Summary:
Cache locality information under /sys is dispersed across a
very large number of files. This is a problem for short-lived processes
due to direct overheads and lock contention in the kernel. This diff
switches to a heuristic strategy that infers the interference pattern from
/proc/cpuinfo instead of computing it exactly. This doesn't necessarily
produce exactly the correct cache hierarchy info, but it yields the
correct topological sort for machines that have only core-local and
socket-local cache locality.

Differential Revision: D16459331

fbshipit-source-id: a322c126d1a4775d015bfb81451dbc6ad6fcc0fd

Summary:
flake8 (in Python 2.7 mode) complains that `typing` is mentioned in type annotations but is not defined:

```
fbcode_builder/getdeps/buildopts.py:251:21: F821 undefined name 'typing'
    subst_mapping,  # type: typing.Mapping[str, str]
                    ^
fbcode_builder/getdeps/buildopts.py:253:5: F821 undefined name 'typing'
    # type: (...) -> typing.Optional[str]
    ^
2     F821 undefined name 'typing'
2
```

Import `typing` explicitly to silence this warning.

Because `typing` may be unavailable, import it conditionally. (Because it's only referenced in comments, failing to import `typing` should have no effect at run time.)

Reviewed By: snarkmaster

Differential Revision: D16435696

fbshipit-source-id: 78a4a7b07acc46aa998f02b54b1a6e52c1daafde

Summary:
We need to wrap coro_.resume() with a RequestContextScopeGuard so that
if the task changes the request context, we restore it after the coroutine
suspends itself.

Reviewed By: lewissbaker

Differential Revision: D16497375

fbshipit-source-id: 439a6f41b2294e56d2ec1b0013a6013524b5ae6a

Summary:
Allow builds for 64-bit ARM (MSVC) to use the same implementation of `__builtin_clzll` as ARM and x86 architectures.
Pull Request resolved: https://github.com/facebook/folly/pull/1191

Test Plan: Imported from GitHub, without a `Test Plan:` line.

Reviewed By: yfeldblum

Differential Revision: D16350573

Pulled By: Orvid

fbshipit-source-id: f25087b3a53de09767f505ee8c7d971a6d957189

Summary:
Note that we only provide the strong exception guarantee for single key insertion and emplacement when `std::is_nothrow_constructible<value_type>::value == true`. We could `static_assert()` that in `sorted_vector_set`, since it seems to hold for all contbuilds. But for `sorted_vector_map` there are quite a few contbuilds broken by that static assert. So I've chosen not to static assert in `sorted_vector_set` for consistency.

`insert()`, `emplace*()` and `erase()` take `const_iterator` in compliance with C++11. `erase()` has an overload that takes an `iterator` in C++17, but since we are backed by a vector I don't think that is necessary.

Add `emplace_hint()` variants.

Reviewed By: yfeldblum

Differential Revision: D16427231

fbshipit-source-id: 4bcf6fd01d9b1320548d12b152d1cef9291c2dd2

Summary:
While wrapping my head around the TDigest code, I made a few small
drive-by clean-ups.

Reviewed By: yfeldblum

Differential Revision: D16478284

fbshipit-source-id: aaf289a57ef633f0f1527263beffb7c0edf596d1

Summary:
The standard requires that operator== for associative containers
check key equality using operator==, which may be a refinement of the
container's key_eq.  (The requirement is expressed via equal_range and
std::is_permutation, but has the same effect.)  The straightforward way to
implement this results in duplicate key comparisons.  This diff adds a
containsEqualValue method that can avoid the second check, and uses
it for map and set operator==.

Reviewed By: yfeldblum

Differential Revision: D16455735

fbshipit-source-id: 8ccd0743f8c11bee5d91c065214cead96dd8b72c

Summary: Eden no longer depends on fboss/common/ 🌈

Reviewed By: simpkins

Differential Revision: D16062879

fbshipit-source-id: e6cf6eed408ca6e688ff2b9b532b15aa96902ac1

Summary: Add a dependency from the eden open source build to the fb303 open source build and switch EdenServiceHandler to BaseService.

Reviewed By: simpkins

Differential Revision: D15528156

fbshipit-source-id: 2ca5c31dd9fcc9bac43fd399b27f33b6f2c5ebfc

Summary: Given a cpp folly::coro::AsyncGenerator, a python async generator/iterator may be built and bridged to it. Users may use `async for ...` syntax to consume data from it.

Reviewed By: yfeldblum

Differential Revision: D16404979

fbshipit-source-id: dd2dfd52d00bd1a3e30e838359a27458744f6461

Summary:
Some people want to use getdeps with Python 2.7. This looks easy to do, so take a step toward Python 2.7 support by fixing getdeps' tests when run with Python 2.7.

For Python 3, this diff should not change behavior.

This diff should address https://github.com/facebook/bistro/issues/35.

Reviewed By: snarkmaster

Differential Revision: D16435667

fbshipit-source-id: f5c262b12995b609263341c4de26dac7f9b12b70

Summary:
This is just https://github.com/facebook/bistro/pull/37/, backported to `fbcode_builder`.

snarkmaster
Pull Request resolved: https://github.com/facebook/bistro/pull/37

Test Plan: Push and watch Travis

Reviewed By: simpkins

Differential Revision: D16453080

Pulled By: snarkmaster

fbshipit-source-id: a15eaead931f046c41e50f8e3b412ef68b172d65

Summary:
`dynamic::hash()` would copy every key-value pair in the object accumulator hash because of two bugs in the code:
1. The lambda took `auto` instead of `auto const&`
2. The hasher was `hash<pair<dynamic, dynamic>>` not `hash<pair<dynamic const, dynamic>>` meaning a conversion was needed.

These bugs together caused 2 copies for each sub-object. Since the copies are recursive, each object got copied 2*depth times.

Reviewed By: yfeldblum

Differential Revision: D16452213

fbshipit-source-id: 64a55e1640abb022c148183646e9f9720fd8482e

Summary:
This is a proposed fix for issue: https://github.com/facebook/folly/issues/1193
Using `folly::resizeWithoutInitialization` with libc++ and the address sanitizer causes false positive 'container overflow' failures.

This PR only addresses the problem for resizing `std::vector`. Presumably the version of `folly::resizeWithoutInitialization` for `std::string` has the same problem, but I have not verified this.

### Potential issues
This PR assumes that the `__sanitizer_annotate_contiguous_container` is available in libc++, and that the `std::vector` implementation has been annotated with it. The latter occurred in ~2014:
http://llvm.org/viewvc/llvm-project/libcxx/trunk/include/vector?r1=191987&r2=208319&pathrev=208319

If there are people using C++14 compilers with versions of libc++ predating the addition of the `std::vector<T>::__annotate_contiguous_container` member function then this fix will fail to compile.

I'm not aware of any sane way to detect the exact version of libc++ being used. Perhaps a suitable mitigation for those (few?) people with a C++14 compiler and pre-2015 libc++ implementation would be to provide a `FOLLY_` macro to disable the annotation.
Pull Request resolved: https://github.com/facebook/folly/pull/1194

Reviewed By: yfeldblum

Differential Revision: D16445890

Pulled By: nbronson

fbshipit-source-id: ef33bbd5cc12c71822985c5174bf8e47e6eb3757

Summary: [Folly] Extract type aliases in `LtHash` members to avoid unnatural line breaks and indentation.

Reviewed By: modocache

Differential Revision: D16453375

fbshipit-source-id: 8d68458ba3ba7bf79fef82986f4f170e3b74ddd6

Summary: [Folly] Apply `clang-format` to `folly/experimental/crypto/`.

Reviewed By: modocache

Differential Revision: D16453122

fbshipit-source-id: 5a24b20dba3ab472cd490afaaa24e7ab1c1295ec

Summary:
getdeps allowed common arguments to be specified either before or after
the subcommand name.  However, the arguments were only actually honored
if they were specified on the command line after the subcommand name.
Specifying any of these arguments before the subcommand would simply be
ignored, and the default value would be used.  This happened since the
subcommand argument's default value overwrote the value specified by the
user from the main command parser.

This fixes the behavior by using `argparse.SUPPRESS` so that argparse
won't set the value at all if it was not specified.  This allows the
value from the main parser to be used correctly if the argument was not
seen by the subcommand parser.

Reviewed By: chadaustin

Differential Revision: D16435358

fbshipit-source-id: f3893ef7a27fb018247f249f313d6438472e0140

Summary: F14NodeSet::insert() leaks memory when construction throws.

Reviewed By: yfeldblum, nbronson

Differential Revision: D16399317

fbshipit-source-id: 4c9385288dd212469dfbd10b3e990b3533dc66d3

Summary: eachToTuple: use to<fbstring>(delim) instead of delim

Reviewed By: yfeldblum

Differential Revision: D16419866

fbshipit-source-id: 7a57f41ea210c569284f6c619784adc95f03b26e

Reviewed By: yfeldblum

Differential Revision: D16426555

fbshipit-source-id: 0211c58b6919dc412b9cd7a029f0543d0060b40b

Differential Revision: D16368120

fbshipit-source-id: 0511e2f84b3bf8d1affbfbeecd05c04cb0269037

Summary: This file is included in `coro/TimedWait.h` and `coro/Collect.h`, so inclusion of those two headers is causing redefinition errors.

Reviewed By: yfeldblum

Differential Revision: D16414841

fbshipit-source-id: 103fa87a7c978d256e58eb67089b4cb984d362e4

Summary:
The /EHa flag tells MSVC to catch asynchronous (structured) exceptions
in `catch(...)` blocks. This is generally a bad idea, because
structured exceptions include page faults, stack overflow, division by
zero, and illegal instructions, at which point attempting to run
further arbitrary C++ code will have wonky effects in the best
case. It would probably also clobber any state required to debug the
original failure.

Reviewed By: Orvid

Differential Revision: D16390933

fbshipit-source-id: 2997daaf350680dc9d0c7da3cabab8749e8ecb82

Summary:
The documentation claims that the constructor only throws
`std::invalid_argument`, but it could throw a `ConversionError`.

Reviewed By: Orvid

Differential Revision: D16387303

fbshipit-source-id: 887a84f48a49acdc0516605f8204432bfd8ca800

Summary: Cut remaining explicit class template instantiations in futures.

Reviewed By: mengz0

Differential Revision: D16366146

fbshipit-source-id: e755188bd4378ca713b2d73b74def76f9ad837a7

Summary: It doesn't have any thrift specific code. Moving it to folly allows other folly Cython code use cFollyOptional.

Reviewed By: fried

Differential Revision: D16388752

fbshipit-source-id: 459c564571ba01fb8f22996227120ba631ffc05c

Summary:
If `path_search()` returns that CMake is not available, raise an Exception and
fail the build.  This makes the failure somewhat easier to to identify.
Without this the code would continue and would try to invoke `subprocess` with
a value of `None` in the command argument list.  This error is slightly harder
to debug, since it isn't clear which command argument or environment variable
is not a string.

Reviewed By: chadaustin

Differential Revision: D16354623

fbshipit-source-id: be972b02cb13bc70db0f867da70e5bf4c6cec46d

Summary:
The scratch path is used as part of the hash computation for each project.  We
need to make sure this path is always normalized to ensure that we compute the
hashes consistently.

Reviewed By: chadaustin

Differential Revision: D16354624

fbshipit-source-id: 39b5362620bdc247cd7e7f1333dac319b354dc6f

Summary:
getdeps currently ends up calling `path_search()` repeatedly searching for
various C/C++ compilers in $PATH.  It ends up doing this multiple times for
each dependency as it computes the project hashes.  This updates the
`path_search()` function to cache its results so that we don't keep performing
the same searches on the file system over and over again.

Reviewed By: chadaustin

Differential Revision: D16354625

fbshipit-source-id: 116293bd2f636632517d26436b2332e6c10624f1

Summary: Fix mem leak due to observer not being freed

Reviewed By: yfeldblum

Differential Revision: D16234386

fbshipit-source-id: 54ea9e904cf6cde0e24edf1511c2e3b46a58376f

Summary: This particular `std::min` is ambiguous on macOS.

Reviewed By: simpkins

Differential Revision: D16352637

fbshipit-source-id: d444b4b5f22c94b76da5463493efe433829bbba5

Summary:
1. Py_INCREF when holding py future object in bridge function. Py_DECREF in a scope guard.
2. fix typo("folly::coroTask" should be "folly::coro::Task")

Reviewed By: yfeldblum, andriigrynenko

Differential Revision: D16341866

fbshipit-source-id: 1d5b946596cd86c4a949e2c66f007a6140046506

Summary: eden.thrift includes fb303_core.thrift, so any cmake target that depends on eden.thrift should pull in fb303_core.thrift's include directories and libraries. Implement that machinery in CppThriftLibrary.cmake.

Reviewed By: wez

Differential Revision: D16062657

fbshipit-source-id: d5d962960e767a138a9b634a12aebccf72d6ef43

Summary: Add a manifest and CMakeLists.txt for building fb303 with cmake.

Reviewed By: wez

Differential Revision: D15480895

fbshipit-source-id: d47f6ef9b9383b79b31a6170c7d4c9e8337de4d8

Summary:
Deprecate Future::then(executor, callback) pending removal.

This form of Future::then is ambiguous, and does not yet implement the stronger
typing of thenValue and thenTry. It is also tempting to use instead of via,
where it is not obvious that it has the behaviour of wrapping a via call in a
push and pop of the current executor:
.pushCurrentExecutor().via(executor).then(callback).popCurrentExecutor().

With the addition of inline continuations, we can instead make the nesting
explicit at low cost by making it an inline continuation that launches an
asynchronous task on the passed executor.

Reviewed By: yfeldblum

Differential Revision: D16282826

fbshipit-source-id: a94c994cf02f43f7a2bd857237885b25591fa9aa

Summary:
Fix an error about source directories being included in the include
path in add_thrift_cpp2_library. I don't know why this was failing for
fb303 and not for other projects, but adding a generation expression
appears to fix the issue.

Reviewed By: strager

Differential Revision: D15725024

fbshipit-source-id: 6c02fed6c6703733cf9e0b130c0f90b70e3ea300

Summary: Adds a callback interface to facilitate adding elision support for folly::Singleton. Planned to add context-aware elision support which will use an adaptation method based on the interface's callsite information (ie, the type of the Singleton instance, T, and the type of the callback function, F).

Reviewed By: nbronson

Differential Revision: D15970597

fbshipit-source-id: 1010322c64d1b7900b4bef5b35059238efabb071

Reviewed By: yfeldblum, luciang

Differential Revision: D16164723

fbshipit-source-id: ed0c4b2dddb09c8d97a8aca9379e4ea9b107a92c

Reviewed By: andriigrynenko

Differential Revision: D14637661

fbshipit-source-id: 0f8852e30ff554e9f0125e2d918054e4bc1e5af5

Reviewed By: yfeldblum

Differential Revision: D16214195

fbshipit-source-id: 4841b0ba3188f19aba7a0e2c2ffd04d2f5f55220