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>
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>
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>
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>
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>
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>
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>
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>
* 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>
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>
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>
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>
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>
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>
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>
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>
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>
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>
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>
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>
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>
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>
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>
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>
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>
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>
`'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>
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>
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>
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>
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>
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>
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>
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/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>
Fedor Indutny
Brian White
Johan Bergström
Christian Tellnes
Ben Noordhuis
Petka Antonov
cjihrig
Chris Dickinson
Bert Belder
Jonathan Ong
Ken Perkins
Trevor Norris
Rod Vagg
Ruben Verborgh
Vladimir Kurchatkin
Brendan Ashworth
Sam Roberts
Michael Dawson