Summary:
Sometimes it'll be handy to have a custom operator which is little more than the
composition of a few base operators. This makes that really easy to do, as shown
in examples, tests, and benchmarks.

Test Plan: Unit tests, benchmarks

Reviewed By: rsagula@fb.com

FB internal diff: D617152

Summary: We should be able to print ##Foo*##, even if we can't print ##Foo##.

Test Plan: Unit Tests.

Reviewed By: tudorb@fb.com

FB internal diff: D625977

Summary:
If your format string ends up being conditional, it's handy to be able to chain
together formatters.

Test Plan: Unit tests

Reviewed By: tudorb@fb.com

FB internal diff: D625502

Summary:
Convert a IOBuf chain into a fbstring, with zero copies in the common case
(unshared, unchained, no headroom) and one copy in all other cases.

Test Plan: tests added

Reviewed By: simpkins@fb.com

FB internal diff: D621210

Summary: fixing clang errors (used clang 3.1)

Test Plan: doh

Reviewed By: tudorb@fb.com

FB internal diff: D622593

Summary:
Yes, I admit, I was dumb enough to try it in a benchmark's
body.

Test Plan: review

Reviewed By: tudorb@fb.com

FB internal diff: D622613

Summary:
DynamicConverter uses some simple heuristics to determine if a
class is a container. One of those tests was to check that the
constructor 'template <class Iterator> [container-name](Iterator first,
Iterator last)' was present. That test was performed by checking if the
class could be constructed by two parameters of some dummy class.

However, it is possible to restrict the template parameter such that it
only accepts iterators, and not any arbitrary dummy class. This would be
useful, for example, to solve overload ambiguity with constructors like
'vector(const T& val, size_type n)', where T and size_type are the same
(see N3337 23.2.3 item 14). It also (I believe) produces more meaningful
compiler errors when a non-iterator is supplied, since it errors at the
function callsite instead of inside the constructor itself.

The new FBVector implementation uses such a feature, and so checking for
[container-name](dummy, dummy) will fail. Hence the dummy class has been
upgraded to reverse_iterator<T*>, a valid iterator class which almost
certainly does not have a specialized contructor in any class (and hence
will not cause any visible change in class_is_container's value).

Test Plan:
Run DynamicConverterTest; it has tests for the correctness of
class_is_container.

Reviewed By: delong.j@fb.com

FB internal diff: D620607

Summary:
Changed communicate() flags from int to a class.
Made Options and CommunicateFlags composable with |
Simplified API so you don't have to type Subprocess::Options().stdout(Subprocess::PIPE)

Test Plan: subprocess_test

Reviewed By: chip@fb.com

FB internal diff: D620186

Test Plan: subprocess_test

Reviewed By: delong.j@fb.com

FB internal diff: D619713

Test Plan: subprocess_test

Reviewed By: chip@fb.com

FB internal diff: D619189

Summary:
Surprised we don't have one.  The API is modeled after Python's
subprocess module, http://docs.python.org/2/library/subprocess.html

Inspired by
https://www.facebook.com/groups/fbcode/permalink/445399858830192/, plus
I needed this functionality now.

Test Plan: test added

Reviewed By: chip@fb.com

FB internal diff: D614056

Summary: parseInt128 is only available for GCC 4.7 and above.

Test Plan: .

Reviewed By: igorzi@fb.com

FB internal diff: D618456

Blame Revision: rFBCODEeaec97e2991f

Summary:
Add benchmarks for throwing an exception versus returning
std::exception_ptr or other types of return values.

Throwing an exception is very expensive.  Calling std::make_exception_ptr()
without ever throwing is nearly as bad.

The exc_ptr_param_return tests were the ones I was most interested in.
(Accepting a std::exception_ptr* argument so the caller can indicate if they
want to receive an exception or not.)  This is fast if the caller doesn't care
about the exception value, but very slow if an error occurs and an
exception_ptr is required.

Test Plan:
======================================================================
folly/test/function_benchmark/main.cpp    relative  time/iter  iters/s
======================================================================
throw_exception                                        3.90us  256.25K
catch_no_exception                                     1.88ns  533.25M
return_exc_ptr                                         2.79us  357.85K
exc_ptr_param_return                                   2.83us  353.25K
exc_ptr_param_return_null                              2.25ns  444.38M
return_string                                         69.39ns   14.41M
return_string_noexcept                                69.39ns   14.41M
return_code                                            1.50ns  666.54M
return_code_noexcept                                   1.50ns  666.54M

Reviewed By: rajat@fb.com

FB internal diff: D616474

Summary: Just that.

Test Plan: .

Reviewed By: igorzi@fb.com

FB internal diff: D613291

Summary:
folly::toJson as demonstrated by the test cases was wrongly encoding utf8 strings.
specifically, a utf8 char made up of x bytes was encodeded into x unicode chars.

for example, the char:
\u2665
which is made of 3 bytes:
\xe2\x99\xa5
was encoded correctly when using encode_non_ascii = true:
"\u2665"
but when encode_non_ascii = false, the json value was wrongly set as:
"\u00e2\u0099\u00a5"

because we use an signed char that wrongly detects non-readable chars with
ascii value > 127 as control chars with ascii value < 32 (\t, \n, etc.)

Test Plan: run the test

Reviewed By: delong.j@fb.com

FB internal diff: D612782

Summary: Tests were run in 'opt', which masked issues alterted by asserts.

Test Plan: Unit tests

Reviewed By: delong.j@fb.com

FB internal diff: D611957

Summary:
the currently code calls both ~SkipListNode() and node->data_.~value_type()
causes double destructing the object.

Test Plan: adding dihde's testing code to a test case

Reviewed By: emailweixu@fb.com

FB internal diff: D612289

Summary: fix Benchmark docs

Test Plan: read

Reviewed By: andrei.alexandrescu@fb.com

FB internal diff: D611820

Summary:
Moving Comprehensions library into Folly and refactoring its interface to be much more modular and composable. There are now two core abstractions:

# Generators: Standalone classes supporting ##apply()## and optionally ##foreach()##. These all inherit from ##GenImpl<T, Self>##.
# Operators: Standalone classes which, when composed with a generator, produce a new generator. These all inherit from ##OperatorImpl<Self>##.

These generators may be composed with ##operator|## overloads which only match ##const GenImpl<T, Self>&## on the left like ##gen | op##.  Additionally, generator may be consumed inline with ##gen | lambda## like ##gen | [](int x) { cout << x << endl; };##.

With this design, new operators may be added very simply without modifying the core library and templates are instantiated only exactly as needed.

Example:

```lang=cpp
auto sum = seq(1, 10) | filter(isPrime) | sum;
seq(1, 10) | [](int i) {
cout << i << endl;
};
```

Test Plan: Unit tests

Reviewed By: andrei.alexandrescu@fb.com

FB internal diff: D542215

Summary: C++11 added front() and back() that return by reference. fbstring also added front() and back() as non-standard convenience functions, which returned by value. This diff aligns fbstring with the standard.

Test Plan: added and ran unittest

Reviewed By: delong.j@fb.com

FB internal diff: D607574

Summary: Optional, like boost::optional, but with full rvalue support.

Test Plan: Unit tests

Reviewed By: delong.j@fb.com

FB internal diff: D571810

Summary: For std::exception_ptr.

Test Plan: .

Reviewed By: davejwatson@fb.com

FB internal diff: D600345

Summary:
Understood how traits are implemented in folly/Traits.h and added
examples on how to use them.

Test Plan: no specific tests

Reviewed By: andrei.alexandrescu@fb.com

FB internal diff: D546631

Summary:
SpookyHashV2 is backwards incompatible with V1.  I renamed the existing
SpookyHash class to SpookyHashV1 (and fixed all uses); the new class is
called SpookyHashV2.

From http://burtleburtle.net/bob/hash/spooky.html:

Both V1 and V2 pass all the tests. V2 corrects two oversights in V1:

In the short hash, there was a d = length that should have been d += length,
which means some entropy got dropped on the floor. It passed the tests anyhow,
but fixing this probably means more distinct info from the message makes it
into the result.

The long hash always ended in mix()+end(), but only end() was needed. Removing
the extra call to mix() makes all long hashes faster by a small constant
amount.

Test Plan: test added; unicorn compiles

Reviewed By: andrei.alexandrescu@fb.com

FB internal diff: D597173

Summary: '\0' may actually be part of string. We cannot assume its null terminated and should use another form of fnv32.

Test Plan: FBStringTest

Reviewed By: xliux@fb.com

FB internal diff: D595287

Summary:
It seems it is not std::bind which is slow but the construction of
std::function from std::bind that is slow.

Test Plan: ran benchmarks

Reviewed By: delong.j@fb.com

FB internal diff: D581967

Summary: The Arena allocator knows how much memory it's using up, so create a function that allows clients to figure out how much it's using. Also create a unit test to sanity check this stuff.

Test Plan: folly/test/ArenaTest.cpp

Reviewed By: tudorb@fb.com

FB internal diff: D579399

Summary:
Background: we want to use folly::json but not change the way data is represented.
Since we need to store \r and \n in strings and the library we're currently using does this,
let's do it in folly too.

Test Plan: unit tests pass

Reviewed By: delong.j@fb.com

FB internal diff: D584960

Summary: These are necessary for string-keyed maps, since maps const their keys.

Test Plan: I wrote a test.

Reviewed By: njormrod@fb.com

FB internal diff: D582283

Summary:
Fixed escaping bug where the `>` in `operator->` was
interpreted as starting a code block.  Folly github pull #15.

Test Plan: no

Reviewed By: njormrod@fb.com

FB internal diff: D582873

Summary:
After a conversation with njormrod I remembered why we didn't
do this.  Let's document it.

Test Plan: It's docs.

Reviewed By: njormrod@fb.com

FB internal diff: D582651

Summary:
Array of spinlocks where each one is padded to prevent false
sharing.

Useful for shard-based locking implementations in environments where
contention is unlikely.

Test Plan: ran tests in 'common'

Reviewed By: soren@fb.com

FB internal diff: D582149

Summary:
In https://phabricator.fb.com/D511928 Brian mentioned the current API for string append is insufficient for appending to a buffer. That made me curious about the relative performance of classic and table-based number to ASCII conversions.

The results were interesting as on the average (over all digit lengths) the table-based conversion was faster, but performance was lackluster (in the worst case half as fast as the classic implementation) for large numbers, I presume due to the cache misses incurred by the tables.

This diff proposes an improved unsigned-to-ASCII primitive that is much faster than both table-based (previous Folly) and classic primitive. The key is a fast digits10() implementation that precomputes the space required by the conversion. After that, the digits are issued in place, no more reverse required. The new routine is up to 14x faster than the classic implementation, depending on the number of digits (benchmarks in comments).

Adding a few people who may be interested in the matter. Brian, thanks for bringing this matter up; if this gets in you may want to use the folly routine in proxygen.

Test Plan: unittest and benchmarks.

Reviewed By: simpkins@fb.com

FB internal diff: D515572

Summary:
The following asserts:
fbstring str(1337, 'f');
fbstring cp = str;
cp.push_back('f');

This is problematic since ml_.capacity() != capacity() inside fbstring_core
for shared strings, which causes us not to un-share prior to push_back.

Test Plan: Existing tests, add unittest case.

Reviewed By: tudorb@fb.com

FB internal diff: D580267

Summary:
implement ##find_first_of## and optimize ##Range.find(char)##

============================================================================
folly/test/RangeBenchmark.cpp                   relative  time/iter  iters/s
============================================================================
LongFindSingleCharDirect                                     2.76ms   362.63
LongFindSingleCharRange                           15.88%    17.37ms    57.58
ShortFindSingleCharDirect                                   53.41fs   18.72T
ShortFindSingleCharRange                           0.00%    29.22ns   34.22M
============================================================================

Test Plan:
- added new tests

- ran all folly tests

fbconfig -r folly/ && mkk runtests_opt

Reviewed By: tudorb@fb.com

FB internal diff: D576720

Summary: This makes code easy to read for eyes used to unique/shared ptrs.

Test Plan: unit-tests

Reviewed By: delong.j@fb.com

FB internal diff: D575997

Summary: Add a pop_back() function to dynamic arrays.

Test Plan: - used it

Reviewed By: delong.j@fb.com

FB internal diff: D569440

Summary: Couldn't get an HPHP extension to compile against this

Test Plan: compiled HPHP including folly/dynamic.h

Reviewed By: delong.j@fb.com

FB internal diff: D574364

Test Plan: by hand

Reviewed By: philipp@fb.com

FB internal diff: D570233

Summary:
Folly supports -inf and nan but not -nan, instead it raised unexpected exceptions not handled in the upstream.

This diff is separated out from D569816 -- Diff1.

Test Plan: std::isnan(folly::to<double>(-nan))

Reviewed By: tudorb@fb.com

FB internal diff: D569939