Summary: It's timing out in open source test runs because each individual test was taking 10 seconds. They each now take 2.

Reviewed By: yfeldblum

Differential Revision: D7841938

fbshipit-source-id: eb81c67cc26e98fb0b4b0c736bea747e985f2fcf

Summary:
`{Future,SemiFuture}::poll()`:
- minor chg to impl from assignment to stateless ?:

Reviewed By: yfeldblum

Differential Revision: D7840466

fbshipit-source-id: 201dd7121b713324bb054b72c8ecdf6f4f163b97

Summary: SerialExecutor is always wrapping another Executor, so we shouldn't force users to coordinate their lifetimes.

Reviewed By: yfeldblum

Differential Revision: D7856146

fbshipit-source-id: ac7caaa0f181406dfb6b59d36ae4efe6d1343590

Summary:
This allows using KeepAlive token as a pointer to an specific Executor sub-type (e.g. SequencedExecutor).
getKeepAliveToken() has to be a free function to achieve this goal.

Reviewed By: yfeldblum

Differential Revision: D7847884

fbshipit-source-id: 95dc62433d30212ceb9b9b2fc7d6b38cb5425311

Summary:
This diff adds dynamic thread creation/destruction to ThreadPoolExecutor (mostly CPU).

1) Threads are lazily created when the first task is add()ed (or getEventBase is called), applies to both IO and CPU

and

2) after a timeout in LifoSem, threads are joined in CPUThreadPoolExecutor.

Most of the logic is in ThreadPoolExecutor.  A separate dynamic IOThreadPoolExecutor will be a follow on diff.

To keep the previous behavior, users can set minthread == maxthreads, or gflags --dynamic_cputhreadpoolexecutor=false

Reviewed By: magedm

Differential Revision: D7477390

fbshipit-source-id: 855a35380861111de33656e8fb1b681ae8cd2807

Summary: Fixes two tiny formatting issues with the folly::Function documentation.

Reviewed By: yfeldblum

Differential Revision: D7856507

fbshipit-source-id: dc82256a35f6c2b1eb66a868f419f2c800f1ddc5

Summary: [Folly] Avoid duplicating throw_exception instances taking string literal, which differ only in the length of the string literal.

Reviewed By: marshallcline

Differential Revision: D7853216

fbshipit-source-id: a4648952cc10661322e5e23c6317ba4b75c268fb

Summary:
This diff optimizes the operation of F14 set emplace when
the argument list happens to be a single reference to a key_type.
An intermediate key_type value is removed in that case, which avoids a
move in all cases and avoids a copy when the emplace finds that the key
is already present in the set.

Reviewed By: yfeldblum

Differential Revision: D7845720

fbshipit-source-id: 16a639f53993d9843ffe265edd58c0d74de0c1f7

Summary:
When used in a QuantileEstimator, some sliding window buckets can be sparse or all together empty. In this case, reserving memory in the constructor and in merge is wasteful. Don't reserve in the constructor (this is a complete waste) and call shrink_to_fit in merge.

It's worth noting that the benchmarks *are* shrinking the digest, because the digests are only 85 or so items big, so it is exercising this path.

Reviewed By: yfeldblum

Differential Revision: D7855242

fbshipit-source-id: 61bcbaaf17dd46f8b98d171af43516ceb8d84510

Summary: Included but not used.  Dropped this (expensive) include

Reviewed By: Orvid

Differential Revision: D7781674

fbshipit-source-id: 11180d34cda63ef91f70c64c84bf89346e5d1729

Summary:
- Promise move ctor: use folly::exchange()
- Promise move assign: make 33% faster, more eager to release memory, use folly::exchange()

Reviewed By: yfeldblum

Differential Revision: D7838311

fbshipit-source-id: 6d0ee8d3121ab703b3c54c72532f34d66bd4259c

Summary:
Fix CoreAndCallbackReference move-ctor so it doesn't call this->detach() (my bad; see D7723078).

Also (for clarity/pedantry/obviousness), explicitly delete the class' move-assignment op.

Reviewed By: yfeldblum

Differential Revision: D7848370

fbshipit-source-id: bc051120b3df0557ce9b01cbad8c00a57b8780d8

Summary: Only wake a single waiter.

Reviewed By: davidtgoldblatt

Differential Revision: D6595881

fbshipit-source-id: 9393618bb619cb22b75ca91b27291b784fc5097c

Summary: Re-run small locks bench.  Add server type to comment.

Reviewed By: yfeldblum

Differential Revision: D7830168

fbshipit-source-id: 4b60ec6845bb3518a98f4645e521e95599072433

Summary: [Folly] No need to optimize `MicroSpinLock::lock` in `folly/futures/` after {D7828158} (facebook/folly@6f9d8a145e1c37b2b47ff91491c647fe321de787).

Reviewed By: nbronson

Differential Revision: D7843192

fbshipit-source-id: b40bcc0abcfed561a141864e135d72fe33f4067d

Summary:
This implements a new AsyncioLoopController for FiberManager, wrapping AsyncioExecutor.
Thanks to recent simplication of LoopController, if I haven't missed anything, code is fairly simple, though not optimizer for thread-local.

fiber_manager.pxd|pyx allow to link the life-time of AsyncioLoopController/FiberManager to the asyncio event loop.
Code is similar to (and rely on) executor.pxd|pyd, which does the same for AsyncioExecutor.

fibers.h|pxd provide a helper function for C++/python callers.
Code is similar to futures.h|pyd, which does the same for folly Future.
This is the main reason to keep a separate function/callback parameters. Not necessary for fibers, but a generic callback implementation can work for both fibers/future, which is convenient.

There are a few lines that feel duplicated between future/fibers implementation, but it felt overkill to abstract further.

Reviewed By: andriigrynenko

Differential Revision: D7811432

fbshipit-source-id: af0a07d7554acbab1fac44b7a9a6a98340501ef4

Summary: [Folly] Remove deactivation, a rarely-used facility, from `Future`.

Reviewed By: marshallcline

Differential Revision: D7788059

fbshipit-source-id: 38db16fc3c2694b3cd772844fdafe71f6033a97c

Summary: Nothing breaks so....

Reviewed By: yfeldblum, siyengar

Differential Revision: D4377965

fbshipit-source-id: 53c42e7f6f3a71aedd4b9a089ffca9d3c8472029

Reviewed By: yfeldblum

Differential Revision: D7832437

fbshipit-source-id: 036a89693e247e0d6356980c3bf9db84d322f408

Summary: This abstraction is currently only used in functions executed on Python threads, so it should be unecessary.

Reviewed By: andriigrynenko

Differential Revision: D7828136

fbshipit-source-id: a9ea8b23ffc26505c576106c49e850e1894648d5

Summary: This should allow better differentation between compilers in situations where warnings need to be disabled for a specific compiler. I've reclaimed the GCC_DISABLE version to have it now refer to only GCC, rather than GCC+CLANG, and also added a version for clang only.

Reviewed By: yfeldblum

Differential Revision: D7834730

fbshipit-source-id: 7e0b076f5d9bc55870cfb7e67281190b4add82e6

Summary:
future/semifuture move-assignment operators call this->assign(). The latter's arg was passed as an lvalue-ref. Result: now that arg is an rvalue-ref.

Note: I adjusted the assign() calling convention to avoid a forwarding ref. It would have been okay but it's easier to avoid a forwarding-ref than to explain why it happens to always be an rvalue-ref.

Reviewed By: yfeldblum

Differential Revision: D7831539

fbshipit-source-id: cf1c66cd498f768905ac3d21e39f693d7a652b39

Summary:
This diff adds support for the F14 algorithm using the advanced
SIMD instructions (NEON) that are part of aarch64.

Reviewed By: yfeldblum

Differential Revision: D7765273

fbshipit-source-id: 866a8a3481ad60b8aadcfb39718d6a5e62bbe07c

Summary:
Forwarding refs may hold lvalue-refs as well as rvalue-refs, so generally should be forward'd not move'd (to avoid erroneously consuming an lvalue).

The `initial` param of `folly::reduce(first, last, initial, func)` (also used by `folly::reduce(container, initial, func)`) is a forwarding-ref but the code used `std::move(initial)`. Goal: change that to `std::forward<T>(initial)`.

Reviewed By: yfeldblum

Differential Revision: D7827721

fbshipit-source-id: 12d4ee0ae3cfb0de71269575289c7cd8246eb468

Summary: (self explanatory)

Reviewed By: yfeldblum

Differential Revision: D7825123

fbshipit-source-id: 60c99950b3af351536c0fd0ba306b5796572e1b6

Summary:
Make these const:
- {Future,SemiFuture}::hasValue()
- {Future,SemiFuture}::hasException()
- {Future,SemiFuture}::isActive()
- Promise::throwIfFulfilled()

Reviewed By: yfeldblum

Differential Revision: D7824879

fbshipit-source-id: 035b4bbcfd5b88825580e22b37b13e07158898f3

Summary:
Avoid swap() in move-ctor:
  - slightly faster
  - frees resources eagerly/sooner vs. later

Side benefit: added `noexcept` to a few appropriate methods.

Reviewed By: yfeldblum

Differential Revision: D7723078

fbshipit-source-id: f2d08b5ce094c42cb114deda7a6841052d1b785a

Summary:
Timekeeper.h had a weird circular dependence issue that would show up when
included from within the future implementation headers.  The stuff in
Timekeeper.h should ideally be within the future headers themselves

Reviewed By: yfeldblum, Orvid

Differential Revision: D7787275

fbshipit-source-id: a6309c0cc5edca4f5bd12d2651c828299681df54

Summary: Remove the Logging.h shim that was simply redirecting includes to GLog.h

Reviewed By: yfeldblum

Differential Revision: D7783879

fbshipit-source-id: 8d1c1d8501c6a2b067ef72471885fedb3e9f7b10

Reviewed By: yfeldblum

Differential Revision: D7819076

fbshipit-source-id: 1dae43c27b88ea1a78d2a54f556687e8c4bbc008

Summary: Add previous for EliasFano reader. Here the equation is: reader == reader.next().previous();

Reviewed By: ot

Differential Revision: D7497171

fbshipit-source-id: 33dc3316036b94a5da278c328322ea4c3322588e

Summary:
- Fix all cases of "segfault if consumed/moved-out"
  - In Future, SemiFuture, Promise.
  - Those now throw NoState instead.

Reviewed By: yfeldblum

Differential Revision: D7721921

fbshipit-source-id: 76fafcbd30e23d63150196d3d7e404f348b220f2

Summary: There isn't a good way to access ThreadPoolExecutors to add instrumentation, such as grabbing basic stats on all executors to better understand pool capacity. This provides a way to do that.

Reviewed By: yfeldblum, djwatson

Differential Revision: D7658144

fbshipit-source-id: 3e72acdccf25742b76d438d2895d586049a0cc17

Summary:
MicroSpinLock was doing a non-relaxed load before cas, which seems odd because typically we expect spin locks to be unlocked. This diff flips the order and assumes the lock is unlocked before doing a load-loop.

While the benchmark results for SmallLocksBenchmark are a little mixed, the results in a more real world use case in stats/DigestBuilderBenchmark show that this is a real winner.

Reviewed By: djwatson

Differential Revision: D7828158

fbshipit-source-id: e195b9739ce931a67a39ffb9715a228aef59f6bc

Summary: Some infrastructure sends `SIGQUIT` to stuck processes and we would like to be able to log stacktraces for these.

Reviewed By: yfeldblum

Differential Revision: D7819980

fbshipit-source-id: d99a1232ca259bd95aed82505751e6591af748ff

Summary:
Just re-running it on newer hardware:

32 processors
model name      : Intel(R) Xeon(R) CPU E5-2660 0 @ 2.20GHz

Reviewed By: yfeldblum

Differential Revision: D7828138

fbshipit-source-id: 30a25c6bf7bd0294f32f7598c701b49f045ab235

Summary: Small change so that a template class `ExtractFirstArg` can do the same thing without relying on an expensive header, `boost/type_traits.hpp`.

Reviewed By: yfeldblum

Differential Revision: D7781675

fbshipit-source-id: 1e5f431217da6cdd31ba1d77a8fc2b19a43808aa

Summary: In a test build `<iterator/iterator_facade>` showed up as the most heavyweight include.  Remove this include, dependency on the boost lib, and implement whatever's needed for the bidirectional iterator interface.

Reviewed By: yfeldblum

Differential Revision: D7781659

fbshipit-source-id: 380e0ae85b18c245845f479b2d268fdd3e16d54a

Summary:
Promote the folly logging code out of the experimental subdirectory.
We have been using this for several months in a few projects and are pretty
happy with it so far.

After moving it out of the experimental/ subdirectory I plan to update
folly::Init() to automatically support configuring it via a `--logging` command
line flag (similar to the initialization it already does today for glog).

Reviewed By: yfeldblum, chadaustin

Differential Revision: D7755455

fbshipit-source-id: 052db34c97f7516728f7cbb1a5ad959def2f6efb

Summary: As title. We were special casing q == 1 or q == 0 improperly before. This led to some strange results where p100 < p99.9. Track min and max (useful anyway), and use those for boundaries.

Reviewed By: anakryiko

Differential Revision: D7800645

fbshipit-source-id: 5fa3e1d0d286d4b31442196ba74b072030706786