Summary: Instead of using setCallbackAndDetach (which is hard to apply in all cases because of cancellation), this diff fixes releaseDefferedExecutor.

Reviewed By: yfeldblum

Differential Revision: D16189338

fbshipit-source-id: 47ef192dd0cec3c13824071c8ba7e1dd06968a70

Reviewed By: yfeldblum

Differential Revision: D15915100

fbshipit-source-id: afc131bdf143ac0eec15523e257545d63207cb34

Summary:
This are noisy and don't provide real value, so remove
the webhooks.

Reviewed By: yns88

Differential Revision: D16115486

fbshipit-source-id: 1daa2d78f4e786a42dcc5779ccd30c36e02d2445

Summary:
Some changes to get Poly compiling with VS2019 (and correcting an oversight from https://github.com/facebook/folly/pull/1174)

The change to `detail/TypeList.h` is straightforward I think.

The introduction of `getOpsImpl` I find a bit distasteful...I will remark that as of Visual Studio 16.2.0 Preview 3.0 it is not necessary, but it is still needed on Visual Studio 16.1.5 (latest release, non preview).

Orvid in your comment [here](https://github.com/facebook/folly/pull/1178#issuecomment-506864383)  you mentioned the question of supporting VS 2017 or VS 2019 given that Poly had no MSVC support at all beforehand. I re-downloaded VS 2017 on my machine to test against the fixes added here and the situation seemed discouraging with lots more errors and perhaps less obvious workarounds.

I have not made any changes to the CMake stuff for Poly on MSVC: https://github.com/facebook/folly/blob/master/CMakeLists.txt#L235

I'm not sure if once we resolve the issue above we might want to add some logic maybe [here](https://github.com/facebook/folly/blob/master/CMakeLists.txt#L63) to handle which version of MSVC is approved for poly; will wait for further direction on that.
Pull Request resolved: https://github.com/facebook/folly/pull/1182

Reviewed By: Orvid

Differential Revision: D16109504

Pulled By: yfeldblum

fbshipit-source-id: d20b242494b075ce91452215c2025584fe488d65

Summary:
Test AsyncSocketTest.getBufInUse was throwing exceptions, I believe
this should fix those issues. The problem appeared to be that the socket
connection did not always work immediately so we needed to call loop() on the
event base so that an async connection could complete.

Reviewed By: yfeldblum

Differential Revision: D16052111

fbshipit-source-id: 9793e613eea5b17cb3ab00605355102b2f8807d8

Summary:
Folly already has its own lightweight implementation `IteratorFacade`. It depends only a few STL headers thus saves compilation time.
Pull Request resolved: https://github.com/facebook/folly/pull/1187

Reviewed By: LeeHowes

Differential Revision: D16147453

Pulled By: yfeldblum

fbshipit-source-id: 9559a095e242ee7e16659558d505777159f6e534

Summary: Add MicroSpinLock TSAN annotations

Reviewed By: yfeldblum

Differential Revision: D14578441

fbshipit-source-id: e2c9bc1806550d003cc659a86d214c089464f126

Summary:
Context: migration of fbcode from folly::gen to range-v3 by way of pipe-adapters ("pipe fittings") that allow a `|`-based pipeline mixing {std-containers and/or range-v3} with folly::gen, e.g.,

```
auto result = myVec              // anything consistent with range-v3 or these adapters
    | ranges::view::filter(...)  // zero-or-more range-v3 pipe-elems
    | <adapter-goes-here>        // <==**one of the pipe-adapters provided by this task**
    | folly::gen::blah();        // trailing pipe-elems are folly::gen
```

Goal of this task: rename the pipe-adapter folly::gen::detail types so those names are consistent with the names of the corresponding folly::gen::objects.

Reviewed By: yfeldblum

Differential Revision: D16138397

fbshipit-source-id: 3e6145f5bb35c8d2bd14a9a3eb645e153743d499

Summary:
Relocate "folly::gen to range-v3" pipe-adapters into the corresponding adapter-struct (fix compile-time errors, plus improve clarity about the linkage between the `operator|` and corresponding type).

Context: pipe-adapters ("pipe fittings") that allow a `|`-based pipeline mixing {std-containers and/or range-v3} with folly::gen, e.g.,

```
auto result = myVec              // anything consistent with range-v3 or these adapters
    | ranges::view::filter(...)  // zero-or-more range-v3 pipe-elems
    | <adapter-goes-here>        // <==**one of the pipe-adapters provided by this task**
    | folly::gen::blah();        // trailing pipe-elems are folly::gen
```

This diff relocates those folly::gen-to-range-v3 adapters.

Reviewed By: yfeldblum, ericniebler

Differential Revision: D16053008

fbshipit-source-id: 078a95990d5277ac0c4e02452cd8fd36bead8306

Reviewed By: yfeldblum

Differential Revision: D16138503

fbshipit-source-id: 49243eeb6eb24417c912b89beb84d15bae73053f

Summary:
StandardLogHandlerFactory has a good amount of logic in it for handling
options. It would be nice to reuse it when specifying a custom Formatter.

This also exposes the WriterFactory from StreamHandlerFactory, so that a user
can re-use the code to create the default LogWriter when using a custom
Formatter.

Reviewed By: yfeldblum

Differential Revision: D16074437

fbshipit-source-id: d79947e2fe93b8920b83294438a88c4dd871986b

Summary:
Mapping scratch dir <from> -> <to> show up in the output of all the show dir commands on Windows, so removing it.
example: getdeps.py show-inst-dir eden

Reviewed By: wez

Differential Revision: D16092494

fbshipit-source-id: 910288a8d23c1d68c5e70b7b2dbb36ef53a326fc

Summary: This adds a subclass to copy the dynamic dependencies on Windows.

Reviewed By: wez

Differential Revision: D16110433

fbshipit-source-id: 14d876947f3ec504382fef0d459367a7119ff6cb

Reviewed By: yfeldblum

Differential Revision: D16099521

fbshipit-source-id: da89c9954037c168f7591353e42b2632af0b8b31

Summary:
This diff adds a `fixup-dyn-deps` subcommand that is intended to
aid in packaging on multiple platforms.

Its purpose is to copy a set of executable object files from the getdeps
installation directories and place them into an installation staging
area that will then be used to create some kind of package (rpm, tarball etc.).

The dynamic dependencies of the executables are determined and also copied
into the destination area, and the important part: the execute is rewritten
such that it will load the deps out of an alternate installation prefix.

The implementation of this command draws on similar scripts in use for
the watchman and eden packaging on windows and macos.  This diff adds
linux support using the `patchelf` utility.

Reviewed By: pkaush

Differential Revision: D16101902

fbshipit-source-id: 5885125971947139407841e08c0cf9f35fdf5895

Summary: I want to use this logic outside of a builder implementation

Reviewed By: pkaush

Differential Revision: D16101914

fbshipit-source-id: db3c9ac6c84a92ab84a18dddb931953b0a51f127

Summary:
The intent is to use this in a future diff to make packaging a first
class concept; it can be used similarly to how we munge dynamic deps on darwin
systems.

Reviewed By: shri-khare

Differential Revision: D16099931

fbshipit-source-id: c66f994b7c07e903fbf2989946b17a0991e12a91

Summary:
wait.h is a GNU header. The POSIX one is sys/wait.h. Fixes musl warning:

warning redirecting incorrect #include <wait.h> to <sys/wait.h>
Pull Request resolved: https://github.com/facebook/folly/pull/1183

Reviewed By: wez

Differential Revision: D16097571

Pulled By: yfeldblum

fbshipit-source-id: 751e4929bc9f00c99c160be9a7dd904e57ad40ce

Summary:
We're seeing a failure on macOS in a CI environment but don't have much context:

```
Stderr:
+ GETDEPS=opensource/fbcode_builder/getdeps.py
+ opensource/fbcode_builder/getdeps.py test --facebook-internal watchman
Traceback (most recent call last):
  File "opensource/fbcode_builder/getdeps.py", line 436, in <module>
    sys.exit(main())
  File "opensource/fbcode_builder/getdeps.py", line 422, in main
    return args.func(args)
  File "opensource/fbcode_builder/getdeps.py", line 342, in run
    builder.run_tests(install_dirs, schedule_type=args.schedule_type)
  File "/data/sandcastle/boxes/trunk-hg-fbcode-fbsource/fbcode/opensource/fbcode_builder/getdeps/builder.py", line 372, in run_tests
    buck_test_info = list_tests()
  File "/data/sandcastle/boxes/trunk-hg-fbcode-fbsource/fbcode/opensource/fbcode_builder/getdeps/builder.py", line 352, in list_tests
    data = json.loads(output.decode("utf-8"))
  File "/opt/homebrew/Cellar/python27/2.7.10_2/Frameworks/Python.framework/Versions/2.7/lib/python2.7/json/__init__.py", line 338, in loads
    return _default_decoder.decode(s)
  File "/opt/homebrew/Cellar/python27/2.7.10_2/Frameworks/Python.framework/Versions/2.7/lib/python2.7/json/decoder.py", line 366, in decode
    obj, end = self.raw_decode(s, idx=_w(s, 0).end())
  File "/opt/homebrew/Cellar/python27/2.7.10_2/Frameworks/Python.framework/Versions/2.7/lib/python2.7/json/decoder.py", line 384, in raw_decode
    raise ValueError("No JSON object could be decoded")
ValueError: No JSON object could be decoded
```

Add some more context to that error message so we can learn more.

Reviewed By: pkaush

Differential Revision: D16061528

fbshipit-source-id: 23603a5d18651d20641ef1987b7094e73a9b1dbe

Summary: On Windows "--vcvars-path" can be passed to point to the toolchain we want to use.

Reviewed By: wez

Differential Revision: D15926044

fbshipit-source-id: 2b0cde793f7c7f8473b78afde8794640bae351f3

Reviewed By: udippant

Differential Revision: D16075865

fbshipit-source-id: 172aed7ebbfce987c78eea06e588048b0e2a755f

Summary:
NotificationQueue, although mostly not used as such, is a multi consumer queue and it's missing benchmarks for that part.

There where also a couple of issues with original benchmark which this diff addresses:
1. Benchamrks used to do N * number of producers number of iterations and didn't report increased iteration count back to benchmark
2. Benchmark would not run the same case during whole benchmark time as producers where stopped as soon as they produced enough request and consumer would finish without contention

Since these are very tight loops I added a busy loop param which can be varied the same way as batching parameter.
This param adds some work in a form of a busy loop in both producers and consumer to make benchmarking cases, where producers/consumers are actually doing something except from posting to the queue, easier.

Reviewed By: yfeldblum

Differential Revision: D16071580

fbshipit-source-id: c75f2dd9ae8617be09834cdf153a4b4ae06aaaca

Summary: While congested throughput is important it is also important not to regress non congested path too much.

Reviewed By: aary

Differential Revision: D16005428

fbshipit-source-id: 7b98a7ef253fc306b3097f7236b57a5d453ed15d

Summary: [Folly] Check for max value in test semaphore, beyond which posts cannot be tolerated.

Reviewed By: aary

Differential Revision: D15938166

fbshipit-source-id: 1f3fee97a9662dafc5b434c0449831147aa3c976

Summary:
Remove all the autoconf files.
And fully switch to the cmake build

Reviewed By: udippant

Differential Revision: D16066986

fbshipit-source-id: 9e3082dded77ce85449cf5a3a03bed31c16b711f

Summary:
Folly's `folly/lang/Align.h` header defines `folly::max_align_t`, which is usually taken from stdlib. This is to correct an error in legacy implementations of `std::max_align_t`, by logically reconstructing the maximum alignment value in a constexpr context.

`max_align_t` depends on the function `detail::max_align_`, which is supposed to implement a recursive max value function using templates. However, the expression

```
!(a < e) ? a : max_align_(e, es...)
```

does not recursively compute the maximum value over `a`, `e`, `es...`. This expression actually selects the first value that is followed by a lower value.

This error prevents folly from compiling on newer versions of clang, targeting x86, because `folly/CachelinePadded.h` statically checks `max_align_t` in an assertion, and then fails, because it thinks that `AtomicStruct<LifoSemHead, atomic>` is over-aligned.

This corrects the implementation to:

```
!(a < e) ? max_align_(a, es...) : max_align_(e, es...)
```

Pull Request resolved: https://github.com/facebook/folly/pull/1180

Reviewed By: Orvid

Differential Revision: D16079107

Pulled By: yfeldblum

fbshipit-source-id: 13adbd6d1c117b3c547e7116216ee1069129b3c7

Summary:
The source we were downloading from (netfilter.org) should work, but
unfortunately, is donw. This is causing getdeps FBOSS builds to fail. For now,
download from a mirror which seems to be up.

If this becomes a frequent issue with this repository/others, we can consider a
different solution.

Also, not sure how lfs pointer edit was missed by previous commit, added here.

Reviewed By: phshaikh

Differential Revision: D16070052

fbshipit-source-id: dd9278254c77e1a5845ca331a9eeb36f2bf03071

Reviewed By: yfeldblum

Differential Revision: D16054275

fbshipit-source-id: 266f2244fd4c28e1726f4f84d32d4f5d6856ec26

Summary:
Second attempt at https://github.com/facebook/folly/pull/1173

This time the class template specializations are guarded with the `__cpp_noexcept_function_type` feature test macro to maintain compatibility with C++14.

Open to replacing the raw macro with something like `FOLLY_POLY_NOEXCEPT_FUNCTION_TYPE` but that was not my first approach since it's more characters and there is not an extra case to test like with `FOLLY_POLY_NTTP_AUTO`.
Pull Request resolved: https://github.com/facebook/folly/pull/1178

Reviewed By: yfeldblum

Differential Revision: D16057403

Pulled By: Orvid

fbshipit-source-id: 9b538c0f1b0614e69447202d1e4d60ed3fa709cb

Summary: `_mm_crc32_u64` requires SSE 4.2, not just any SSE.

Reviewed By: vener91

Differential Revision: D16032673

fbshipit-source-id: 13cb6299e14184b30a9a04f2644aee2f0cd1a60e

Summary:
Clang defines `_MSC_VER` on Windows, but it is not MSC.

```
% ./third-party/toolchains/llvm/8.0.0/bin/windows/clang.exe -march=corei7 -dM -E -x c empty | grep MSC
#define _MSC_BUILD 1
#define _MSC_EXTENSIONS 1
#define _MSC_FULL_VER 191225835
#define _MSC_VER 1912
```

We are not allowed to define `__SSE4_2__` in clang when clang assumes older arch, because in that case it refuses to compile SSE intrinsics, there is a couple of places in folly where we assume defined `__SSE4_2__` to use SSE4.2 intrinsics.

On the other hand clang defines `__SSE4_2__` itself when it is available:

```
% ./third-party/toolchains/llvm/8.0.0/bin/windows/clang.exe -march=corei7 -dM -E -x c empty | grep SSE
#define __SSE2_MATH__ 1
#define __SSE2__ 1
#define __SSE3__ 1
#define __SSE4_1__ 1
#define __SSE4_2__ 1
#define __SSE_MATH__ 1
#define __SSE__ 1
#define __SSSE3__ 1
```

This is an example clang error on Windows:

```
folly/container/detail/F14Table.h:1098:23: error: always_inline function '_mm_crc32_u64' requires target feature 'sse4.2', but would be inlined into function 'splitHash' that is compiled without support for 'sse4.2'
      std::size_t c = _mm_crc32_u64(0, hash);
```

Reviewed By: vener91

Differential Revision: D16038436

fbshipit-source-id: 8728a8d91cadb936c9e244795c9270943fe005ad

Summary: https://github.com/google/sanitizers/issues/970

Reviewed By: yfeldblum

Differential Revision: D16037542

fbshipit-source-id: 8c2feb85454ae582498d38a474e8d5913c9438b1

Summary: Allows futures::retrying to take a continuation that returns a SemiFuture, and in that case to return a SemiFuture as the result.

Reviewed By: andriigrynenko

Differential Revision: D15876742

fbshipit-source-id: da6f1614dd940b83ca048ecedf3d3645aa69b1f0

Summary:
Adding ability to get the current number of bytes used in the tx
(transmit/send) socket buffer and the rx (receive) socket buffer.

Reviewed By: yfeldblum

Differential Revision: D15966995

fbshipit-source-id: 29a51cd4caad265590baa45385712659fb3f3dac

Summary: ported the properties header from unifex and added some simple compile tests

Reviewed By: ericniebler

Differential Revision: D15144375

fbshipit-source-id: 580f8c8977abd070333779898f3ac47510932f87

Summary: `FiberManagerInternal.h` uses `FOLLY_TLS` and `FOLLY_EXPORT` defines which are defined in `folly/Portability.h`.

Reviewed By: vener91, Orvid

Differential Revision: D16020265

fbshipit-source-id: 7f80b397bd364cc53efb67568293da5c90fe97d9

Summary: Original commit changeset: bb2af75d24d6

Reviewed By: yfeldblum

Differential Revision: D15998293

fbshipit-source-id: 9a5ba878f7571bd86dc4c2b6aa72b8764f631ffc

Summary:
adds customizations for folly::Executor folly::EventBase and related types.
The customizations make folly::Executor::KeepAlive<T> model the pushmi Executor, TimeExecutor, and Strand concepts as appropriate

Reviewed By: ericniebler

Differential Revision: D15418629

fbshipit-source-id: d74a777ca743b5fb72664d9c1527837c3fbdb3ba

Reviewed By: snarkmaster

Differential Revision: D15730421

fbshipit-source-id: 51eaac2ad443aadbab3ba772e651287c6b330408

Summary: interval_repeat is an algorithms for single senders. it repeats the predecessor single at the start time and again on an interval relative to that start time. the last parameter is an optional function that allows the time used for the next schedule to be modified.

Reviewed By: ericniebler

Differential Revision: D15945618

fbshipit-source-id: d03cf079d870ceb65e9502c9ff0edcb87b6fd25e