folly.git
9 years agoBump version to 25:0 v0.25.0
Sara Golemon [Wed, 11 Feb 2015 02:02:15 +0000 (18:02 -0800)]
Bump version to 25:0

9 years agoAutodetect pthread_atfork() support
Sara Golemon [Wed, 11 Feb 2015 02:00:05 +0000 (18:00 -0800)]
Autodetect pthread_atfork() support

Summary: This is usually available on linux

Test Plan: Existing tests and manual test on arclinux

Reviewed By: mpawlowski@fb.com

Subscribers: folly-diffs@, yfeldblum

FB internal diff: D1840081

Signature: t1:1840081:1423619986:911c95f9f04f286e9f113543ab4b705cf1b17396

9 years agoremove stray / in include path
Ranjeeth Dasineni [Fri, 6 Feb 2015 06:15:05 +0000 (22:15 -0800)]
remove stray / in include path

Summary: removing // as it could break some comment parser

Test Plan: unitz

Reviewed By: dancol@fb.com

Subscribers: trunkagent, doug, seanc, folly-diffs@, yfeldblum

FB internal diff: D1830591

Signature: t1:1830591:1423203436:9e8b28b714eb2dc080462ab3e8ce04c176df8586

9 years agoswitch order of method/object in Future::then to match std::bind
Hans Fugal [Tue, 10 Feb 2015 20:32:04 +0000 (12:32 -0800)]
switch order of method/object in Future::then to match std::bind

Summary:
I have half a mind to just rip this out and let people use std::bind if they need this. But I won't be so cruel.

Why isn't this just implemented as `then(std::bind(method, object))` anyway? Is the template soup we have now faster?

Test Plan:
Fixed the unit tests to use the new format.
Will look to contbuild to catch all the things this might break (if anyone is using it at all?), and will fix them.

Reviewed By: hannesr@fb.com

Subscribers: trunkagent, exa, folly-diffs@, yfeldblum, jsedgwick, davejwatson

FB internal diff: D1831118

Signature: t1:1831118:1423243771:65db9a89daf14d8bd88331c503ba1ea7ab03b679

9 years agofutures::chain
Hans Fugal [Tue, 10 Feb 2015 20:24:07 +0000 (12:24 -0800)]
futures::chain

Summary:
`futures::chain<A,Z>(a, b, c, d, e, f, ..., z)` where `a` is a callback suitable for `Future<A>::then` and `z` is a callback suitable for `Future<Y>::then<Z>`.

This will be important, at least in the background, for the new `via`. It will probably also be useful to some user sometime.

I imagine this will be tweaked over time if people find magic ways to get it to deduce the types better. But this works and it's not *too* much trouble to specify A and Z.

Test Plan: unit

Reviewed By: davejwatson@fb.com

Subscribers: trunkagent, exa, folly-diffs@, yfeldblum, jsedgwick

FB internal diff: D1831073

Tasks: 6048744

Signature: t1:1831073:1423259292:711be0e047a2acb706fd74e529d5d5fb6abda566

9 years agofolly::Singleton leak diagnostics utils
Andrii Grynenko [Fri, 6 Feb 2015 03:41:31 +0000 (19:41 -0800)]
folly::Singleton leak diagnostics utils

Summary: This adds different types of messages depending on whether Singleton was depending on other Singleton or just leaked. It also adds destruction stack trace for such Singletons (if they were ever destroyed) to help debug such leaks/broken dependencies.

Test Plan: unit test

Reviewed By: chip@fb.com

Subscribers: trunkagent, folly-diffs@, yfeldblum

FB internal diff: D1830526

Signature: t1:1830526:1423266462:ba328b0da0bf4030b1c4f686d8f7b609fd20683c

9 years agoMaking each SingletonEntry a singleton
Andrii Grynenko [Wed, 4 Feb 2015 17:48:12 +0000 (09:48 -0800)]
Making each SingletonEntry a singleton

Summary:
Most of the singleton construction logic is moved to SingletonEntry, and each SingletonEntry is now also a singleton.
SingletonVault becomes only responsible for keeping singleton construction order (and potentially dependencies) and destoying them in correct order.
This also significantly improves perf of get() / get_weak() (not-fast)

This diff is based on D1823663.

Test Plan:
unit test

============================================================================
folly/experimental/test/SingletonTest.cpp       relative  time/iter  iters/s
============================================================================
NormalSingleton                                            333.35ps    3.00G
MeyersSingleton                                   99.99%   333.39ps    3.00G
FollySingletonSlow                                49.99%   666.84ps    1.50G
FollySingletonFast                                95.90%   347.61ps    2.88G
FollySingletonFastWeak                             2.22%    15.00ns   66.66M
============================================================================

Reviewed By: chip@fb.com

Subscribers: trunkagent, folly-diffs@, yfeldblum

FB internal diff: D1827390

Signature: t1:1827390:1423268514:da322d1dcaba54905d478b253f26dd76f890fb4e

9 years agomake decodeVarint accept StringPiece
Philip Pronin [Tue, 10 Feb 2015 01:54:29 +0000 (17:54 -0800)]
make decodeVarint accept StringPiece

Summary: `decodeVarint` now accepts all of `StringPiece`, `MutableStringPiece`, `ByteRange`, and `MutableByteRange`.

Test Plan: fbconfig -r folly unicorn/utils && fbmake runtests_opt -j32

Reviewed By: ott@fb.com, lucian@fb.com

Subscribers: trunkagent, chaoyc, search-fbcode-diffs@, unicorn-diffs@, folly-diffs@, yzhan, yfeldblum

FB internal diff: D1836805

Signature: t1:1836805:1423534085:cca5c3e83ad699e5d56e1d1e3394644ec3f94dab

9 years agotypo
Andre Pinto [Tue, 10 Feb 2015 03:01:47 +0000 (19:01 -0800)]
typo

Summary: Correct typo on docs.

Test Plan: N/A

Reviewed By: alikhtarov@fb.com

Subscribers: folly-diffs@, yfeldblum

FB internal diff: D1836264

Signature: t1:1836264:1423527402:4e5f263f446346f7145c33c671f0b9e694db15fd

9 years agofix open source build
Marcin Pawlowski [Mon, 9 Feb 2015 21:41:02 +0000 (13:41 -0800)]
fix open source build

Summary:
include was missing, causing build to fail on
Arch Linux with gcc 4.9.2 -- found this while resolving
other issues.

Test Plan: build folly on my personal linux and on dev server

Reviewed By: viswanath@fb.com

Subscribers: fugalh, folly-diffs@, jsedgwick, yfeldblum

FB internal diff: D1835143

Signature: t1:1835143:1423513128:5dcff69a82878f47b40512711ed254f239399c2c

9 years agoRevert "Revert "Using type-tags for test SingletonVaults""
Andrii Grynenko [Fri, 6 Feb 2015 23:48:19 +0000 (15:48 -0800)]
Revert "Revert "Using type-tags for test SingletonVaults""

Summary:
This reverts commit 4893c09795ad4b1187518b184ac4812079039988.

Fix unit test in D1823663.

Test Plan: fbconfig -r folly fbmake dbg

Reviewed By: alikhtarov@fb.com

Subscribers: folly-diffs@, yfeldblum

FB internal diff: D1832645

Signature: t1:1832645:1423267466:6012f1d7700d540c7290c29b01b33148cf91183c

9 years agoTLS cache for AccessSpreader
Nathan Bronson [Fri, 6 Feb 2015 23:16:53 +0000 (15:16 -0800)]
TLS cache for AccessSpreader

Summary:
Under Linux the process gtod doesn't contain the current cpu, so
__vdso_getcpu uses RDTSCP, which makes it cost about 20 nanos.  The gtod
_does_, however, contain a nanosecond time (for CLOCK_REALTIME_COARSE)
updated during context switches.  This diff adds a TLS cache that uses
__vdso_clock_gettime_ns(CLOCK_REALTIME_COARSE) to detect context switches.
The end result is that AccessSpreader goes from ~20 nanos to ~6.

Test Plan: unit tests

Reviewed By: davejwatson@fb.com

Subscribers: yfeldblum, trunkagent, folly-diffs@

FB internal diff: D1798922

Signature: t1:1798922:1423264298:32312a5e9bddb3b8aa630c146ef708164a6a4651

9 years agoRevert "Using type-tags for test SingletonVaults"
Nicholas Ormrod [Fri, 6 Feb 2015 21:34:15 +0000 (13:34 -0800)]
Revert "Using type-tags for test SingletonVaults"

Summary: This reverts commit 315cfed207895455ecd359d0c1b9d98f28ed0519.

Test Plan:
fbconfig -r folly && fbmake dbg

Reviewed By: robbert@fb.com

Subscribers: sdwilsh, folly-diffs@, yfeldblum

FB internal diff: D1831932

Signature: t1:1831932:1423258329:0962b939a93bbf1722a9e1c90090dcc024b63765

Blame Revision: D1823663

9 years agoDedupe the implementations of EventBase::runInEventBaseThreadAndWait.
Yedidya Feldblum [Fri, 6 Feb 2015 19:37:36 +0000 (11:37 -0800)]
Dedupe the implementations of EventBase::runInEventBaseThreadAndWait.

Summary: [Folly] Dedupe the implementations of EventBase::runInEventBaseThreadAndWait.

Test Plan:
Unit tests:
* `folly/io/async/test/EventBaseTest.cpp`

Reviewed By: subodh@fb.com

Subscribers: trunkagent, folly-diffs@, yfeldblum, dougw, brettp

FB internal diff: D1826291

Signature: t1:1826291:1423225534:42264d8dcc8adec6b90ac8a3d6ce1f4b98f29297

Blame Revision: D1810764D1823407

9 years agoUsing type-tags for test SingletonVaults
Andrii Grynenko [Wed, 4 Feb 2015 00:48:31 +0000 (16:48 -0800)]
Using type-tags for test SingletonVaults

Summary: This makes test SingletonVaults more similar to global SingletonVault. It also ensures that every SingletonVault is a singleton, which allows potential optimization of making each SingletonEntry a Singleton too, which itself will make get/get_weak as efficient as get_fast/get_weak_fast.

Test Plan: unit test

Reviewed By: chip@fb.com

Subscribers: folly-diffs@, yfeldblum

FB internal diff: D1823663

Signature: t1:1823663:1423188884:4ab2b47cfba270a66613f6f2458b6affc0ed96ad

9 years agomake wait() and friends return reference to self instead of new Future
Hans Fugal [Fri, 6 Feb 2015 00:20:51 +0000 (16:20 -0800)]
make wait() and friends return reference to self instead of new Future

Summary:
we still make the new Future, but assign it to ourselves. this avoids the following buggy pattern that people might expect to work

```
auto f = ...
f.wait();

// Careful. f.value() was moved out into the new Future, so you may have lost something
someOperationOn(f.value());

// Nope. We already set a callback internally in wait()
f.then(...);
```

Test Plan: unit

Reviewed By: davejwatson@fb.com

Subscribers: exa, yfeldblum, trunkagent, fbcode-common-diffs@, sammerat, cold-storage-diffs@, folly-diffs@, jsedgwick, aflock

FB internal diff: D1809040

Tasks: 6048284

Signature: t1:1809040:1422900812:1b416408eb5eaa71e88778c9c22ed8bfba087efe

9 years agoAdd StringKeyed(Unordered){Set,Map} to folly
Andrei Alexandrescu [Wed, 4 Feb 2015 21:09:20 +0000 (13:09 -0800)]
Add StringKeyed(Unordered){Set,Map} to folly

Summary: We're using StringKeyed* from common/datastruct to avoid unnecessary string creation whenever we're looking up string keys. C++14 does offer a solution, see e.g. http://stackoverflow.com/questions/10536788/avoiding-key-construction-for-stdmapfind. That is not supported by current compilers.

Test Plan: unittests

Reviewed By: pavlo@fb.com

Subscribers: trunkagent, net-systems@, folly-diffs@, yfeldblum

FB internal diff: D1825700

Signature: t1:1825700:1423086724:530550c3c80e33c80900f31c0ade05c66b22cbe8

9 years agoDisallow assignment to rvalue Range objects (and StringPiece in particular)
Andrei Alexandrescu [Wed, 4 Feb 2015 19:57:51 +0000 (11:57 -0800)]
Disallow assignment to rvalue Range objects (and StringPiece in particular)

Summary:
Before this diff the code below compiled and did absolutely nothing of interest:

StringPiece fun();
...
fun() = "hello";

i.e. assignment to an rvalue range was allowed. Such code is almost always, if not always, in error. This diff fixes that.

Test Plan: ran unittests

Reviewed By: ldbrandy@fb.com

Subscribers: mpawlowski, net-systems@, folly-diffs@, yfeldblum

FB internal diff: D1825360

Signature: t1:1825360:1423097817:fdaaf893cd1abbe71dc857a315df7c45cb6a0bd0

9 years agoSubprocess Process Group Improvements
Darren Mo [Thu, 5 Feb 2015 17:56:04 +0000 (09:56 -0800)]
Subprocess Process Group Improvements

Summary: Eliminated a potential race condition by removing unnecessary code.

Test Plan: Unit tests pass.

Reviewed By: lesha@fb.com

Subscribers: tudorb, trunkagent, folly-diffs@, yfeldblum

FB internal diff: D1824924

Signature: t1:1824924:1423102017:6bce388dccf9126d6ac3a1741d8300a51c5276ed

Blame Revision: D1492526

9 years agoAvoid EventBase depending on Baton (mutex/condvar variant).
Yedidya Feldblum [Wed, 4 Feb 2015 23:10:51 +0000 (15:10 -0800)]
Avoid EventBase depending on Baton (mutex/condvar variant).

Summary: [Folly] Avoid EventBase depending on Baton (mutex/condvar variant).

Test Plan:
Unit tests:
* folly/io/async/test/EventBaseTest.cpp

Reviewed By: subodh@fb.com

Subscribers: trunkagent, fma, folly-diffs@, yfeldblum, dougw, brettp

FB internal diff: D1823407

Signature: t1:1823407:1423088450:71ae1673ed8067103e2aaa1ac9b239eae9ebe9de

Blame Revision: D1810764

9 years agoFix comment
Hans Fugal [Wed, 4 Feb 2015 21:03:26 +0000 (13:03 -0800)]
Fix comment

Summary: It is true. Also, add a useful message about which thread this Future will complete in.

Test Plan: it should still compile

Reviewed By: davejwatson@fb.com

Subscribers: trunkagent, exa, folly-diffs@, jsedgwick, yfeldblum

FB internal diff: D1819849

Tasks: 57052866008465

Signature: t1:1819849:1423074012:79bf11c877eb2f061cb5d07f7ab45fa4cc4a894b

9 years agoFuture<T>::onTimeout(Duration, function<T()>, Timekeeper*=nullptr)
Hans Fugal [Wed, 4 Feb 2015 20:55:23 +0000 (12:55 -0800)]
Future<T>::onTimeout(Duration, function<T()>, Timekeeper*=nullptr)

Summary:
(or func returns Future<T>)

Invoke and respond to a timeout with a callback, rather than using `within` and adding `onError` or `then` or something.

Test Plan: new tests

Reviewed By: davejwatson@fb.com

Subscribers: jsedgwick, yfeldblum, trunkagent, fugalh, exa, folly-diffs@

FB internal diff: D1763174

Tasks: 4548494

Signature: t1:1763174:1423074062:05cec1dfb1110b31b599033949ebe0ee70dd0552

9 years agoFix another comment
Hans Fugal [Wed, 4 Feb 2015 20:54:10 +0000 (12:54 -0800)]
Fix another comment

Summary:
I don't know what this paragraph is even saying. It looks like
copypasta

Test Plan: still compiles or my uncle is a monkey

Reviewed By: davejwatson@fb.com

Subscribers: trunkagent, exa, folly-diffs@, jsedgwick, yfeldblum

FB internal diff: D1819905

Tasks: 5705286

Signature: t1:1819905:1423074021:a095fe2a17d5698678a527b60c9384625df869fd

9 years agoFixing find_first_of O(n) case
Tom Jackson [Wed, 4 Feb 2015 20:51:11 +0000 (12:51 -0800)]
Fixing find_first_of O(n) case

Summary:
The `memchr()`-based `find_first_of()` behaves extremely badly when it's used in a loop and the input string doesn't contain all the needles. This was discovered when a reasonable line-breaking routine tried to use it to break lines by a mix of '\r' and '\n'. The entire remainder of the file was scanned each time a line was read.

Before:

```
CountDelimsBase                    1.26s   794.86m
CountDelimsNoSSE         100.03%   1.26s   795.12m
CountDelimsStd         40501.63%   3.11ms  321.93
CountDelimsMemchr         98.31%   1.28s   781.41m
CountDelimsByteSet     23162.41%   5.43ms  184.11
```

After:

```
CountDelimsBase                   3.20ms  312.08 <-- Base impl no longer considers memchr
CountDelimsNoSSE         102.37%  3.13ms  319.47
CountDelimsStd           103.19%  3.11ms  322.05
CountDelimsMemchr          0.25%  1.27s   788.39m
CountDelimsByteSet        59.68%  5.37ms  186.27
```

Test Plan: Benchmarks

Reviewed By: njormrod@fb.com

Subscribers: folly-diffs@, yfeldblum

FB internal diff: D1823536

Signature: t1:1823536:1423081687:bb2ec8cdea477ee9b9c8d8ad2bfdecdc52657515

9 years agoRevert "fix configure script to check pthread_atfork(3) in the correct way"
Sara Golemon [Wed, 11 Feb 2015 01:54:30 +0000 (17:54 -0800)]
Revert "fix configure script to check pthread_atfork(3) in the correct way"

This reverts commit 462e13b135cfa772403fddc16105bf23ea78105a.

9 years agoMerge pull request #125 from 0mok/fix-configure-pthread-atfork
Marcin [Mon, 9 Feb 2015 20:02:49 +0000 (20:02 +0000)]
Merge pull request #125 from 0mok/fix-configure-pthread-atfork

fix configure script to check pthread_atfork(3) in the correct way

9 years agoMerge commit '64f2f2734ad853784bdd260bcf31e065c47c0741' into fix-configure-pthread...
Brad Kim [Mon, 9 Feb 2015 05:04:00 +0000 (14:04 +0900)]
Merge commit '64f2f2734ad853784bdd260bcf31e065c47c0741' into fix-configure-pthread-atfork

9 years agoBump version to 24:0 v0.24.0
Andrew Cox [Wed, 4 Feb 2015 21:04:02 +0000 (13:04 -0800)]
Bump version to 24:0

9 years agoFix folly::Singleton error message
Andrii Grynenko [Wed, 4 Feb 2015 20:01:39 +0000 (12:01 -0800)]
Fix folly::Singleton error message

Test Plan: fbmake runtests

Reviewed By: chip@fb.com, pavlo@fb.com

Subscribers: folly-diffs@, yfeldblum

FB internal diff: D1825369

Signature: t1:1825369:1423080185:8507643a3f8860bae6a2240e0ae28fbaf2885b12

9 years agoFuture constructor
Dave Watson [Wed, 4 Feb 2015 18:52:49 +0000 (10:52 -0800)]
Future constructor

Summary:
As a replacement for makeFuture().  The main advantage is for non-void futures, implicit conversion works pretty well.

See unittest for examples

Test Plan: fbconfig -r folly/futures; fbmake runtests

Reviewed By: hannesr@fb.com

Subscribers: yfeldblum, trunkagent, doug, folly-diffs@, jsedgwick

FB internal diff: D1806575

Signature: t1:1806575:1422465608:6099f791591b70ce1bcda439b49307b8f3187d89

9 years agoAdding demonstrative test of O(n^2) string split case
Tom Jackson [Wed, 4 Feb 2015 18:52:34 +0000 (10:52 -0800)]
Adding demonstrative test of O(n^2) string split case

Summary:
Even though the benchmark only keeps the first 10 lines, it gets slower the bigger the "file" is.

```
folly/gen/test/StringBenchmark.cpp              relative  time/iter  iters/s
----------------------------------------------------------------------------
Lines_Gen(1e3)                                               5.53us  180.94K
Lines_Gen(2e3)                                    66.43%     8.32us  120.21K
Lines_Gen(3e3)                                    48.26%    11.45us   87.33K
```

Test Plan: Benchmarks

Reviewed By: lesha@fb.com

Subscribers: folly-diffs@, yfeldblum

FB internal diff: D1823545

Tasks: 6155600

Signature: t1:1823545:1423023712:86fdb3dadbec44195e4b3596cf793cea80ae3a76

9 years agofix AsyncServerSocket::bind issue
Philip Pronin [Tue, 3 Feb 2015 07:53:46 +0000 (23:53 -0800)]
fix AsyncServerSocket::bind issue

Summary:
When closing sockets on retry, we should use `shutdownSocketSet_`
if it is set (so socket will properly be removed from the set).

Also I changed all `::close()` and `::shutdown()` calls to `*NoInt()`
version.

Test Plan:
fbconfig unicorn/test:basic_compression_test && fbmake opt -j32
while _bin/unicorn/test/test:basic_compression_test; do echo DONE; done

Reviewed By: mcduff@fb.com

Subscribers: trunkagent, folly-diffs@, yzhan, yfeldblum

FB internal diff: D1821104

Tasks: 47525796123510

Signature: t1:1821104:1423001050:5bc09ffdd6666c2884ea82dbd70831a56513cfc9

Blame Revision: D1795120

9 years agoprint better error message
Tianjiao Yin [Tue, 3 Feb 2015 22:03:18 +0000 (14:03 -0800)]
print better error message

Summary:
before:

E0203 12:32:34.548096 3354863 [admonitor-local] Singleton.cpp:72] Singleton of type N8facebook12configerator22ConfigeratorStaticDataE has a living reference at destroyInstances time; beware! Raw pointer is 0x7fd6ccc81000. It is very likely that some other singleton is holding a shared_ptr to it. Make dependencies between these singletons are properly defined.

after:

E0203 13:48:09.013022 3913115 Singleton.cpp:72] Singleton of type facebook::configerator::ConfigeratorStaticData has a living reference at destroyInstances time; beware! Raw pointer is 0x7f6f7dc4c000. It is very likely that some other singleton is holding a shared_ptr to it. Make dependencies between these singletons are properly defined.

Test Plan: run it

Reviewed By: chip@fb.com

Subscribers: folly-diffs@, yfeldblum

FB internal diff: D1822466

Signature: t1:1822466:1423000686:345f40fa706701476256a7157468521bc69166a0

9 years agokill asyncsslserversocket
Dave Watson [Mon, 2 Feb 2015 22:10:44 +0000 (14:10 -0800)]
kill asyncsslserversocket

Summary: This was only used in a unittest, every real use case uses AsyncServerSocket, and decides on its own to do SSL or not.

Test Plan: fbconfig -r thrift/lib/cpp/test; fbmake runtests

Reviewed By: alandau@fb.com

Subscribers: fugalh, trunkagent, doug, alandau, bmatheny, ssl-diffs@, mshneer, jsedgwick, folly-diffs@

FB internal diff: D1806807

Signature: t1:1806807:1422396209:02333c736e1ef10963e3fe0b4ed6ecf6122475bb

9 years agoRemove request context enable flag
Dave Watson [Tue, 27 Jan 2015 21:05:18 +0000 (13:05 -0800)]
Remove request context enable flag

Summary:
Originally meant as a kill switch.  It's turned on in everything except adfinder, so we can probably remove this flag after running a couple adfinder canaries

I have no idea who should review this in ads team, guessed and added some people

Test Plan: https://our.intern.facebook.com/intern/tupperware/canary/view/?experiment_id=32937

Reviewed By: haijunz@fb.com

Subscribers: trunkagent, doug, ruibalp, nli, alandau, bmatheny, wormhole-diffs@, mshneer, folly-diffs@

FB internal diff: D1809255

Tasks: 5645329

Signature: t1:1809255:1422490068:5b4551e6c126d86cac8f8e7b06da6474e6b8a25a

9 years agoWait for some time if Singleton isn't destroyed immediately
Andrii Grynenko [Wed, 28 Jan 2015 04:09:27 +0000 (20:09 -0800)]
Wait for some time if Singleton isn't destroyed immediately

Summary: This diff introduces 5 seconds wait time to let other threads release Singleton which may be temporarily locked. This helps prevent most of "Singleton object alive after destruction" warnings in cases where weak_ptr API is used correctly. Abusive use of folly::Singletons, where dependencies between singletons are not properly defined will still cause a warning.

Test Plan: unit test

Reviewed By: chip@fb.com

Subscribers: trunkagent, folly-diffs@

FB internal diff: D1808371

Signature: t1:1808371:1422487261:573eb40b6a260e428d96be476659335250c7ea76

9 years agoSet CLOEXEC for connected fd
Haijun Zhu [Mon, 2 Feb 2015 19:41:38 +0000 (11:41 -0800)]
Set CLOEXEC for connected fd

Summary:
AsyncSocket sets CLOEXEC if it connects an address, but won't
if it uses a connected fd. This sets CLOEXEC for such fd.

Test Plan: fbconfig folly/io/async/test && fbmake runtests

Reviewed By: davejwatson@fb.com

Subscribers: trunkagent, folly-diffs@, zacm

FB internal diff: D1809300

Signature: t1:1809300:1422900997:40fcc84506582ac67076fd975a36d094522c35c2

9 years agoEventBase::runInEventLoopThreadAndWait.
Yedidya Feldblum [Mon, 2 Feb 2015 19:37:23 +0000 (11:37 -0800)]
EventBase::runInEventLoopThreadAndWait.

Summary:
[Folly] EventBase::runInEventLoopThreadAndWait.

Useful for when some code needs to be run in the event loop thread, but another thread needs to trigger the code and then wait for it to be done.

Test Plan:
Unit tests:
* `folly/io/async/test/EventBaseTest.cpp`

Reviewed By: davejwatson@fb.com

Subscribers: trunkagent, folly-diffs@, brettp, dougw

FB internal diff: D1810764

Signature: t1:1810764:1422900654:7ff0aa7feb2792266f620b344cf8a1110a09f7ef

9 years agofix configure script to check pthread_atfork(3) in the correct way
Brad Kim [Wed, 4 Feb 2015 15:27:55 +0000 (00:27 +0900)]
fix configure script to check pthread_atfork(3) in the correct way

9 years agoBump version to 23:0 v0.23.0
woo [Mon, 2 Feb 2015 21:15:32 +0000 (13:15 -0800)]
Bump version to 23:0

9 years agofixing readme and configure script links to double-conv
Marcin Pawlowski [Wed, 28 Jan 2015 00:01:46 +0000 (16:01 -0800)]
fixing readme and configure script links to double-conv

Summary:
Fixing following the issues raised by nsuke on github
following his diff (https://github.com/facebook/folly/pull/112)

Test Plan: visual inspection, tested build on arch linux

Reviewed By: davejwatson@fb.com

Subscribers: folly-diffs@

FB internal diff: D1807378

Tasks: 5909189

Signature: t1:1807378:1422403093:ca024f396a27ecd9b304c57643f0107764a38ca1

9 years agoadd some missing methods to DeterministicAtomic
Nathan Bronson [Mon, 2 Feb 2015 16:46:18 +0000 (08:46 -0800)]
add some missing methods to DeterministicAtomic

Summary:
fetch_and and fetch_or are actually required in some instances,
since the operation they perform isn't reversible.  This also adds some
other missing functions, ^= and fetch_{add,sub,xor}.

Test Plan: use from other code's unit tests

Reviewed By: mssarang@fb.com

Subscribers: yfeldblum, folly-diffs@

FB internal diff: D1798924

Signature: t1:1798924:1422892374:a9859039075ddcac54f75259b9bc6a29bb963a09

9 years agoRevert "[folly::gen] Making 'just()' reference arguments like 'from()'"
Chip Turner [Sat, 31 Jan 2015 03:51:20 +0000 (19:51 -0800)]
Revert "[folly::gen] Making 'just()' reference arguments like 'from()'"

Summary: This reverts commit 90da92b05762f8032560d0d6a66237ab2772d7f2.

Test Plan: compiles

Reviewed By: tjackson@fb.com

Subscribers: lins, anca, folly-diffs@, yfeldblum

FB internal diff: D1817376

Tasks: 6118752

9 years agoRework the Future::Core state machine
Hans Fugal [Fri, 30 Jan 2015 23:23:24 +0000 (15:23 -0800)]
Rework the Future::Core state machine

Summary:
There was a race reading `callback_` in `maybeCallback` and setting `callback_` in `setCallback`. This diff reworks the state machine to make this unpossible. To avoid the explosion of states due to the cross-product of has-interrupt-handler/has-been-interrupted/etc. I introduce a separate lock for setting interrupt handler and interruption, since this is primarily orthogonal. Other attributes (active, for example) are still atomic variables, and while somewhat tied into the state machine logically (e.g. transitioning from Armed to Done only happens when active) they are mostly independent, keeping the state machine simple (and probably faster).

I think it may even be possible to do some things cheaper. In some states, we may not need to protect the writing of `callback_` and `result_`. But we'd need to enforce some ordering so I'm not going to try to tackle that. But that could be some speedup if we can do it cheaply.

Test Plan:
Builds and all existing tests pass.

Reviewed By: rockyliu4@fb.com

Subscribers: yfeldblum, stepan, trunkagent, exa, folly-diffs@, jsedgwick

FB internal diff: D1807854

Tasks: 6087856

Signature: t1:1807854:1422656713:25b62706cd7952b2dde06dab08074f8030db456b

9 years agocautionary comment in futexWake
Nathan Bronson [Thu, 29 Jan 2015 20:02:45 +0000 (12:02 -0800)]
cautionary comment in futexWake

Summary:
Add a description of the fault-after-unlock problem that can
occur when a synchronization object is used to guard its own
destruction.

Test Plan: comment change only

Reviewed By: alikhtarov@fb.com

Subscribers: trunkagent, folly-diffs@

FB internal diff: D1808446

Signature: t1:1808446:1422481173:83c0369e2277cd5b8cf6377fcd92b132019611a7

9 years agoMaking 'just()' reference arguments like 'from()'
Tom Jackson [Thu, 29 Jan 2015 18:56:30 +0000 (10:56 -0800)]
Making 'just()' reference arguments like 'from()'

Summary: That is, for all inputs except r-values.

Test Plan:
Run tests
@override-test-failures

Reviewed By: ajaymenon@fb.com

Subscribers: trunkagent, folly-diffs@

FB internal diff: D1807789

Signature: t1:1807789:1422410689:442d50ab17f8e3b08e34f33318a4dc7f23b4c6cc

9 years agoenable ssl false start with Next Protocol Negotiation (NPN) extension
Shijin Kong [Thu, 29 Jan 2015 18:54:53 +0000 (10:54 -0800)]
enable ssl false start with Next Protocol Negotiation (NPN) extension

Summary:
This speeds up TLS handshake and might be a factor of liger perf regression.
The enabling is guarded by an #ifdef. The condition itself is defined in an
openssl patch.

Test Plan: folly unit tests pass. Tried on devices as well and from tcpdump data was sent before the new ticket was received, essentially speeded up the handshake process.

Reviewed By: subodh@fb.com

Subscribers: trunkagent, kmdent, seanc, benyluo, ssl-diffs@, ranjeeth, folly-diffs@

FB internal diff: D1806856

Tasks: 5284979

Signature: t1:1806856:1422494521:0a048ea9001da13b5d698b5a764d1e66dcbedc99

9 years agoindirect(), for making pointers out of refs
Tom Jackson [Thu, 29 Jan 2015 18:22:59 +0000 (10:22 -0800)]
indirect(), for making pointers out of refs

Test Plan: Unit tests

Reviewed By: ajaymenon@fb.com

Subscribers: trunkagent, folly-diffs@

FB internal diff: D1808023

Signature: t1:1808023:1422410709:8db2d73d5b4c0c2eab563643e5fa1557ebfd4730

9 years agoCleanup duplicate include
Sean Cannella [Wed, 28 Jan 2015 23:28:58 +0000 (15:28 -0800)]
Cleanup duplicate include

Summary:
Included climits and limits.h due to non-conflicting changes,
clean this up.

Test Plan: compiled

Reviewed By: kelarini@fb.com

Subscribers: trunkagent, folly-diffs@

FB internal diff: D1810133

Signature: t1:1810133:1422483209:0826d2198769efed86607de6c5b28ce0bc4ac04b

9 years agofix deprecated Promise::setException()
James Sedgwick [Wed, 28 Jan 2015 21:55:45 +0000 (13:55 -0800)]
fix deprecated Promise::setException()

Summary: should have been using fulfilTry anyways. and fulfilTry needs to take the Try by value so fix that too

Test Plan: unit

Reviewed By: hans@fb.com

Subscribers: fugalh, folly-diffs@, jsedgwick

FB internal diff: D1808923

Tasks: 6098987

Signature: t1:1808923:1422478981:d4b9727394339c996ebccb7841b94e0c7b2bffb4

9 years agoFixing break in multifeed. CHAR_BIT was previously unfound
Kyle Dent [Wed, 28 Jan 2015 21:47:43 +0000 (13:47 -0800)]
Fixing break in multifeed. CHAR_BIT was previously unfound

Summary: Now CHAR_BIT will be found

Test Plan: ran multifeed and folly

Reviewed By: seanc@fb.com

Subscribers: seanc, folly-diffs@

FB internal diff: D1810027

Signature: t1:1810027:1422480946:4568c5190e318ff097d33b3d0434593b72f8205e

9 years agofix missing header in Range.h
Khalid El-Arini [Wed, 28 Jan 2015 21:47:44 +0000 (13:47 -0800)]
fix missing header in Range.h

Summary: ^

Test Plan: fbmake

Reviewed By: woo@fb.com

Subscribers: folly-diffs@

FB internal diff: D1810061

Signature: t1:1810061:1422481560:089c9186834f2bf87ac6d10b71d53784299b7992

Blame Revision: D1806632

9 years agoConvert TransportInfo SSL fields to shared_ptrs
Viswanath Sivakumar [Wed, 28 Jan 2015 20:08:36 +0000 (12:08 -0800)]
Convert TransportInfo SSL fields to shared_ptrs

Summary:
We do a lot of copying of TransportInfo in proxygen, and in most cases
the SSL structs don't change after connection establishment. We could
cut down on memory usage by sharing these huge strings. This is
especially true with SPDY where all streams belonging to a session could
share these fields.

Facebook:

Test Plan: Unit tests, will canary

Reviewed By: afrind@fb.com

Subscribers: fugalh, bmatheny, ssl-diffs@, folly-diffs@, jsedgwick, woo

FB internal diff: D1807557

Tasks: 5343753

Signature: t1:1807557:1422472932:53038345fca620632097586fb9e410bca8fe748d

9 years agoFixing a -Wshorten-64-32 issues in folly for liger.
Kyle Dent [Wed, 28 Jan 2015 19:33:49 +0000 (11:33 -0800)]
Fixing a -Wshorten-64-32 issues in folly for liger.

Summary: Normally I would use the folly::to<> function, but it would cause a circular dependency, so I just added a static cast

Test Plan: Ran on iOS and ran the folly tests

Reviewed By: seanc@fb.com

Subscribers: lucian, mpawlowski, marcelo, tudorb, aalexandre, seanc, folly-diffs@

FB internal diff: D1806632

Signature: t1:1806632:1422416646:b8104f18f90eb7457f2f358428f2bd800f8f1db5

9 years agokill a couple unnecessary rethrows
James Sedgwick [Wed, 28 Jan 2015 19:01:44 +0000 (11:01 -0800)]
kill a couple unnecessary rethrows

Summary: as above, there were less of these than I expected, nice

Test Plan: unit

Reviewed By: hans@fb.com

Subscribers: trunkagent, folly-diffs@, jsedgwick

FB internal diff: D1789332

Tasks: 5949939

Signature: t1:1789332:1421878033:d7c2979a77b51a5257b8bcd910ad9296ca1aa7e0

9 years agoFix Conv.h compilation on Android
Sean Cannella [Wed, 28 Jan 2015 16:47:51 +0000 (08:47 -0800)]
Fix Conv.h compilation on Android

Summary:
std::to_string doesn't exist on Android so don't use it.

Facebook: Did a sync to fbandroid and confirmed liger compiles with this fix.

Test Plan: existing tests

Reviewed By: ranjeeth@fb.com

Subscribers: trunkagent, folly-diffs@, shikong, kmdent, fma, pgriess

FB internal diff: D1808037

Signature: t1:1808037:1422410556:d78e0633a1554254b1a1f25bef49a4550a1817c6

9 years agoAnother stab at waitWithSemaphore -> Future<T>::wait()
James Sedgwick [Wed, 28 Jan 2015 16:05:30 +0000 (08:05 -0800)]
Another stab at waitWithSemaphore -> Future<T>::wait()

Summary:
See D1785572. Check out the new Future test and the commented portion of wait(Duration) for the fix

The test only fails a few times out of a hundred before the fix, but hasn't failed yet after

Test Plan: futures unit, wait for contbuild

Reviewed By: hans@fb.com

Subscribers: trunkagent, rushix, fbcode-common-diffs@, hero-diffs@, cold-storage-diffs@, adamsyta, zhuohuang, darshan, micha, folly-diffs@, lins, tingy, hannesr, jsedgwick

FB internal diff: D1803526

Tasks: 59400086059995

Signature: t1:1803526:1422309486:3613c59a708ecac312d241723828763feb2a57aa

9 years agoAdd writeFile function to folly
Andrei Alexandrescu [Wed, 28 Jan 2015 00:21:22 +0000 (16:21 -0800)]
Add writeFile function to folly

Summary: Reciprocal of readFile

Test Plan: fbmake runtests

Reviewed By: tudorb@fb.com

Subscribers: trunkagent, net-systems@, folly-diffs@

FB internal diff: D1807551

Signature: t1:1807551:1422410565:f5bda294deb788da9f3881c19bb20cfa1f588c09

9 years agofixing readme and configure script links to double-conv
Marcin Pawlowski [Wed, 28 Jan 2015 00:01:46 +0000 (16:01 -0800)]
fixing readme and configure script links to double-conv

Summary:
Fixing following the issues raised by nsuke on github
following his diff (https://github.com/facebook/folly/pull/112)

Test Plan: visual inspection, tested build on arch linux

Reviewed By: davejwatson@fb.com

Subscribers: folly-diffs@

FB internal diff: D1807378

Tasks: 5909189

Signature: t1:1807378:1422403093:ca024f396a27ecd9b304c57643f0107764a38ca1

9 years agomake getFD() a const method on AsyncUDPServerSocket
Claudiu Gheorghe [Mon, 26 Jan 2015 22:59:46 +0000 (14:59 -0800)]
make getFD() a const method on AsyncUDPServerSocket

Summary: no need to be mutable, since it only calls AsyncUDPSocket::getFD() which is a const method

Test Plan: just compiled -- simple change

Reviewed By: afrind@fb.com

Subscribers: trunkagent, doug, folly-diffs@

FB internal diff: D1802765

Signature: t1:1802765:1422307972:eab6ab4b16407df6e5034cef78c79f45cd9fa46f

9 years agofix folly build under python 3
Marcin Pawlowski [Mon, 26 Jan 2015 21:57:21 +0000 (13:57 -0800)]
fix folly build under python 3

Summary:
Make python scripts compatible woth both python 3 and python 2.
Added ignore rule for test/glog directory.

Facebook:
While verifying some issues reported on github i noticed that
folly does not build when python 3 is the default (at least on arch
linux). As it turns out it is fairly easy to make our python files
compatible with both python 2 and python 3.
I also added test/glog*/* to gitignore so that it is easier to work
on folly outside of facebook.

Test Plan:
unit tests in fb envirnoment and unit test on arch linux
(python 3.4.2, gcc 4.9.2)

Reviewed By: njormrod@fb.com

Subscribers: trunkagent, folly-diffs@

FB internal diff: D1801368

Signature: t1:1801368:1422294777:4e7b6b5634e8dccd849194e171c9a342bd1cb8b8

9 years agofolly: to: make exceptions more informative
Lucian Grijincu [Mon, 26 Jan 2015 21:43:36 +0000 (13:43 -0800)]
folly: to: make exceptions more informative

Summary: print the value which failed conversion to aid debugging.

Test Plan: run code which throws in to<> and contbuild

Reviewed By: tudorb@fb.com, andrei.alexandrescu@fb.com

Subscribers: trunkagent, folly-diffs@

FB internal diff: D1786294

Signature: t1:1786294:1421795912:00cc10923990e5aac0a15eedec09deb8e8d470d1

9 years ago(Wangle) Then Magic
Hannes Roth [Mon, 26 Jan 2015 20:53:42 +0000 (12:53 -0800)]
(Wangle) Then Magic

Summary: All is good now.

Test Plan: Run all the tests.

Reviewed By: hans@fb.com

Subscribers: jsedgwick, trunkagent, fugalh, folly-diffs@

FB internal diff: D1758272

Signature: t1:1758272:1421889405:a015d30783715e106a1e6667e971f7cf47c8392d

9 years ago(Wangle) Don't add an extra value() call for makeFuture(Try)
Hannes Roth [Mon, 26 Jan 2015 17:08:49 +0000 (09:08 -0800)]
(Wangle) Don't add an extra value() call for makeFuture(Try)

Summary: Not sure why we'd need that? This avoids a move, I think.

Test Plan:
Run all the tests. Also unbreaks this code, which used to work at some
point.
https://phabricator.fb.com/diffusion/FBCODE/browse/master/taoThriftService/TaoServiceHandler.cpp;a27bc2ef36cd671d65ae0fd2cc1fb1f823e68ae4$246

Reviewed By: hans@fb.com

Subscribers: trunkagent, folly-diffs@, jsedgwick

FB internal diff: D1800619

Signature: t1:1800619:1422053903:f9e65a0be3d820a9a9989b3cf5dfaf2a61e2e900

9 years agofix memory leak in AsyncServerSocket
Mark McDuff [Sat, 24 Jan 2015 09:23:20 +0000 (01:23 -0800)]
fix memory leak in AsyncServerSocket

Test Plan: g-unittest

Reviewed By: philipp@fb.com

Subscribers: folly-diffs@

FB internal diff: D1801665

Signature: t1:1801665:1422093104:eecf86ea8c1ab39dcb3d01442ed66ae45ef3bede

9 years agomake AsyncServerSocket bind to same port on ipv4 and ipv6 with port=0
Mark McDuff [Thu, 22 Jan 2015 00:18:25 +0000 (16:18 -0800)]
make AsyncServerSocket bind to same port on ipv4 and ipv6 with port=0

Summary: I'm in unfamiliar territory, so shout if I'm doing something dumb.  Perhaps it's a bad assumption that if the ipv4 port is free that the ipv6 port is also free?

Test Plan: g-unittest

Reviewed By: davejwatson@fb.com

Subscribers: trunkagent, ps, bmatheny, folly-diffs@

FB internal diff: D1795120

Signature: t1:1795120:1422034693:bd315023ab6cd9e9bda12161d05dd781dc401546

9 years agospecify connection's idle tiemout
Woo Xie [Fri, 23 Jan 2015 18:21:38 +0000 (10:21 -0800)]
specify connection's idle tiemout

Summary: Now all connections are always scheduled with connectionManager's default timeout.  This diff enables us to specify the different timeouts for each managed connections.

Test Plan: prospect testing

Reviewed By: davejwatson@fb.com

Subscribers: trunkagent, fugalh, folly-diffs@, jsedgwick

FB internal diff: D1797193

Tasks: 5343753

Signature: t1:1797193:1422034092:ed67b96efe8af8f8d1355d3f86fb1289daafb178

9 years agoFinagle interfaces
Dave Watson [Fri, 23 Jan 2015 17:50:35 +0000 (09:50 -0800)]
Finagle interfaces

Summary: Future service interfaces similar to finagle.  Service creators for client, filters

Test Plan: Unittests included - also sets up a simple pipeline to test a full stack client/server.

Reviewed By: hans@fb.com

Subscribers: jsedgwick, trunkagent, njormrod, folly-diffs@, doug, fugalh

FB internal diff: D1573086

Tasks: 5002456

Signature: t1:1573086:1421970698:328453c4a980bb6950fc9aeed6a2b6d9819c20db

9 years agoRevert "[futures] waitWithSemaphore -> Future<T>::wait()"
Dave Watson [Thu, 22 Jan 2015 20:32:26 +0000 (12:32 -0800)]
Revert "[futures] waitWithSemaphore -> Future<T>::wait()"

Summary:
Subject:

This reverts commit d8eeb9a12f932b74d268c74ce7ce2610dc14b152.

Test Plan: Watch contbuild?

Reviewed By: njormrod@fb.com

Subscribers: trunkagent, doug, fbcode-common-diffs@, hero-diffs@, cold-storage-diffs@, adamsyta, zhuohuang, darshan, micha, folly-diffs@, lins, tingy, hannesr, jsedgwick

FB internal diff: D1797163

Tasks: 5940008

Signature: t1:1797163:1421959040:c9d528367fed2de2f7dafdac9416aa01bdb47df9

9 years agoFix ASAN failures in OBCClient
Andrii Grynenko [Thu, 22 Jan 2015 19:07:08 +0000 (11:07 -0800)]
Fix ASAN failures in OBCClient

Summary:
This also adds a dependency from all folly::Singletons on RequestContext. Better fix would probaly be to switch RequestContext to be managed by folly::Singleton, however that would require us to switch all singletons which use it to be folly::Singletons too.

Facebook:
Making sure we initialize SR singletons when OBCClient singleton is created, so that SR is destoyed after OBCClient.

Test Plan:
fbconfig -r --sanitize=address --platform-all=gcc-4.8.1-glibc-2.17 dragon/
fbmake runtests

Reviewed By: meyering@fb.com, davejwatson@fb.com

Subscribers: trunkagent, folly-diffs@, meyering

FB internal diff: D1796847

Tasks: 5986816

Signature: t1:1796847:1421954531:e9ec177d5289a33c12fe0947426d89963c8e262c

9 years agoAdd additional instantiations in Instantiations.cpp
Adam Simpkins [Fri, 16 Jan 2015 02:35:56 +0000 (18:35 -0800)]
Add additional instantiations in Instantiations.cpp

Summary:
Histogram::getPercentileBucketIdx() and getPercentileEstimate() are
implemented using template methods of HistogramBuckets.

This updates Instantiations.cpp to also instantiate the versions of
these template methods that are required by Histogram<int64_t>.  Without
this, anyone using these methods was required to also include
Histogram-defs.h, or otherwise they may get link errors.  This makes it
so they no longer need to include Histogram-defs.h

The C++ syntax required is pretty horrendous.  We could potentially
avoid this by refactoring the code so that these methods are no longer
templates (and require that the Bucket class provide methods for this
purpose).  At the moment adding instantiations seems simpler, and
doesn't change any public APIs.

Test Plan:
Successfully linked code using
Histogram<int64_t>::getPercentileEstimate() without including
Histogram-defs.h

Reviewed By: ldbrandy@fb.com, meyering@fb.com

Subscribers: trunkagent, doug, net-systems@, exa, folly-diffs@

FB internal diff: D1786346

Signature: t1:1786346:1421429734:da17b7ea326c64a1e158fec3b87dc49c3db3d848

9 years agoadd IOBuf::appendToIov()
Adam Simpkins [Thu, 15 Jan 2015 00:00:31 +0000 (16:00 -0800)]
add IOBuf::appendToIov()

Summary:
This splits out some code from the existing getIov() method, to support
appending to an existing iovec array, rather than always returning a new
one.

I have a use case where I have multiple IOBuf chains to write at once,
preceded by some header data.  appendToIov() makes it easier to populate
a single iovec array with this data, and then perform a single writev()
call.

Test Plan: Updated the unit tests to exercise appendToIov()

Reviewed By: jasmeetbagga@fb.com

Subscribers: trunkagent, doug, exa, net-systems@, folly-diffs@

FB internal diff: D1783090

Tasks:

Signature: t1:1783090:1421450476:d5c89d6e1adbd2f2239057238e020b302df09f23

Blame Revision:

9 years agogetVia() and waitVia()
James Sedgwick [Wed, 21 Jan 2015 21:39:03 +0000 (13:39 -0800)]
getVia() and waitVia()

Summary:
Introduce ProgressableExecutor, which is an Executor that can be driven somehow. Examples include EventBase and ManualExecutor
Then introduce Future<T>::getVia(ProgressableExecutor*) and Future<T>::waitVia(ProgressableExecutor*) that drive the given executor until the Future is complete, with the usual semantics of get and wait respectively
This is a really common pattern in tests and you can see in the various changes to other projects lends sopme nice redness and cleanliness

Some notes:
1. I don't like ProgressableExecutor::makeProgress() that much. Too verbose. Maybe DrivableExecutor::drive()? Something else? Thoughts?
2. I still need to integrate this with some stuff in Zeus (SessionFuture) and Zookeeper (ZookeeperFuture) but I'm going to do that in a separate diff because it's going to be a little more intrusive
3. These APIs take a raw ptr so that they are consistent with via()
4. See inline note on ManualExecutor
5. See inline note in added unit tests

Test Plan: add unit for new API, wait for contbuild

Reviewed By: hans@fb.com

Subscribers: trunkagent, dresende, pzq, tdimson, fbcode-common-diffs@, targeting-diff-backend@, alandau, apollo-diffs@, bmatheny, everstore-dev@, zhuohuang, laser-diffs@, mshneer, folly-diffs@, hannesr, jsedgwick

FB internal diff: D1789122

Tasks: 5940008

Signature: t1:1789122:1421868315:6ea2fc2702be1dc283c24a46d345fb5da3935b32

9 years agouse wait()/get() instead of while(!f.isReady()) { ... }
James Sedgwick [Wed, 21 Jan 2015 19:41:50 +0000 (11:41 -0800)]
use wait()/get() instead of while(!f.isReady()) { ... }

Summary: as above. nice redness. couple of comments inline though

Test Plan: contbuild

Reviewed By: hans@fb.com

Subscribers: trunkagent, zeus-diffs@, cold-storage-diffs@, targeting-diff-backend@, apollo-diffs@, micha, folly-diffs@, jsedgwick

FB internal diff: D1787776

Tasks: 5940008

Signature: t1:1787776:1421776843:13772348538a0af27ceb5a363b818ece0cdfffc6

9 years agowaitWithSemaphore -> Future<T>::wait()
James Sedgwick [Wed, 21 Jan 2015 19:35:42 +0000 (11:35 -0800)]
waitWithSemaphore -> Future<T>::wait()

Summary:
see task. also adjust all callsites to get() or wait() as appropriate. this is a lot better, especially in various tests

Facebook:

cc @rushix, @wez as this was quite nice for docstore and novak, and you might want to be aware of new conventions

Test Plan: futures unit, wait for contbuild

Reviewed By: hans@fb.com

Subscribers: trunkagent, fbcode-common-diffs@, hero-diffs@, cold-storage-diffs@, adamsyta, zhuohuang, darshan, micha, folly-diffs@, lins, tingy, hannesr, jsedgwick, wez, rushix

FB internal diff: D1785572

Tasks: 5940008

Signature: t1:1785572:1421866794:a879de4d0bc14e96c434f623ed2a74361e25f28c

9 years agocodecs
Dave Watson [Wed, 21 Jan 2015 16:12:33 +0000 (08:12 -0800)]
codecs

Summary:
Start of codec framework.  Copied the frame based codecs almost exactly from netty, but made to fit our pipeline management.

BytesToMessageCodec is slightly different: Netty preprocesses all the available data to a List<Message>, while this codec a) Only does one message at a time to avoid queueing issues, and b) doesn't template the message type, and just passes an IOBuf

I'm fighting the type system to get the pipelines to play nice to each other:  I'd rather template the message type, but it prevents stacking outgoing handlers on top of it, since you ahve to specify both the input/output type for each handler, even if you only care about the output type.  Suggestions to fix?   Netty gets around this by lots of dynamic casting to Object type, but we can't do that in C++ since we don't have a base object type

Test Plan: Includes lots of tests

Reviewed By: jsedgwick@fb.com

Subscribers: jsedgwick, doug, fugalh, folly-diffs@

FB internal diff: D1758189

Tasks: 50023615002316

Signature: t1:1758189:1421170225:6bc8cc6a0bb461a965665bc88f7009033b215da9

9 years agofolly: added C interface to SingletonVault
Bob Haarman [Tue, 20 Jan 2015 20:36:40 +0000 (12:36 -0800)]
folly: added C interface to SingletonVault

Test Plan:
fbconfig folly/experimental/test:singletonvault_c_test && fbmake runtests
buck test folly/experimental/test:singletonvault_c_test

Reviewed By: andrii@fb.com

Subscribers: trunkagent, sdwilsh, folly-diffs@

FB internal diff: D1780054

Signature: t1:1780054:1421187023:b18c1d5575b2864b3d3df001a0024b91eb321c46

9 years agoTry to fix IdleTime unit test
Haijun Zhu [Tue, 20 Jan 2015 18:54:59 +0000 (10:54 -0800)]
Try to fix IdleTime unit test

Summary:
IdleTime unit test has been failing randomly (#5996886). The
event base loop calculates an exponentially moving average of loop busy
time in each loop. In the test, the busy time is initialized to 5900 and
each loop runs more than 8000 us. In normal case decaying of the
previous loop busy time will take 6 loops to  move this average to >6000.
But if the test is running on a heavily loaded machine the busy+idle time
a loop is longer than it should be, causing the decaying happen faster.
This diff skips this test if this happens.

Test Plan:
It won't affect anything if running on my devserver. It only
helps on a heavily loaded contbuild host.

Reviewed By: alandau@fb.com

Subscribers: folly-diffs@

FB internal diff: D1788450

Tasks: 5996886

Signature: t1:1788450:1421460774:1ec575c50f881e10b5a0208717fe68164f0d0f57

9 years agoFix comment typo
Alexey Spiridonov [Sat, 17 Jan 2015 00:04:22 +0000 (16:04 -0800)]
Fix comment typo

Test Plan: checked this is the right function name

Reviewed By: ngbronson@fb.com

Subscribers: folly-diffs@

FB internal diff: D1788386

Signature: t1:1788386:1421453582:82d0c4685c1bd2d888f5974fd2a998ebbc27a9b5

9 years agofix and improve MPMCQueue comment
Nathan Bronson [Fri, 16 Jan 2015 23:44:39 +0000 (15:44 -0800)]
fix and improve MPMCQueue comment

Summary:
The big comment block at the top of MPMCQueue had a non-sense
sentence resulting from a merge conflict, and also didn't include
details about ordering or recommended patterns for handling consumer
shutdown.

Test Plan: comment change

Reviewed By: lesha@fb.com

Subscribers: folly-diffs@

FB internal diff: D1788230

Signature: t1:1788230:1421451643:7c768c2b127499e3f2b06ee241fba851b4ed2a29

9 years agoapply all sockopts to listening sockets
Satadru Pan [Fri, 16 Jan 2015 20:05:10 +0000 (12:05 -0800)]
apply all sockopts to listening sockets

Summary:
We want all socket options to be applied to the listen sockets too.

The old code here that only applied SO_KEEPALIVE is a bit superstitious.
Not all options will be inherited by accept()ed sockets, but that's
fine.  We set them again post-accept().

Applying all these options to the listening sockets will fix at least
one long standing bug, which is that QoS is not set correctly for the
first few packets in a flow.  This also will let us do the right thing
when we gain the ability to set TCP_MAXSEG in D1741753.

Test Plan: fbmake runtests & canary in prod

Reviewed By: afrind@fb.com, davejwatson@fb.com

Subscribers: jsedgwick, satadru, trunkagent, fugalh, exa, folly-diffs@, agartrell

FB internal diff: D1745182

Tasks: 2911597

Signature: t1:1745182:1418860157:045680a91b153482bcd4a014894fb28059955d06

9 years agofolly: Bits: restore type of bitOffset_ to shorter "unsigned int"
Jim Meyering [Sat, 10 Jan 2015 06:28:00 +0000 (22:28 -0800)]
folly: Bits: restore type of bitOffset_ to shorter "unsigned int"

Summary:
My initial change to avoid the -Wsign-compare error neglected
the fact that the difference of two unsigned types is also
unsigned. That broke some unicorn tests, as seen in D1776343,
which worked around the problem by changing the type of
bitOffset_ like this: s/unsigned int/size_t.

The real problem lay in the distance_to function, which computed
the signed difference of two bitOffset_ values.
To work properly with the narrower unsigned type,
we could either first widen each operand to ssize_t
before computing their parenthesized difference, or
(as Philip suggested and I prefer), just drop the
parentheses and let the signedness of the preceding
operands ensure that each successive intermediate
result is also signed.

Test Plan:
Choose one of the failing tests, and ensure it now passes:

fbconfig -r unicorn/diskindex4/test:doc_map_test
fbmake runtests

Also run all unicorn tests, for good measure (still waiting):
fbconfig -r unicorn && fbmake runtests

Reviewed By: philipp@fb.com, andrei.alexandrescu@fb.com

Subscribers: trunkagent, net-systems@, folly-diffs@

FB internal diff: D1776624

Tasks: 5941250

Signature: t1:1776624:1420872874:b1ea8a3f23f21269aa65cbe1647e86a8f622a710

Blame Revision: D1770613

9 years agoremove outer Try from whenAll/whenN/whenAny callbacks
James Sedgwick [Fri, 16 Jan 2015 14:37:21 +0000 (06:37 -0800)]
remove outer Try from whenAll/whenN/whenAny callbacks

Summary:
it's redundant and we shouldn't encourage it, so i'm changing all callsites
i changed docs as well

Test Plan: wait for contbuild

Reviewed By: hans@fb.com

Subscribers: trunkagent, fbcode-common-diffs@, hero-diffs@, cold-storage-diffs@, targeting-diff-backend@, adityab, everstore-dev@, zhuohuang, darshan, micha, folly-diffs@, wch, lins, tingy, jsedgwick

FB internal diff: D1784667

Tasks: 5936469

Signature: t1:1784667:1421364620:83169739320e5342d28a744e3689794f16108fea

9 years agokill remaining Futures-related Wangle references
James Sedgwick [Fri, 16 Jan 2015 14:33:13 +0000 (06:33 -0800)]
kill remaining Futures-related Wangle references

Summary: grep around for wangle, adjust where appropriate

Test Plan: contbuild

Reviewed By: hans@fb.com

Subscribers: trunkagent, fbcode-common-diffs@, ruibalp, zeus-diffs@, vikas, targeting-diff-backend@, alandau, mwa, jgehring, zhguo, jying, darshan, fuegen, mshneer, folly-diffs@, lins, maxwellsayles, jsedgwick

FB internal diff: D1777581

Tasks: 5960242

Signature: t1:1777581:1421361395:2ee8bfeca76dafe7376a217c66e0a4eaf3ef416a

9 years agoMPMCQueue.h: fix a bug introduced while accommodating -Wsign-compare
Jim Meyering [Thu, 15 Jan 2015 03:54:27 +0000 (19:54 -0800)]
MPMCQueue.h: fix a bug introduced while accommodating -Wsign-compare

Summary:
* folly/MPMCQueue.h (waitForTurn): My changes to accommodate
-Wsign-compare introduced a bug that could cause lots of wasted CPU.
(See @scottf's comments on the blamed patch.)
Subtraction involving an unsigned type will always yield an unsigned
result, and there was at least this one instance where the result
would sometimes have to be a small negative number.  Instead, we'd
get approximately UINT32_MAX.  The fix is to explicitly cast the
difference to a signed type.

Test Plan:
Run this and note there is no error:
(this ensures it remains -Wsign-compare-clean with gcc-4.9)

fbconfig --platform-all=gcc-4.9-glibc-2.20 tao/server && fbmake dbgo

Before this change, running the mpmc_queue_test test would get
timeout errors ~8 in 100 runs.  Confirm that with this patch, now
there is no type of failure. I.e., run these commands and ensure
the grep prints nothing:

fbconfig folly/test:mpmc_queue_test
for i in $(seq 100); do fbmake runtests_dbgo >& makerr-$i; done
grep -E '(TIMEOUT|FATAL|FAIL).* [1-9]' makerr-*

Reviewed By: hans@fb.com

Subscribers: trunkagent, scottf, folly-diffs@

FB internal diff: D1783672

Tasks: 5941250

Signature: t1:1783672:1421350260:d7d306e8cf3ed3ee636d06228e1f24a2c2796b54

Blame Revision: D1766722

9 years ago'fix' threadlocal fork test
Dave Watson [Thu, 15 Jan 2015 21:25:02 +0000 (13:25 -0800)]
'fix' threadlocal fork test

Summary:
configure seems to think ubuntu 14.04 doesn't have pthread_atfork, which seems wrong?! But this test seems to fail if we don't have it on some platform anyway.  Not sure the right fix, just disable it for now?

Test Plan: watch https://ci-builds.fb.com/job/folly/

Reviewed By: hans@fb.com

Subscribers: doug, folly-diffs@

FB internal diff: D1783046

Signature: t1:1783046:1421348575:d279877abcc597a04c285092a5e5e53406922d5a

Blame Revision: D1768669

9 years agofix the manual config and flipped condition in SpinLock.h
Ranjeeth Dasineni [Thu, 15 Jan 2015 01:06:40 +0000 (17:06 -0800)]
fix the manual config and flipped condition in SpinLock.h

Summary:
i) FOLLY_HAVE_INT128_T definition is duplicated and is getting set to 1 on android. deleted that.
ii) we dont have pthread_spinlock_t and pthread_atfork on android, the former is absent apple as well
iii) a condition got flipped in D1768669. __ANDROID__ implies !FOLLY_HAVE_PTHREAD_SPINLOCK_T. fixed that

Test Plan:
- fbconfig -r folly && fbmake
- ran folly build with buck as well
- synced to fb4a and it builds

Reviewed By: seanc@fb.com

Subscribers: doug, seanc, folly-diffs@

FB internal diff: D1783579

Signature: t1:1783579:1421350138:9301303f77784e7bdb851e3f643751078457b025

9 years agoglobal executors with weak_ptr semantics
James Sedgwick [Wed, 14 Jan 2015 15:08:50 +0000 (07:08 -0800)]
global executors with weak_ptr semantics

Summary:
unfortunately, can't use atomics now that a weak_ptr is stored instead of a raw ptr, hence the additional singleton locks
might want to make an overload of via() that takes shared_ptr to integrate more easily? or change via() to store a shared_ptr (and make it non-owning if a raw ptr is passed)

Test Plan: unit

Reviewed By: davejwatson@fb.com

Subscribers: jsedgwick, trunkagent, fugalh, folly-diffs@

FB internal diff: D1764359

Tasks: 5002442

Signature: t1:1764359:1420845340:349ea88091d7ca4ee386b54aec599647341fadd4

9 years agoAdd folly::sorted_vector_map::at()
Louis Kruger [Tue, 13 Jan 2015 20:03:43 +0000 (12:03 -0800)]
Add folly::sorted_vector_map::at()

Summary: http://www.cplusplus.com/reference/map/map/at/

Test Plan: unittest

Reviewed By: ldbrandy@fb.com

Subscribers: trunkagent, jkong, folly-diffs@

FB internal diff: D1773996

Signature: t1:1773996:1421176834:33c8b39f03bf085ca2dd45df113cba948d0f2dd0

9 years agoBump version to 22:0 v0.22.0
Viswanath Sivakumar [Tue, 13 Jan 2015 19:02:44 +0000 (11:02 -0800)]
Bump version to 22:0

9 years ago(folly) osx bootstrap script
Hans Fugal [Tue, 6 Jan 2015 00:21:50 +0000 (16:21 -0800)]
(folly) osx bootstrap script

Summary:
This is a bootstrap script for building folly on osx. I attempted to make it Just Work™ for anyone who uses Homebrew (http://brew.sh/) but even if it doesn't work out of the box it should be instructive for the motivated developer on OSX.

Facebook:
In #5134151 Wangle used to not build, because something was using a futex. But that seems to have been rectified (or the build problems are hiding in headers that aren't used in the libfolly build).

I'll also update https://our.intern.facebook.com/intern/wiki/Folly/BuildingOnMac to point people to this script instead.

Test Plan:
Tested on my laptop which is running Yosemite (10.10.1)
Build, then inspect `otool -L .libs/*dylib`

Reviewed By: davejwatson@fb.com

Subscribers: folly-diffs@, exa

FB internal diff: D1765248

Tasks: 5134151

Signature: t1:1765248:1420500905:bc5da97a0446f502cb2badc85c9004ff0a62027a

9 years agoAdd libgtest to Makefile to unbreak OSS build
Viswanath Sivakumar [Tue, 13 Jan 2015 18:57:03 +0000 (10:57 -0800)]
Add libgtest to Makefile to unbreak OSS build

Summary:
SpookyHashTest now needs libgtestmain to be linked
(https://phabricator.fb.com/D1775555).

https://ci-builds.fb.com/job/folly/1570/console

Test Plan:
Don't have Ubuntu setup, but this should fix it. Will kick off jenkins
build.

Reviewed By: andrewjcg@fb.com

Subscribers: folly-diffs@

FB internal diff: D1779604

Signature: t1:1779604:1421174608:de865529d728e943288b52c0d63ebb712eda094f

Blame Revision: D1775555

9 years agomove Futures from folly::wangle to folly
James Sedgwick [Mon, 12 Jan 2015 14:48:32 +0000 (06:48 -0800)]
move Futures from folly::wangle to folly

Summary: Namespace change. Next up is to extract the Executors into folly/executors/

Test Plan: wait for contbuild

Reviewed By: davejwatson@fb.com

Subscribers: jsedgwick, trunkagent, fbcode-common-diffs@, chaoyc, search-fbcode-diffs@, lars, davejwatson, ruibalp, hero-diffs@, zeus-diffs@, andrewcox, vikas, mcduff, cold-storage-diffs@, unicorn-diffs@, ldbrandy, targeting-diff-backend@, netego-diffs@, abirchall, fugalh, adamsyta, atlas2-eng@, chenzhimin, mpawlowski, alandau, bmatheny, adityab, everstore-dev@, zhuohuang, mwa, jgehring, prometheus-diffs@, smarlow, akr, bnitka, jcoens, benj, laser-diffs@, zhguo, jying, darshan, micha, apodsiadlo, alikhtarov, fuegen, dzhulgakov, jeremyfein, mshneer, folly-diffs@, wch, lins, tingy, hannesr, maxwellsayles

FB internal diff: D1772779

Tasks: 5960242

Signature: t1:1772779:1420751149:bc0b9220be25f1d46d9cef3fdeaa9c3681217aff

9 years agofolly: make remaining TARGETS files auto converitble to BUCK files
Andrew Gallagher [Sat, 10 Jan 2015 03:25:31 +0000 (19:25 -0800)]
folly: make remaining TARGETS files auto converitble to BUCK files

Summary: Also add the `@buckify` tag to keep these automatically synced.

Test Plan: built and tested with buck and fbconfig/fbmake

Reviewed By: robbert@fb.com

Subscribers: trunkagent, sdwilsh, fugalh, folly-diffs@, jsedgwick

FB internal diff: D1775555

Signature: t1:1775555:1420840948:e9a2e86379be71b7d953fa53a727e51d2541b011

9 years agofolly: Bits: make bitsPerBlock 64 bit back
Lucian Grijincu [Sat, 10 Jan 2015 01:18:36 +0000 (17:18 -0800)]
folly: Bits: make bitsPerBlock 64 bit back

Summary: D1770613 made bitsPerBlock 32 bit. Sadpanda.

Test Plan: _bin/unicorn/diskindex4/test/doc_map_test --gtest_filter=DocMap/DocMapTest.Deleted/23

Reviewed By: meyering@fb.com

Subscribers: net-systems@, folly-diffs@

FB internal diff: D1776343

Tasks: 5968993

Signature: t1:1776343:1420852923:f35166dd9023a701ff16467710131f821810215a

Blame Revision: D1770613

9 years agofolly/dynamic-inl.h: avoid -Wsign-compare error
Jim Meyering [Wed, 7 Jan 2015 16:16:07 +0000 (08:16 -0800)]
folly/dynamic-inl.h: avoid -Wsign-compare error

Summary:
* folly/dynamic-inl.h (format): Add an explicit
int-to-size_t cast (that is ok because we've just ensured
it is non-negative) to avoid this error from gcc-4.9:

folly/dynamic-inl.h:953:29: error: comparison between signed and unsigned integer expressions [-Werror=sign-compare]

Test Plan:
Run this and note there are fewer errors than before:
fbconfig --platform-all=gcc-4.9-glibc-2.20 -r folly && fbmake dbgo

Reviewed By: delong.j@fb.com

Subscribers: trunkagent, folly-diffs@

FB internal diff: D1773242

Tasks: 5941250

Signature: t1:1773242:1420822301:b1e3c065cd25728c77957ce9e1f52e884ba8693b

9 years agofolly: ProducerConsumerQueue: avoid many -Wsign-compare errors
Jim Meyering [Wed, 7 Jan 2015 21:45:19 +0000 (13:45 -0800)]
folly: ProducerConsumerQueue: avoid many -Wsign-compare errors

Summary:
There were many templates instantiated with signed types.
Changing them to unsigned avoids the compilation failures
noted below.  However, we'll have to watch arc-diff-spawned
tests for new failures, just in case.

This is by far the largest undecomposable patch of those
I've written for this task so far.  It is possible that
some subset of these changes accomplishes the same goal,
but once I managed to get a successful compilation, I have
only re-reviewed it once and did not try to shrink it.

Test Plan:
Without these changes, we would get the following 32 errors from gcc-4.9:
https://phabricator.fb.com/P19692642
With them, (and all of the other patches attached to the parent task,
all of fbcode compiles and passes its tests (modulo preexisting
failures and aborts).

Reviewed By: delong.j@fb.com

Subscribers: trunkagent, folly-diffs@

FB internal diff: D1773461

Tasks: 5941250

Signature: t1:1773461:1420821964:61635f3a2efc95e8d4457c36e7a5ee3a0a50df23

9 years agomove wangle/futures to futures
James Sedgwick [Fri, 9 Jan 2015 16:27:02 +0000 (08:27 -0800)]
move wangle/futures to futures

Summary:
First step of moving futures out of wangle into folly

Namespace change will be another diff

This was just a straight codemod of "wangle/futures" to "futures"

Test Plan: wait for contbuild

Reviewed By: davejwatson@fb.com

Subscribers: trunkagent, fbcode-common-diffs@, chaoyc, search-fbcode-diffs@, lars, ruibalp, hero-diffs@, zeus-diffs@, vikas, mcduff, hitesh, cold-storage-diffs@, unicorn-diffs@, ldbrandy, targeting-diff-backend@, netego-diffs@, fugalh, adamsyta, atlas2-eng@, alandau, apollo-diffs@, bmatheny, adityab, everstore-dev@, zhuohuang, sweeney, mwa, jgehring, prometheus-diffs@, smarlow, akr, bnitka, jcoens, zhguo, jying, darshan, apodsiadlo, alikhtarov, fuegen, dzhulgakov, alihussains, jeremyfein, mshneer, folly-diffs@, wch, lins, tingy, maxwellsayles, hannesr

FB internal diff: D1772041

Tasks: 5960242

Signature: t1:1772041:1420739169:36126b1264c5d1747d241921d02b13e306c73fe1

9 years agofolly/Bits.h (BitIterator): avoid -Wsign-compare error
Jim Meyering [Wed, 7 Jan 2015 03:25:48 +0000 (19:25 -0800)]
folly/Bits.h (BitIterator): avoid -Wsign-compare error

Summary:
* folly/Bits.h (BitIterator) [bitOffset_]: Change type of this
member from ssize_t to "unsigned int".
(bitsPerBlock): Change return type from size_t to "unsigned int".
These changes avoid some -Wsign-compare-highlighted issues, e.g.,
folly/Bits.h:456:18: error: comparison between signed and unsigned integer expressions [-Werror=sign-compare]

Test Plan:
Run this and note there are fewer errors than before:
fbconfig --platform-all=gcc-4.9-glibc-2.20 -r folly && fbmake dbgo

I've also run "fbmake runtests", and it failed precisely the way it does
without this patch.

Reviewed By: andrei.alexandrescu@fb.com

Subscribers: trunkagent, net-systems@, folly-diffs@

FB internal diff: D1770613

Tasks: 5941250

Signature: t1:1770613:1420757397:7ee7e5de500c8991af2597a6245d12d8f286b4b7

9 years agofolly: fix many -Wsign-compare warning/errors reported by gcc-4.9
Jim Meyering [Tue, 6 Jan 2015 17:33:03 +0000 (09:33 -0800)]
folly: fix many -Wsign-compare warning/errors reported by gcc-4.9

Summary:
With the upgrade to gcc-4.9 fbcode's default is to enable
-Wsign-compare by default.  That exposes nontrivial technical debt.
This is part of that clean-up.  Some of these changes fix the problem
where there's an "int" for-loop index, yet an unsigned limit.
The fix is to use an unsigned type when the limit is also unsigned.
I usually choose size_t because of the general recommendation (when
writing portable code) to avoid size-tied types like uint32_t and
uint64_t unless you have a very good reason to require them.

Test Plan:
Run this and note there are fewer errors than before:
fbconfig --platform-all=gcc-4.9-glibc-2.20 tao/server && fbmake dbgo

Also, run this, compiling and running tests with gcc-4.8.1.
Verify that there are no more failures than without this patch:
fbmake opt -k && fbmake runtests_opt
Here's the tail of that process, (same with and without the patch):
Summary (total time 47.45s):
PASS: 1949
FAIL: 1
SKIP: 0
FATAL: 9
TIMEOUT: 0

Reviewed By: hans@fb.com

Subscribers: trunkagent, fugalh, njormrod, folly-diffs@

FB internal diff: D1766722

Tasks: 5941250

Signature: t1:1766722:1420762240:a8545648ddb4fd0b2adf0d3147a84409c0f706a8