3x slower than stdfs::remove_all. Since this was the case, I removed all of the parallelism and rewrote it serially, and ended up about 30% faster than stdfs::remove_all (in addition to supporting symlinks). In addition, I did the following three orthogonal changes: - simplified the work queue, basing it on Billy O'Neal's idea - Fix warnings on older versions of compilers in tests, by splitting the pragmas out of pch.h. - Ran clang-format on some files In fixing up remove_all, the following changes were made: - On Windows, regular symlinks and directory symlinks are distinct; as an example, to remove directory symlinks (and junctions, for that matter), one must use RemoveDirectory. Only on Windows, I added new `file_type` and `file_status` types, with `file_type` including a new `directory_symlink` enumerator, and `file_status` being exactly the same as the old one except using the new `file_type`. On Unix, I didn't make that change since they don't make a distinction. - I added new `symlink_status` and `status` functions which use the new `file_status` on Windows. - I made `Filesystem::exists` call `fs::exists(status(p))`, as opposed to the old version which called `stdfs::exists` directly. - Added benchmarks to `vcpkg-test/files.cpp`. They test the performance of `remove_all` on small directories (~20 files), with symlinks and without, and on large directories (~2000 files), with symlinks and without.
2019-08-03 07:49:45 +08:00
void create_symlink(const fs::path& file, const fs::path& target, std::error_code& ec);
[vcpkg] Make Filesystem::remove_all faster #7570 I added benchmarks to measure how fast the parallel remove_all code was -- it turns out, about 3x slower than stdfs::remove_all. Since this was the case, I removed all of the parallelism and rewrote it serially, and ended up about 30% faster than stdfs::remove_all (in addition to supporting symlinks). In addition, I did the following three orthogonal changes: - simplified the work queue, basing it on Billy O'Neal's idea - Fix warnings on older versions of compilers in tests, by splitting the pragmas out of pch.h. - Ran clang-format on some files In fixing up remove_all, the following changes were made: - On Windows, regular symlinks and directory symlinks are distinct; as an example, to remove directory symlinks (and junctions, for that matter), one must use RemoveDirectory. Only on Windows, I added new `file_type` and `file_status` types, with `file_type` including a new `directory_symlink` enumerator, and `file_status` being exactly the same as the old one except using the new `file_type`. On Unix, I didn't make that change since they don't make a distinction. - I added new `symlink_status` and `status` functions which use the new `file_status` on Windows. - I made `Filesystem::exists` call `fs::exists(status(p))`, as opposed to the old version which called `stdfs::exists` directly. - Added benchmarks to `vcpkg-test/files.cpp`. They test the performance of `remove_all` on small directories (~20 files), with symlinks and without, and on large directories (~2000 files), with symlinks and without.
2019-08-03 07:49:45 +08:00
void create_directory_symlink(const fs::path& file, const fs::path& target, std::error_code& ec);
}