win: fix buffer leak on error in pipe.c

stream: do not leak `alloc_cb` buffers on error

Always pass previously allocated (if any) buffer, when calling
`read_cb()` on error.

unix: fix setting written size on uv_wd

Now working on v0.11.23

2014.03.11, Version 0.11.22 (Unstable)

Changes since version 0.11.21:

* unix, windows: map ERANGE errno (Saúl Ibarra Corretgé)

* unix, windows: make uv_cwd be consistent wi

2014.03.11, Version 0.11.22 (Unstable)

Changes since version 0.11.21:

* unix, windows: map ERANGE errno (Saúl Ibarra Corretgé)

* unix, windows: make uv_cwd be consistent with uv_exepath (Saúl Ibarra
Corretgé)

* process: remove debug perror() prints (Fedor Indutny)

* windows: fall back for volume info query (Isaiah Norton)

* pipe: allow queueing pending handles (Fedor Indutny)

* windows: fix winsock status codes for address errors (Raul Martins)

* windows: Remove unused variable from uv__pipe_insert_pending_socket
(David Capello)

* unix: workaround broken pthread_sigmask on Android (Paul Tan)

* error: add ENXIO for O_NONBLOCK FIFO open() (Fedor Indutny)

* freebsd: use accept4, introduced in version 10 (Saúl Ibarra Corretgé)

* windows: fix warnings of MinGW -Wall -O3 (StarWing)

* openbsd, osx: fix compilation warning on scandir (Saúl Ibarra
Corretgé)

* linux: always deregister closing fds from epoll (Geoffry Song)

* unix: reopen tty as /dev/tty (Saúl Ibarra Corretgé)

* kqueue: invalidate fd in uv_fs_event_t (Fedor Indutny)

show more ...

Merge branch 'v0.10'

Conflicts:
test/test-spawn.c

kqueue: invalidate fd in uv_fs_event_t

Invalidate file descriptor when closing `uv_fs_event_t` handle. Note
that `uv__io_stop` is just removing `fd` from `loop->watchers` and not
act

kqueue: invalidate fd in uv_fs_event_t

Invalidate file descriptor when closing `uv_fs_event_t` handle. Note
that `uv__io_stop` is just removing `fd` from `loop->watchers` and not
actually invalidating all consequent events in a `kevent()` results.

fix joyent/node#1101

show more ...

unix: reopen tty as /dev/tty

Reopen the file descriptor when it refers to a tty. This lets us put the
tty in non-blocking mode without affecting other processes that share it
with us

unix: reopen tty as /dev/tty

Reopen the file descriptor when it refers to a tty. This lets us put the
tty in non-blocking mode without affecting other processes that share it
with us.

This brings back commit 31f9fbc, which was reverted in 20bb1bf. The OSX
select trick is working now.

Original patch by @bnoordhuis

show more ...

linux: always deregister closing fds from epoll

If the same file description is open in two different processes, then
closing the file descriptor is not sufficient to deregister it from

linux: always deregister closing fds from epoll

If the same file description is open in two different processes, then
closing the file descriptor is not sufficient to deregister it from the
epoll instance (as described in epoll(7)), resulting in spurious events
that cause the event loop to spin repeatedly. So always explicitly
deregister it.

Fixes #1099.

Conflicts:
test/test-spawn.c

show more ...

openbsd, osx: fix compilation warning on scandir

The select function takes a const struct on newer OSX versions but it
doesn't on OSX <= 10.7 or OpenBSD.

windows: fix warnings of MinGW -Wall -O3

build: allow CC to be overridden on Makefile.mingw

build: remove unneeded EXTRA_CFLAGS from Makefile.mingw

freebsd: use accept4, introduced in version 10

Merge remote-tracking branch 'origin/v0.10'

Conflicts:
AUTHORS
ChangeLog
include/uv.h
src/unix/error.c
src/unix/process.c

Merge remote-tracking branch 'origin/v0.10'

Conflicts:
AUTHORS
ChangeLog
include/uv.h
src/unix/error.c
src/unix/process.c
src/version.c

show more ...

error: add ENXIO for O_NONBLOCK FIFO open()

When opening FIFO with `O_NONBLOCK` flag, `ENXIO` could be returned if
the readable side hasn't yet opened this FIFO.

unix: workaround broken pthread_sigmask on Android

On android, if a call to `pthread_sigmask` fails with EINVAL, fallback
to `sigprocmask`. This workarounds the broken behaviour of
`

unix: workaround broken pthread_sigmask on Android

On android, if a call to `pthread_sigmask` fails with EINVAL, fallback
to `sigprocmask`. This workarounds the broken behaviour of
`pthread_sigmask` on Android versions < 4.1. See:

* https://android.googlesource.com/platform/bionic/+/9bf330b5
* https://code.google.com/p/android/issues/detail?id=15337

This is implemented by replacing all calls of `pthread_sigmask` to a new
function `uv__pthread_sigmask` which implements this fallback mechanism.

show more ...

windows: Remove unused variable from uv__pipe_insert_pending_socket

windows: fix winsock status codes for address errors

- STATUS_TOO_MANY_ADDRESSES now returns WSAEADDRINUSE
- STATUS_CONFLICTING_ADDRESSES now returns WSAEADDRNOTAVAIL

test: fix compiation warning

-Wunused-function would kick in because of the close_loop function,
but there tests and benchmarks which actually don't use any loop.

pipe: allow queueing pending handles

Introduce `int uv_pipe_pending_count(uv_pipe_t*)` and
`uv_handle_type uv_pipe_pending_type(uv_pipe_t*)`. They should be
used in IPC pipe's read c

pipe: allow queueing pending handles

Introduce `int uv_pipe_pending_count(uv_pipe_t*)` and
`uv_handle_type uv_pipe_pending_type(uv_pipe_t*)`. They should be
used in IPC pipe's read cb to accept incoming handles:

int count = uv_pipe_pending_count(pipe);
int i;
for (i = 0; i < count; i++) {
uv_handle_type type = uv_pipe_pending_type(pipe);
/* ... */
uv_accept(...);
}

show more ...

windows: fall back for volume info query

Wine does not currently support FileFsVolumeInformation:

https://github.com/mirrors/wine/blob/0e42fd97c0/dlls/ntdll/file.c#L2679

So

windows: fall back for volume info query

Wine does not currently support FileFsVolumeInformation:

https://github.com/mirrors/wine/blob/0e42fd97c0/dlls/ntdll/file.c#L2679

So check io_status and fall back to previous behavior if not
implemented.

show more ...

process: remove debug perror() prints

fix #1128

build: add required libraries to libuv.pc.in

build: fix configure when PKG_CONFIG is set

If you define the environment variable PKG_CONFIG and assign it to
pkgconf (or pkg-config), configure will say it's OK, but it won't be
us

build: fix configure when PKG_CONFIG is set

If you define the environment variable PKG_CONFIG and assign it to
pkgconf (or pkg-config), configure will say it's OK, but it won't be
used.

show more ...