lib: allow server.listen({ port: "1234" })

net.connect() accepts `{ port: "1234" }` (i.e. a string) as of commit
9d2b89d06 ("net: allow port 0 in connect()") but net.Server#listen()
did not, creating a minor inconsistency.  This commit rectifies that.

Fixes: https://github.com/iojs/io.js/issues/1111
PR-URL: https://github.com/iojs/io.js/pull/1116
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>

doc: move checkServerIdentity option to tls.connect()

PR-URL: https://github.com/iojs/io.js/pull/1107
Reviewed-By: Jeremiah Senkpiel <fishrock123@rocketmail.com>

doc: fix missing periods in url.markdown

PR-URL: https://github.com/iojs/io.js/pull/1115
Reviewed-By: Jeremiah Senkpiel <fishrock123@rocketmail.com>

tls_wrap: proxy handle methods in prototype

Set proxied methods wrappers in `TLSWrap` prototype instead of doing it
on every socket allocation. Should speed up things a bit and will
certainly make heapsnapshot less verbose.

PR-URL: https://github.com/iojs/io.js/pull/1108
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Trevor Norris <trev.norris@gmail.com>

buffer: Don't assign .parent if none exists

The .parent property of the allocated buffer should remain undefined in
the case that it's not a slice. Also included test to verify this.

PR-URL: https://github.com/iojs/io.js/pull/1109
Reviewed-By: Chris Dickinson <christopher.s.dickinson@gmail.com>
Reviewed-By: Fedor Indutny <fedor@indutny.com>

Working on v1.5.2

2015-03-09 io.js v1.5.1 Release

Notable changes:

* tls: The reported TLS memory leak has been at least partially
  resolved via various commits in this release. Current testing indicated
  that there may still be some leak problems. Progress being tracked at:
  https://github.com/iojs/io.js/issues/1075
* http: Fixed an error reported at
  https://github.com/joyent/node/issues/9348 and
  https://github.com/npm/npm/issues/7349
  Pending data was not being fully read upon an 'error' event leading to
  an assertion failure on socket.destroy().
  (Fedor Indutny) https://github.com/iojs/io.js/pull/1103

benchmark: chunky http client benchmark variation

PR-URL: https://github.com/iojs/io.js/pull/228
Reviewed-By: Trevor Norris <trev.norris@gmail.com>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>

http_client: ensure empty socket on error

Read all pending data out of the socket on `error` event and ensure that
no `data`/`end` handlers will be invoked on `socket.destroy()`.
Otherwise following assertion happens:

    AssertionError: null == true
        at TLSSocket.socketOnData (_http_client.js:308:3)
        at TLSSocket.emit (events.js:107:17)
        at TLSSocket.Readable.read (_stream_readable.js:373:10)
        at TLSSocket.socketCloseListener (_http_client.js:229:10)
        at TLSSocket.emit (events.js:129:20)
        at TCP.close (net.js:476:12)

Fix: https://github.com/joyent/node/issues/9348
PR-URL: https://github.com/iojs/io.js/pull/1103
Reviewed-By: Rod Vagg <rod@vagg.org>
Reviewed-By: Nicu Micleușanu <micnic90@gmail.com>

node_crypto_bio: adjust external memory size

Adjust V8's external memory size when allocating buffers for TLS data to
ensure that V8 has enough information to trigger the GC at right time.

PR-URL: https://github.com/iojs/io.js/pull/1085
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>

doc: fix confusion markdown in util.markdown

PR-URL: https://github.com/iojs/io.js/pull/1097
Reviewed-By: Brendan Ashworth <brendan.ashworth@me.com>

fs: fix .write() not coercing non-string values

Fixes: https://github.com/iojs/io.js/issues/1098
PR-URL: https://github.com/iojs/io.js/pull/1102
Reviewed-By: Brendan Ashworth <brendan.ashworth@me.com>

url: remove redundant assignment in url.parse

PR-URL: https://github.com/iojs/io.js/pull/1095
Reviewed-By: Jeremiah Senkpiel <fishrock123@rocketmail.com>
Reviewed-By: Nicu Micleușanu <micnic90@gmail.com>

stream_base: WriteWrap::New/::Dispose

Encapsulate allocation/disposal of `WriteWrap` instances into the
`WriteWrap` class itself.

PR-URL: https://github.com/iojs/io.js/pull/1090
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>

tls: do not leak WriteWrap objects

Kill WriteWrap instances that are allocated in `tls_wrap.cc` internally.

Fix: https://github.com/iojs/io.js/issues/1075
PR-URL: https://github.com/iojs/io.js/pull/1090
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>

doc: update clang version prerequisite

See commit `4877ec0`, where it now warns on a clang++ version older than
3.4.

PR-URL: https://github.com/iojs/io.js/pull/1094
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>

src: fix memory leak in fs.writeSync error path

The SYNC_CALL macro returns on error, bypassing the delete[] call.

Mea culpa, it looks like I introduced this memory leak back in 2013,
in commit d2b80b8a ("src: clean up FSReqWrap").

PR-URL: https://github.com/iojs/io.js/pull/1092
Reviewed-By: Fedor Indutny <fedor@indutny.com>

src: fix mismatched delete[] in src/node_file.cc

Fix a bad delete of a pointer that was allocated with placement new.
Casting the pointer was not the right solution because there was at
least one non-placement new constructor call.

This commit rewrites FSReqWrap to be more explicit about ownership of
the auxiliary data and removes a number of egregious const_casts.
The ASYNC_DEST_CALL macro also gets significantly slimmed down.

PR-URL: https://github.com/iojs/io.js/pull/1092
Reviewed-By: Fedor Indutny <fedor@indutny.com>

doc: replace article "an" with "a" in net docs

PR-URL: https://github.com/iojs/io.js/pull/1093
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>

crypto: fix leak in SafeX509ExtPrint

`ASN1_item_d2i`'s return value must be freed by the owner.

Fix: https://github.com/iojs/io.js/issues/1075
PR-URL: https://github.com/iojs/io.js/pull/1087
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>

src: add missing Context::Scope

Add a Context::Scope that was overlooked in commit 583a868
("stream_wrap: add HandleScope's in uv callbacks").

PR-URL: https://github.com/iojs/io.js/pull/1084
Reviewed-By: Fedor Indutny <fedor@indutny.com>

Working on v1.5.1

2015-03-06 io.js v1.5.0 Release

Notable changes:

* buffer: New `Buffer#indexOf()` method, modelled off `Array#indexOf()`.
  Accepts a String, Buffer or a Number. Strings are interpreted as UTF8.
  (Trevor Norris) https://github.com/iojs/io.js/pull/561
* fs: `options` object properties in `'fs'` methods no longer perform a
  `hasOwnProperty()` check, thereby allowing options objects to have
  prototype properties that apply. (Jonathan Ong)
  https://github.com/iojs/io.js/pull/635
* tls: A likely TLS memory leak was reported by PayPal. Some of the recent
  changes in stream_wrap appear to be to blame. The initial fix is in
  https://github.com/iojs/io.js/pull/1078, you can track the progress
  toward closing the leak at
  https://github.com/iojs/io.js/issues/1075 (Fedor Indutny).
* npm: Upgrade npm to 2.7.0. See npm CHANGELOG.md:
  https://github.com/npm/npm/blob/master/CHANGELOG.md#v270-2015-02-26
  for details including why this is a semver-minor when it could have
  been semver-major.
* TC: Colin Ihrig (@cjihrig) resigned from the TC due to his desire to do
  more code and fewer meetings.

deps: make node-gyp work with io.js

Every npm version bump requires a few patches to be floated on
node-gyp for io.js compatibility. These patches are found in
03d199276e21c1fa08d8df14eeb654c90cc5aa20,
5de334c23096492014a097ff487f07ad8eaee6d2, and
da730c76e98fb9fd18dac445dafbbec74d79f802. This commit squashes
them into a single commit.

PR-URL: https://github.com/iojs/io.js/pull/990
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>

deps: upgrade npm to 2.7.0

PR-URL: https://github.com/iojs/io.js/pull/1080
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>

tls_wrap: do not hold persistent ref to parent

Hold non-persistent reference in JS, rather than in C++ to avoid cycles.

PR-URL: https://github.com/iojs/io.js/pull/1078
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>

js_stream: fix leak of instances

Don't forget to call `MakeWeak` to ensure that instance objects are
garbage collectable.

PR-URL: https://github.com/iojs/io.js/pull/1078
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>

stream_wrap: add HandleScope's in uv callbacks

Ensure that no handles will leak into global HandleScope by adding
HandleScope's in all JS-calling libuv callbacks in `stream_wrap.cc`.

Fix: https://github.com/iojs/io.js/issues/1075
PR-URL: https://github.com/iojs/io.js/pull/1078
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>

benchmark: fix `wrk` check

PR-URL: https://github.com/iojs/io.js/pull/1076
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>

build: remove mdb from io.js

After upgrading to a newer v8 mdb is pretty much broken - even if using
the latest updates from nodejs. If nodejs decides to update their v8 we
can give it another go, but for now it's better to remove it than have
it in our tree unsupported.

PR-URL: https://github.com/iojs/io.js/pull/1023
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Rod Vagg <rod@vagg.org>

http: send Content-Length when possible

This changes the behavior for http to send send a Content-Length header
instead of using chunked encoding when we know the size of the body when
sending the headers.

Fixes: https://github.com/iojs/io.js/issues/1044
PR-URL: https://github.com/iojs/io.js/pull/1062
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Rod Vagg <rod@vagg.org>
Reviewed-By: Brian White <mscdex@mscdex.net>

src: fix ucs-2 buffer encoding regression

StringBytes::Write() did a plain memcpy() when is_extern is true but
that's wrong when the source is a two-byte string and the destination
a one-byte or UTF-8 string.

The impact is limited to strings > 1,031,913 bytes because those are
normally the only strings that are externalized, although the use of
the 'externalize strings' extension (--expose_externalize_string) can
also trigger it.

This commit also cleans up the bytes versus characters confusion in
StringBytes::Write() because that was closely intertwined with the
UCS-2 encoding regression.  One wasn't fixable without the other.

Fixes: https://github.com/iojs/io.js/issues/1024
Fixes: https://github.com/joyent/node/issues/8683
PR-URL: https://github.com/iojs/io.js/pull/1042
Reviewed-By: Trevor Norris <trev.norris@gmail.com>

src: fix external string length calculation

Make StringBytes::GetExternalParts() return the byte length for two-byte
strings, not the character length.  Its callers operate on bytes, not
characters.

This also fixes StringBytes::Size() reporting only half of the actual
number of bytes for external two-byte strings.

PR-URL: https://github.com/iojs/io.js/pull/1042
Reviewed-By: Trevor Norris <trev.norris@gmail.com>

test: simplify parallel/test-stringbytes-external

Make the algorithm that creates the big input strings a little easier
to comprehend.  No functional changes, the string lengths are unchanged.

PR-URL: https://github.com/iojs/io.js/pull/1042
Reviewed-By: Trevor Norris <trev.norris@gmail.com>

src: rename confusingly named local variable

Rename `val_` to `string`.  The underscore suffix is normally reserved
for data members, not locals, and it's not a great name in the first
place.

PR-URL: https://github.com/iojs/io.js/pull/1042
Reviewed-By: Trevor Norris <trev.norris@gmail.com>

src: simplify node::Utf8Value()

* Remove kStorageSize constant.

* Remove superfluous local variable and reinterpret_cast.

* Reorder data members so the length field and data pointer
  (if not the data itself) fit in a single cache line.

PR-URL: https://github.com/iojs/io.js/pull/1042
Reviewed-By: Trevor Norris <trev.norris@gmail.com>

src: remove NODE_INVALID_UTF8 environment variable

Introduced in joyent/node v0.10 as a backwards compatibility measure.
It's an ugly hack and allowing invalid UTF-8 is not a good idea in the
first place, remove it.

PR-URL: https://github.com/iojs/io.js/pull/1042
Reviewed-By: Trevor Norris <trev.norris@gmail.com>

src: fix gc heuristic for external twobyte strings

Large external two-byte strings reported their character length instead
of their byte length, throwing off the garbage collector heuristic by
a factor of two.

PR-URL: https://github.com/iojs/io.js/pull/1042
Reviewed-By: Trevor Norris <trev.norris@gmail.com>

src: remove unused code

PR-URL: https://github.com/iojs/io.js/pull/1042
Reviewed-By: Trevor Norris <trev.norris@gmail.com>

src: extract node env init out of process init

PR-URL: https://github.com/iojs/io.js/pull/980
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>

test: don't spawn child processes in domain test

Make parallel/test-domain-abort-on-uncaught a little easier to debug,
make it execute the tests in the same process instead of each test in
a separate child process.

PR-URL: https://github.com/iojs/io.js/pull/974
Reviewed-By: Chris Dickinson <christopher.s.dickinson@gmail.com>

src: fix -Wempty-body compiler warnings

Turn counter macros into no-op instructions when counters are disabled.
Evaluating to nothing makes gcc complain when the macro is used in a
conditional.  Fixes the following warning:

    ../src/tls_wrap.cc:320:5: warning:
    suggest braces around empty body in an 'if' statement [-Wempty-body]
         NODE_COUNT_NET_BYTES_SENT(write_size_);
         ^

PR-URL: https://github.com/iojs/io.js/pull/974
Reviewed-By: Chris Dickinson <christopher.s.dickinson@gmail.com>

build: add basic mips/mipsel support

Extend the configure script so that it knows how to generate build files
for mips and mipsel.  Actually building io.js is pending MIPS buildbots.

PR-URL: https://github.com/iojs/io.js/pull/1045
Reviewed-By: Rod Vagg <rod@vagg.org>

lib: avoid .toLowerCase() call in Buffer#write()

Avoid a costly String#toLowerCase() call in Buffer#write() in the
common case, i.e., that the string is already lowercase.  Reduces
the running time of the following benchmark by about 40%:

    for (var b = Buffer(1), i = 0; i < 25e6; ++i) b.write('x', 'ucs2');

PR-URL: https://github.com/iojs/io.js/pull/1048
Reviewed-By: Trevor Norris <trev.norris@gmail.com>

lib: hand-optimize Buffer constructor

The Buffer constructor is used pervasively throughout io.js, yet it was
one of the most unwieldy functions in core.  This commit breaks up the
constructor into several small functions in a way that makes V8 happy.

About 8-10% CPU time was attributed to the constructor function before
in buffer-heavy benchmarks.  That pretty much drops to zero now because
V8 can now easily inline it at the call site.  It shortens the running
time of the following simple benchmark by about 15%:

    for (var i = 0; i < 25e6; ++i) new Buffer(1);

And about 8% from this benchmark:

    for (var i = 0; i < 1e7; ++i) new Buffer('x', 'ucs2');

PR-URL: https://github.com/iojs/io.js/pull/1048
Reviewed-By: Trevor Norris <trev.norris@gmail.com>

tty: do not add `shutdown` method to handle

UV_TTY does not support `uv_shutdown()` so adding this method in
StreamBase will cause an `abort()` in C land.

Fix: https://github.com/iojs/io.js/issues/1068
PR-URL: https://github.com/iojs/io.js/pull/1073
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>

net: allow port 0 in connect()

The added validation allows non-negative numbers and numeric
strings. All other values result in a thrown exception.

Fixes: https://github.com/joyent/node/issues/9194
PR-URL: https://github.com/joyent/node/pull/9268
Reviewed-By: Julien Gilli <julien.gilli@joyent.com>
Reviewed-By: Trevor Norris <trev.norris@gmail.com>
Reviewed-By: James M Snell <jasnell@users.noreply.github.com>

buffer: reword Buffer.concat error message

this brings the error messaging in line with
other node TypeError messages.

Fixes: https://github.com/joyent/node/issues/7766
PR-URL: https://github.com/joyent/node/pull/8723
Reviewed-By: James M Snell <jasnell@users.noreply.github.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>

win,test: fix test-stdin-from-file

The test-stdin-from-from-file test runs a subprocess that verifies stdin
can be piped from a file.

The subprocess additionally attempts to verify that the file descriptor
for stdin never gets closed. It used to do this by creating a TCP server
and asserting that the associated file descriptor is greater than two.
However this strategy doesn't work on windows, because servers don't
have an associated file descriptor. With this patch an ordinary file is
opened instead of creating a server.

PR: https://github.com/iojs/io.js/pull/1067
Reviewed-By: Fedor Indutny <fedor@indutny.com>
Reviewed-By: Rod Vagg <rod@vagg.org>
Reviewed-By: Nikolai Vavilov <vvnicholas@gmail.com>

fs: remove unnecessary usage of .hasOwnProperty()

PR-URL: https://github.com/iojs/io.js/pull/635
Reviewed-By: Vladimir Kurchatkin <vladimir.kurchatkin@gmail.com>

test: adds a test for undefined value in setHeader

As a result of 979d0ca8 there is a new check for undefined values on
OutgoingMessage.setHeader. This commit introduces a test for this case.

PR-URL: https://github.com/iojs/io.js/pull/970
Reviewed-By: Rod Vagg <rod@vagg.org>
Reviewed-By: Brendan Ashworth <brendan.ashworth@me.com>

benchmark: check for wrk ahead of running benchmarks

PR-URL: https://github.com/iojs/io.js/pull/982
Reviewed-By: Rod Vagg <rod@vagg.org>

build: remove tools/wrk from the tree

wrk is an optional tool that some of the http benchmarks uses. The removal
doesn't affect any users. Developers are assumed to install it before running
the tests.

This change reduces the tarball by 5%

PR-URL: https://github.com/iojs/io.js/pull/982
Reviewed-By: Rod Vagg <rod@vagg.org>

events: optimize listener array cloning

This both switches to a single algorithm for array cloning and also
speeds up (by ~100% in the ee-listeners-many benchmark) the
"many elements"  case that was previously handled by
`array.slice()`.

PR-URL: https://github.com/iojs/io.js/pull/1050
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Julian Duque <julianduquej@gmail.com>

test: split parts out of host-headers test into its own test

this makes the separation between http and https testing cleaner

PR-URL: https://github.com/iojs/io.js/pull/1049
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Shigeki Ohtsu <ohtsu@iij.ad.jp>

test: refactor all tests that depends on crypto

we had a few ways versions of looking for support before executing a test. this
commit unifies them as well as add the check for all tests that previously
lacked them. found by running `./configure --without-ssl && make test`. also,
produce tap skip output if the test is skipped.

PR-URL: https://github.com/iojs/io.js/pull/1049
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Shigeki Ohtsu <ohtsu@iij.ad.jp>

test: check for openssl cli and provide path if it exists

the previous version checked if io.js was compiled with openssl support which
isn't really relevant since we're starting a http server. we on the other hand
need an openssl-cli which may or may not exist.

PR-URL: https://github.com/iojs/io.js/pull/1049
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Shigeki Ohtsu <ohtsu@iij.ad.jp>

test: remove unused https imports

PR-URL: https://github.com/iojs/io.js/pull/1049
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Shigeki Ohtsu <ohtsu@iij.ad.jp>

test: introduce a helper that checks if crypto is available

since this applies to tls and https (among other things), it'll be used
for those tests as well. if we decouple the build system to somehow support
crypto but not tls, we could refine this.

PR-URL: https://github.com/iojs/io.js/pull/1049
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Shigeki Ohtsu <ohtsu@iij.ad.jp>

test: don't assume process.versions.openssl always is available

PR-URL: https://github.com/iojs/io.js/pull/1049
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Shigeki Ohtsu <ohtsu@iij.ad.jp>

buffer: add indexOf() method

Add Buffer#indexOf(). Support strings, numbers and other Buffers. Also
included docs and tests.

Special thanks to Sam Rijs <srijs@airpost.net> for first proposing this
change.

PR-URL: https://github.com/iojs/io.js/pull/561
Reviewed-by: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Brian White <mscdex@mscdex.net>
Reviewed-By: Chris Dickinson <christopher.s.dickinson@gmail.com>

url: throw for invalid values to url.format

`'use strict'` changes the behavior for `Function.prototype.call` when
the context is `undefined`. In earlier versions of node the value
`undefined` would make `url.format` look for fields in the global scope.

The docs states that `url.format` takes a parsed URL object and returns
a formatted URL string. So with this change it will now throw for other
values.

The exception is if the input is a string. Then it will call `url.parse`
on the string and then format it. The reason for that is that you can
call `url.format` on strings to clean up potentially wonky urls.

Fixes: https://github.com/iojs/io.js/issues/1033
PR-URL: https://github.com/iojs/io.js/pull/1036
Reviewed-By: Trevor Norris <trev.norris@gmail.com>
Reviewed-By: Brian White <mscdex@mscdex.net>
Reviewed-By: Julian Duque <julianduquej@gmail.com>

doc: add TC meeting 2015-02-18 minutes

PR-URL: https://github.com/iojs/io.js/pull/1051
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>

timer: Improve performance of callbacks

setImmediate, setTimeout, and setInterval were called in an inefficient
way, especially in the presence of arguments.  This optimization
improves their performance, with special cases for up to 4 arguments.
Performance of setImmediate increases with 35%, setInterval with 60%,
setTimeout with 70%.

PR-URL: https://github.com/iojs/io.js/pull/406
Reviewed-by: Trevor Norris <trev.norris@gmail.com>
Reviewed-by: Christian Tellnes <christian@tellnes.com>

doc: remove cjihrig from TC

PR-URL: https://github.com/iojs/io.js/pull/1056
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>

src: fix compiler warning in smalloc.cc

Fix the following compiler warning by static_casting the enum values
to an uint32_t:

    ../src/smalloc.cc: In function 'void
    node::smalloc::Initialize(v8::Handle<v8::Object>,
                              v8::Handle<v8::Value>,
                              v8::Handle<v8::Context>)':
    ../src/smalloc.cc:601:203: warning: enumeral and non-enumeral type
    in conditional expression
        EXTERNAL_ARRAY_TYPES(V)

PR-URL: https://github.com/iojs/io.js/pull/1055
Reviewed-By: Vladimir Kurchatkin <vladimir.kurchatkin@gmail.com>

smalloc: export constants from C++

PR-URL: https://github.com/iojs/io.js/pull/920
Reviewed-By: Chris Dickinson <christopher.s.dickinson@gmail.com>
Reviewed-By: Trevor Norris <trev.norris@gmail.com>

smalloc: validate arguments in js

PR-URL: https://github.com/iojs/io.js/pull/920
Reviewed-By: Chris Dickinson <christopher.s.dickinson@gmail.com>
Reviewed-By: Trevor Norris <trev.norris@gmail.com>

http: optimize outgoing requests

This commit does some small optimization changes on
`lib/_http_outgoing.js`. These include switching from `while` loops to
`for` loops, moving away from `util` to `typeof` checks, and removing
dead code. It also includes variable caches to avoid lookups and
generic style changes. All in all, much faster execution.

It gets an across the board increase in req/sec on the benchmarks,
from my experience about a 10% increase.

PR-URL: https://github.com/iojs/io.js/pull/605
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Nicu Micleușanu <micnic90@gmail.com>
Reviewed-By: Christian Vaagland Tellnes <christian@tellnes.com>
Reviewed-By: Brian White <mscdex@mscdex.net>

tls: accept empty `net.Socket`s

Accept `new net.Socket()` as a `socket` option to `tls.connect()`
without triggering an assertion error in C++.

This is done by wrapping it into a JSStream to ensure that there will be
a handle at the time of wrapping the socket into TLSSocket.

Fix: https://github.com/iojs/io.js/issues/987
PR-URL: https://github.com/iojs/io.js/pull/1046
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Rod Vagg <rod@vagg.org>

test: fix racey-ness in tls-inception

Fix test failure on FreeBSD and SmartOS, which happens due to a bad
timing:

    events.js:141
          throw er; // Unhandled 'error' event
                ^
    Error: read ECONNRESET
        at exports._errnoException (util.js:734:11)
        at TLSWrap.onread (net.js:538:26)

The outer `net.conncet()` socket stays alive after the inner socket is
gone. This happens because `.pipe()`'s implementation does not `destroy`
the source side when the destination has emitted `close`.

Fix: https://github.com/iojs/io.js/issues/1012
PR-URL: https://github.com/iojs/io.js/pull/1040
Reviewed-By: Chris Dickinson <christopher.s.dickinson@gmail.com>

node: improve performance of nextTick

Couple micro optimizations to improve performance of process.nextTick().
Removes ~60ns of execution time.

Also added small threshold to test that allows timer to fire early on
the order if microseconds.

PR-URL: https://github.com/iojs/io.js/pull/985
Reviewed-By: Vladimir Kurchatkin <vladimir.kurchatkin@gmail.com>

doc: fix child_process heading depth

Addition of the Sync/Async headers didn't indent all the sub-headers.

Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
PR-URL: https://github.com/iojs/io.js/pull/1038

test: fix test-fs-access when uid is 0

Superusers can open files with W_OK permission even though their
mode is set to 0444. This commit makes the test attempt to change
its uid to nobody on non-Windows platforms. Patch originally from
https://github.com/joyent/node/commit/28d0cbbd.

Fixes: https://github.com/iojs/io.js/issues/1031
PR-URL: https://github.com/iojs/io.js/pull/1037
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>

test: make destroyed-socket-write2.js more robust

test/simple/test-http-destroyed-socket-write2.js validates
that you get an appropriate error when trying to write to
a request when the response on the other side has been destroyed.

The test uses http.request to get a request and then keeps writing
to it until either it hits 128 writes or gets the expected error.
Since the writes are asynchronous we see that the writes just end
up adding events to the event loop, which then later get processed
once the connection supporting the request is fully ready.

The test is timing dependent and if takes too long for the connection
to be made the limit of 128 writes is exceeded and the test fails.
The fact that the test allows a number of writes is probably to allow
some delay for the connection to be ready for writing.

On AIX, in the default configuration using the loopback interface
is slower and the test fails because the delay is such that many
more writes can be queued up before the connection takes place.
If we use the host ip instead of defaulting to the loopback then
the test passes.

The test needs to be made more robust to delays. Since each write
simply enqueues an additional write to the event queue there is
probably no point in doing the second write until the first has
completed. This patch schedules the next write when the first one
completes and allows the test to pass even if it takes longer for
the connection to be ready for writing

PR-URL: https://github.com/joyent/node/pull/9270
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Timothy J Fontaine <tjfontaine@gmail.com>

doc: add explanations for querystring

 - add an article: `decode a non-utf8 string`
 - explain default and fallback behaviour of `querystring.unescape`

PR-URL: https://github.com/joyent/node/pull/9259
Reviewed-by: Julien Gilli <julien.gilli@joyent.com>

doc: fix default value of opts.decodeURIComponent

In the documentation for querystring.parse, the documentation mentions
that the default value for options.decodeURIComponent is the
decodeURIComponent function, but it's actually the querystring.unescape
function.

PR-URL: https://github.com/joyent/node/pull/9259
Reviewed-by: Julien Gilli <julien.gilli@joyent.com>

tests: fix race in test-http-curl-chunk-problem

This test setups two event listeners: one on a child process' exit event
, another for the same child process' stdandard output's 'data' event.
The data even listener writes to a stream, and the exit event listener
ends it.

Because the exit event can be emitted before the data event, there is a
chance that something will be written to the stream after it's ended,
and that an error is thrown.

This change makes the test end the stream in the listener for the child
process' standard output's end event, which is guaranteed to be emitted
after the last data event, thus avoiding the race.

PR: https://github.com/joyent/node/pull/9301
Reviewed-by: Bert Belder <bertbelder@gmail.com>
Reviewed-by: Michael Dawson <michael_dawson@ca.ibm.com>
Reviewed-by: Colin Ihrig <cjihrig@gmail.com>

doc: add missing newline in CHANGELOG

Pushed without review

Working on v1.4.4

2015-03-02 io.js v1.4.3 Release

Notable changes:

* stream: Fixed problems for platforms without `writev()` support,
  particularly Windows. Changes introduced in 1.4.1, via
  https://github.com/iojs/io.js/pull/926, broke some
  functionality for these platforms, this has now been addressed.
  https://github.com/iojs/io.js/pull/1008 (Fedor Indutny)
* arm: We have the very beginnings of ARMv8 / ARM64 / AARCH64
  support. An upgrade to OpenSSL 1.0.2 is one requirement for full
  support. https://github.com/iojs/io.js/pull/1028
  (Ben Noordhuis)
* Add new collaborator: Julian Duque @julianduque

build: add basic arm64 support

This commit adds basic arm64 support to the build.  Building the bundled
openssl is disabled pending an upgrade to openssl 1.2, the currently
bundled version has some hand-rolled assembly that is 32 bits only.

PR-URL: https://github.com/iojs/io.js/pull/1028
Reviewed-By: Johan Bergström <bugs@bergstroem.nu>
Reviewed-By: Rod Vagg <rod@vagg.org>

src: fix --without-ssl build

Don't include tls_wrap.h in stream_base.cc.  It's not used and it breaks
the build when --without-ssl is passed to configure.

Fixes the following build error:

    In file included from ../src/tls_wrap.h:5:0,
                     from ../src/stream_base.cc:10:
    ../src/node_crypto.h:20:25: fatal error: openssl/ssl.h:
    No such file or directory
     #include <openssl/ssl.h>

PR-URL: https://github.com/iojs/io.js/pull/1027
Reviewed-By: Johan Bergström <bugs@bergstroem.nu>

doc: update AUTHORS list

Update AUTHORS file using tools/update-authors.sh

PR-URL: https://github.com/iojs/io.js/pull/1018
Reviewed-By: Chris Dickinson <christopher.s.dickinson@gmail.com>

doc: add julianduque as collaborator

PR-URL: https://github.com/iojs/io.js/pull/1021
Reviewed-By: Evan Lucas <evanlucas@me.com>
Reviewed-By: Christian Tellnes <christian@tellnes.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Rod Vagg <r@va.gg>

doc: fix typos and sources in WORKING_GROUPS.md

`therefor` is a typo of `therefore`, and was fixed. There were also two
places where the website WG was directly linked, where they should have
put the WG's name/repo; that was fixed as well.

PR-URL: https://github.com/iojs/io.js/pull/1022
Reviewed-By: Mikeal Rogers <mikeal.rogers@gmail.com>
Reviewed-By: Christian Tellnes <christian@tellnes.no>

doc: Clean up net.Socket

net.Socket::connect(options[, connectListener]) was missing, with the
relevant details found on the net.connect function. I moved the
appropriate documentation over and then rewrote the documentation for
the function to say that it just creates a socket and calls the connect
method on it. I also changed the other net.Socket::connect variants to
say they are like the options version but called with a specific
options object.

net.connect and other functions were called methods even though they
don't use the `this` binding, so I changed method to function where
appropriate.

Finally, I added a missing period to the end of the module summary.
It's not really related to the rest of the changes, but benjamingr
noticed it.

PR-URL: https://github.com/iojs/io.js/pull/951
Reviewed-By: Sam Roberts <vieuxtech@gmail.com>
Reviewed-By: Christian Tellnes <christian@tellnes.no>

doc: suggest alternatives to deprecated APs

At the moment users who want to use `fs.exists` get a warning that the
method is deprecated but do not get offered an alternative in the page.
This PR suggests `fs.stat` and `fs.access` as alternatives while
keeping the warning about the use case in place.

Fixes: https://github.com/iojs/io.js/issues/1002
PR-URL: https://github.com/iojs/io.js/pull/1007
Reviewed-By: Brian White <mscdex@mscdex.net>
Reviewed-By: Christian Tellnes <christian@tellnes.no>

stream_base: `.writev()` has limited support

Only TCP and JSStream do support `.writev()` on all platforms at the
moment. Ensure that it won't be enabled everywhere.

Fix: https://github.com/iojs/io.js/issues/995
PR-URL: https://github.com/iojs/io.js/pull/1008
Reviewed-by: Bert Belder <bertbelder@gmail.com>
Reviewed-By: Rod Vagg <rod@vagg.org>

Working on v1.4.3

2015-02-28 io.js v1.4.2 Release

Notable changes:

* tls: A typo introduced in the TLSWrap changes in
  https://github.com/iojs/io.js/pull/840 only encountered as a bug on
  Windows was not caught by the io.js CI system due to problems with the
  Windows build script and the Windows CI slave configuration, see
  Fixed in https://github.com/iojs/io.js/pull/994 &
  https://github.com/iojs/io.js/pull/1004 (Fedor Indutny)
* npm: Upgrade npm to 2.6.1. See
  https://github.com/npm/npm/blob/master/CHANGELOG.md#v260-2015-02-12
  for details.
* Add new collaborators:
  - Robert Kowalski (@robertkowalski)
  - Christian Vaagland Tellnes (@tellnes)
  - Brian White (@mscdex)

tls: proxy `handle.reading` back to parent handle

Fix: https://github.com/iojs/io.js/issues/995
PR-URL: https://github.com/iojs/io.js/pull/1004
Reviewed-By: Rod Vagg <rod@vagg.org>

install: new performance counters provider guid

The io.js performance counters manifest conflicts with the one from
node, causing installer failures. Change the GUID in the io.js
performance counters manifest file to resolve the conflicts.

Fixes: https://github.com/iojs/io.js/issues/524
PR: https://github.com/iojs/io.js/pull/1001
Reviewed-by: Bert Belder <bertbelder@gmail.com>

tls: fix typo `handle._reading` => `handle.reading`

The problem does not manifest itself on unixes, because
`uv_read_start()` always return 0 there. However on Windows on a second
call `uv_read_start()` returns `UV__EALREADY` destroying all sockets on
a read attempt.

Set `.reading` property that is already handled by `net.js` code.

Fix: https://github.com/iojs/io.js/issues/988
PR-URL: https://github.com/iojs/io.js/pull/994
Reviewed-By: Chris Dickinson <christopher.s.dickinson@gmail.com>

http: don't confuse automatic headers for others

If you set a custom http header which includes eg. the string `Date`,
then http will not automatically send the `Date` header.

This is also true for other automatic http headers.

PR-URL: https://github.com/iojs/io.js/pull/828
Reviewed-By: Brendan Ashworth <brendan.ashworth@me.com>
Reviewed-By: Jeremiah Senkpiel <fishrock123@rocketmail.com>

build: improve vcbuild.bat

* Support running multiple test suites (e.g. vcbuild test-simple test-gc).
* Removed a nonexisting test suite (test-uv).
* Removed superfluous extra layer of argument parsing.
* Fix the node-weak build.

R=@rvagg

PR-URL: https://github.com/iojs/io.js/pull/998
Reviewed-By: Rod Vagg <rod@vagg.org>

build: reduce tarball size by 8-10%

Slim the tarballs further by removing examples, documentation and test for
third party libraries. Also switch to checkout-index versus archive so we avoid
using tar.

PR-URL: https://github.com/iojs/io.js/pull/961
Reviewed-By: Trevor Norris <trev.norris@gmail.com>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Rod Vagg <rod@vagg.org>

doc: minor formatting fixes.

PR-URL: https://github.com/iojs/io.js/pull/996
Reviewed-By: Rod Vagg <rod@vagg.org>

doc: update stability index

This simplifies the stability index to 4 levels:

0 - deprecated
1 - experimental / feature-flagged
2 - stable
3 - locked

Domains has been downgraded to deprecated, assert has been
downgraded to stable. Timers and Module remain locked. All
other APIs are now stable.

PR-URL: https://github.com/iojs/io.js/pull/943
Fixes: https://github.com/iojs/io.js/issues/930
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Reviewed-By: Jeremiah Senkpiel <fishrock123@rocketmail.com>
Reviewed-By: Vladimir Kurchatkin <vladimir.kurchatkin@gmail.com>

doc: add robertkowalski as collaborator

PR-URL: https://github.com/iojs/io.js/pull/977
Reviewed-By: Evan Lucas <evanlucas@me.com>
Reviewed-By: Christian Tellnes <christian@tellnes.com>
Reviewed-By: Rod Vagg <r@va.gg>