Summary:
Provide translations for gcc noreturn attribute

__attribute__((noreturn)) is gcc specific, clang understands it on nix systems, but for msvc __declspec(noreturn) is the compiler specific version, which clang will imitate/use on windows.  There was already a FOLLY_NORETURN in portability.h, however because of __declspec limitations it must prefix the declaration, not postfix.  The gcc noreturn attribute does not have this limitation and will work prefixed OR postfixed, so to keep from turning code into spaghetti nonsense, the macro was moved to the front of declations where it is currently used.  This will allow it to work with the proper definitions on clang, clang on windows, gcc, and msvc

Test Plan: fbmake runtests

Reviewed By: delong.j@fb.com

FB internal diff: D1279466

Summary:
Provide translations for gcc always_inline and noinline attribute

Change to using a macro from portability.h for attributes for using always_inline and noinline for cross platform support

Test Plan: fbmake runtests

Reviewed By: delong.j@fb.com

FB internal diff: D1279652

Summary:
Using StringPiece makes it easier than std::string

Facebook: Motivation for this diff is found in D1253595

Test Plan:
Added new unittest

fbconfig folly/io/test && fbmake runtests_opt

Reviewed By: simpkins@fb.com

FB internal diff: D1276185

Summary: As part of the windows port, uint is not defined as a type and is not standard, it makes msvc choke, a simple change to unsigned int fixes the issue

Test Plan: fbmake runtests

Reviewed By: delong.j@fb.com

FB internal diff: D1278487

Test Plan: used it

Reviewed By: lucian@fb.com

FB internal diff: D1277063

@override-unit-failures

Summary:
as per discussion from https://phabricator.fb.com/D1274352

After changing the queue length to be a small value (100) I didn't see
improvement from cache alignment of atomics (too much noise). Dropping
that and just fixing the benchmark.

Test Plan: n/a

Reviewed By: ngbronson@fb.com

FB internal diff: D1276591

Summary:
The code originally contained the line

EXPECT_EQ((rcount == lineCount), r); // EOF iff at lineCount

Which asserted that, on the last iteration, EOF was being read (and thus
r was true).

This line has been causing errors in contbuild. There seem to be two
issues at play.

First, the left hand side of the EXPECT_EQ is always false. rcount is
incremented three lines down, and the (implicit) loop ends when rcount
equals lineCount. This means that, on the last iteration, rcount is
actually one less than lineCount. Hence the check is effectively
asserting that r is false.

The second issue is that r is rarely true. Empirically, a necessary but
not sufficient condition is that there be lots of other processes
running at the same time (such as when running fbconfig -r folly &&
fbmake runtests_opt). This would explain why the test passes during
development but fails in contbuild.

I am guessing that, when there are few processes running, Subprocess is
predictably ceasing to read before whichever other process yields EOF,
and that this does not always happen when there is resource contention.

As an immediate fix, I am asserting that r may only be true on the last
iteration. Could someone (cc @tudorb) who is more familiar with the
intricacies of Subprocess please check to see if this is an idiosyncrasy
of the test, or if there is actually an underlying bug.

Test Plan:
Before the change, run fbconfig -r folly && fbmake runtests_opt
repeatedly. Notice that most of the iterations yield an error from line
376 (which is consistent with contbuild error logs).

After the change, notice that the error disappears.

Reviewed By: andrewjcg@fb.com

FB internal diff: D1269124

Summary:
same as map, but runs it's argument in parallel over a number of threads.

@override-unit-failures

Test Plan: added test

Reviewed By: tjackson@fb.com

FB internal diff: D1258364

Summary: Changing the parent's WD is prone to race conditions of various sorts, and needlessly alters parent state. Python's subprocess also has this feature.

Test Plan: fbmake dbg _bin/folly/test/subprocess_test ; ../_bin/folly/test/subprocess_test

Reviewed By: agoder@fb.com

FB internal diff: D1269526

Summary:
* folly/test/CacheLocalityTest.cpp (contentionAtWidth): Work around
clang's prohibition against using flexible arrays in a lambda.

Test Plan:
fbconfig --clang --with-project-version clang:dev folly/test \
&& fbmake runtests

Reviewed By: ngbronson@fb.com

FB internal diff: D1263620

Summary: Allow specifying custom alignment.

Test Plan: fbconfig -r folly/test && fbmake opt -j32

@override-unit-failures

Reviewed By: lucian@fb.com

FB internal diff: D1264851

Summary:
D1261546 introduced regression in `sizeof(ConcurrentSkipList)`.
In case of `NodeAlloc = folly::SysAlloc`, size of an empty `NodeAlloc` struct
is 1 byte, due to the alignment requirements of the next field, it got
translated into wasting 8 bytes.

Test Plan: fbconfig -r folly/test && fbmake opt -j32

@override-unit-failures

Reviewed By: lucian@fb.com

FB internal diff: D1264730

Summary:
* folly/folly-config.h (FOLLY_HAVE_CONSTEXPR_STRLEN): Define to
zero when compiling with clang, since clang's strlen is not constexpr.

Test Plan:
Run this:
fbconfig --clang --with-project-version clang:dev folly/test && fbconfig dbg
and observe that the following no longer strikes:
folly/test/RangeTest.cpp:304:25: error: constexpr variable 'hello1' must be initialized by a constant expression

I also tried using clang-3.4 (the current default),
fbconfig --clang folly/test && fbmake dbg
with these results:
https://phabricator.fb.com/P8583443

Reviewed By: pgriess@fb.com

FB internal diff: D1262840

Summary: Fix make check.  Benchmark.cpp was included twice, eventfd doesn't exist anymore.

Test Plan: autoreconf -i; ./configure; make check on ubuntu

Reviewed By: pgriess@fb.com

FB internal diff: D1259706

Summary:
Automake complains there are two Malloc.cpp rules - rename one of them.

Facebook: Actually this file doesn't even build in fbconfig - is it still necessary?

Test Plan: autoreconf -i on ubuntu doens't fail

Reviewed By: pgriess@fb.com

FB internal diff: D1257372

Summary:
When all the variants I am dreaming of work, there's a little more than
100 tests generated.

Test Plan: fbmake runtests

Reviewed By: hannesr@fb.com

FB internal diff: D1258754

Summary: Now compiles in 18 seconds. Took two minutes before.

Test Plan: `fbmake runtests_opt`

Reviewed By: smith@fb.com

FB internal diff: D1243689

Summary:
Allow specifying `SimpleAllocator` to use for node allocation.
Added optimizations for arena allocators.

Test Plan: fbconfig -r folly/test && fbmake opt -j32

Reviewed By: soren@fb.com

FB internal diff: D1261546

Summary:
There is no reason to force heap allocation for
`ConcurrentSkipList`.

Test Plan: fbconfig -r unicorn/test && fbmake opt -j32

@override-unit-failures

Reviewed By: lucian@fb.com

FB internal diff: D1261183

Summary: For use with Thrift::Frozen, especially.

Test Plan: Unit tests

Reviewed By: philipp@fb.com

FB internal diff: D1255257

Summary:
Subtle bugfix to fbvector

Reproduction requirements:
- emplace into a vector at a non-end position
- the vector has enough capacity for the extra elements
- the value type's internal state is poisoned after a move

What should happen, explained by picture:

INITIAL SETUP

abc1234XY____
^
want to insert two elements here

FIRST STEP: uninitialized move-construction
x and y are initialized, but could be in an invalid state

abc1234xyXY__

SECOND STEP: move-assignment, in reverse order

abc123xy4XY__
abc12xy34XY__
abc1xy234XY__
abcxy1234XY__

What actually happens:
An indexing error causes the entire tail (##1234xy##) to be
move-assigned, instead of the intended head of the tail (##1234##).

If the data type's move operator does not invalidate its own data (i.e.
move is essentially a copy, as is the case with atomic types) then the
move assignment of ##y## onto ##Y## and ##x## onto ##X## is basically a
no-op. However, for types that do invalidate their own data when being
the source of a move (e.g. non-empty vectors and fbvectors), then the
end result will have bad data in place of ##X## and ##Y##.

Detection:
This bug has lain dormant for a while.

Very few people emplace into a vector, and those few who do could be
saved by using a copy-movable type like int.

The original test case did not use a data type which poisoned itself
after being the source of a move, so the tests did not notice any
oddities.

A new testsuite, which does poison itself, did notice.

Test Plan:
re-enable the test/stl_test mega-test. Run it. All tests
pass.

fbconfig -r folly && fbmake runtests_opt

Run fbvector through the new test suite (which initially caught the
error). It now passes.

Reviewed By: robbert@fb.com

FB internal diff: D1257258

Summary: These were required to build folly on OSX.

Test Plan: build

Reviewed By: davejwatson@fb.com

FB internal diff: D1256856

Summary:
There are two operations on `FutureObject` (the backing object that both `Future` and `Promise` front), which are not threadsafe currently. Those are `setContinuation` and `fulfil`. They're not threadsafe because they have a race condition between checking to see whether they should run the continuation and the other side making its modification. This diff introduces a `std::atomic_flag` variable to avoid the race and make these threadsafe.

NB Making `Future`/`Promise` threadsafe does not make this pattern safe:

f1.via(...).then(...).then(...)

(In a hypothetical world where `Future::via` is another name for `executeWithSameThread`)
There is a race between the future from `via` and the call of `then` (and between the first and second `then`s), so the `then`s could execute in the far thread as intended, or they could execute in the current thread (probably not what was intended). This diff does not solve that semantic problem of which thread this runs in. But it does make it safer, in that all those continuations will always execute, just maybe not in the thread you had intended.

You may ask what is the benefit if that semantic problem still exists? That's up for debate. I think the safety is worth it: at least everything a n00b throws at it will *work*. The question of thread semantics can iterate. If we end up with thread semantics that really don't need locks (e.g. maybe because we move to an Rx/Later style "launch" at the end of a chain) we can always toss this atomic.

How does this affect performance? I'm doing some experiments and going to write up my findings. Early results indicate that outlier performance is impacted but not by a lot, but don't count those chickens until I hatch them. Part of the reason for sending this diff out is to run windtunnel experiments.

Test Plan: tests, benchmark, close inspection

Reviewed By: davejwatson@fb.com

FB internal diff: D1241822

Summary:
TIMED_SYNCHRONIZED_CONST is acquiring a read-write lock but releasing a shared lock.  Chaos ensues.
@override-unit-failures

Test Plan: Add unit test, test times out before fix, passes after fix.

Reviewed By: andrei.alexandrescu@fb.com

FB internal diff: D1251518

Blame Revision: D327492

Summary:
Subprocess::communicate callbacks are level-triggered, which makes writing
"chatty" communication protocols difficult -- you often want to silence the
write callback until you read the expected message. Add
Subprocess::enableNotifications() for this purpose.

Test Plan: test added

Reviewed By: lucian@fb.com

FB internal diff: D1251564

@override-unit-failures

Summary: Add loopOnce() method which provides functionality similar to event_base_loop(base, EVLOOP_ONCE);

Test Plan: fbmake

Reviewed By: simpkins@fb.com

FB internal diff: D1249753

Summary: Utility is doing reverse of prettyPrint. This is a first sketch.

Test Plan:
1) We can reverse all tests from prettyPrint test and use for our function. For now I just reversed several of them
2) We can also test prettyPrint and prettyToDouble together by checking whether prettyToDouble(prettyPrint(X)) == X. This is implemented.

Reviewed By: marcelo.juchem@fb.com

FB internal diff: D1159879

Summary:
This should be standard compliant. `__PRETTY_FUNCTION__` seems
unnecessary, it just prints `basic_fbstring`, might as well use that and
simplify the code.

Test Plan:
`fbconfig --platform-all=gcc-4.8.1-glibc-2.17-fb folly/test`
`fbmake runtests_dbg`
`fbmake runtests_opt`

`fbconfig --platform-all=gcc-4.7.1-glibc-2.14.1 folly/test`
`fbmake runtests_dbg`
`fbmake runtests_opt`

`fbconfig --clang folly/test`
`...`

Reviewed By: andrei.alexandrescu@fb.com

FB internal diff: D1243670

Summary:
Change NULLs to nullptrs.

Facebook:
I was tired of seeing lint errors for NULL -> nullptr, so I
fixed it.

How: modified flint's tokenizer to output lines of the form

sed -i '__LINE__s|\<NULL\>|nullptr|g' __FILE__

for each NULL token encountered.

Ran the sed lines, restricted to filepaths with extensions
h,hpp,cc,cpp,tcc. Did not apply to FacebookUpdate, due to weird
formatting; hphp, to be done in a separate diff; payment, due to a large
number of generated text which didn't play well with my flow; and
memcache, which has lots of .h's that cannot contain nullptr's because
they are included from .c's.

@bypass-lint

Test Plan:
fbconfig -r common folly && fbmake opt && fbmake runtests_opt
^ of the 4k+ test cases, about two dozen are failing. Slightly more
failed in master than with my diff applied.

arc unit

Reviewed By: andrei.alexandrescu@fb.com

FB internal diff: D1234261

Test Plan: Unit tests

Reviewed By: marcelo.juchem@fb.com

FB internal diff: D1236630

Summary:
For proper rounding, and also so memOpInChunks doesn't try to mlock / munlock /
madvise / munmap in chunks smaller than the page size.

Test Plan: memory_mapping_test

Reviewed By: lucian@fb.com

FB internal diff: D1232153

Summary:
Let's double check that `small_vector` is doing the right thing. So
it doesn't just accidentally work.

Test Plan:
`fbconfig folly/wangle && fbmake runtests`
`fbconfig --clang folly/wangle && fbmake runtests`

Reviewed By: marccelani@fb.com

FB internal diff: D1231360

Summary: Moar c++11 features.  marccelani's comments in D1195393, but in a separate diff.

Test Plan:
contbuild
fbconfig thrift/lib/cpp/test/
fbmake runtests

Reviewed By: rajat@fb.com

FB internal diff: D1219158

Summary:
This implementation improves the performance of AsciiCaseInsensitive.
AsciiCaseInsensitive uses toupper, and it's slow.
Since toupper only works for ascii characters and this function does exactly the same thing over the range of [0..256), we can gain 3x performance here.

Test Plan: unit test and benchmark

Reviewed By: lucian@fb.com

FB internal diff: D1217596

@override-unit-failures

Summary: `std::async` doesn't play nicely with our clang. Replace it with `std::thread`.

Test Plan: `fbconfig --clang folly/wangle && fbmake runtests`

Reviewed By: marccelani@fb.com

FB internal diff: D1230197

Summary:
In a shared library, we're using whichever memory allocator the main program
uses, but we may still believe we're using jemalloc if we load libjemalloc.so
(because rallocm is defined). This means we'll pass to rallocm pointers that
were not allocated by jemalloc, which is bad.

Test Plan: tests added

Reviewed By: ngbronson@fb.com

FB internal diff: D1228483

Summary: So people can more easily write functions which take contiguous values from memory.

Test Plan: Unit tests

Reviewed By: tudorb@fb.com

FB internal diff: D1217809

Summary: Now we could use small_vector::max_size in constant expressions

Test Plan: unittest

Reviewed By: delong.j@fb.com

FB internal diff: D1224274

Test Plan: fbconfig -r folly/experimental/symbolizer && fbmake runtests_opt && fbmake runtests

Reviewed By: simpkins@fb.com

FB internal diff: D1224552

@override-unit-failures