Skip to content

Conversation

mhinz
Copy link
Member

@mhinz mhinz commented May 4, 2017

This change implicitly adds IPv6 support.

If the address contains ":", we try to use a TCP socket instead of a Unix domain
socket. Everything in front of the last occurrence of ":" is the hostname and
everything after it the port.

If the hostname lookup fails, we fall back to using a Unix domain socket.

If the port is empty ("localhost:"), a random port will be assigned.

Examples:

  NVIM_LISTEN_ADDRESS=localhost:12345  -> TCP (IPv4 or IPv6), port: 12345
  NVIM_LISTEN_ADDRESS=localhost:       -> TCP (IPv4 or IPv6), port: random (> 1024)
  NVIM_LISTEN_ADDRESS=localhost        -> Unix domain socket "localhost" in current dir

@mhinz mhinz requested review from tarruda and removed request for tarruda May 4, 2017 16:08
@jamessan
Copy link
Member

jamessan commented May 4, 2017

Should we be using uv_getaddrinfo instead, which would also support IPv6?

@mhinz
Copy link
Member Author

mhinz commented May 4, 2017

I look into it!

@oni-link
Copy link
Contributor

oni-link commented May 4, 2017

See function socket_connect() in #6594 (https://github.com/neovim/neovim/pull/6594/files#diff-6d5b2f9d654e4e24ddb9a960da02581eR182), it already uses uv_getaddrinfo() for address lookup.

@mhinz
Copy link
Member Author

mhinz commented May 5, 2017

@oni-link Thanks for the pointer!

Nevertheless, I think this will take more time than I'm willed to invest at the moment.

Does someone want to take over? Maybe @bfredl? Seems like a good addition for the already mentioned #6594.

@mhinz mhinz changed the title [RFC] Allow NVIM_LISTEN_ADDRESS=localhost:12345 [RFC] Use getaddrinfo() for $NVIM_LISTEN_ADDRESS May 8, 2017
@mhinz mhinz force-pushed the listen/localhost branch from d5102a9 to d49243b Compare May 8, 2017 12:11
for (; ai; ai = ai->ai_next) {
memcpy(&watcher->uv.tcp.addr, ai->ai_addr, ai->ai_addrlen);
result = uv_tcp_bind(&watcher->uv.tcp.handle,
(const struct sockaddr *)&watcher->uv.tcp.addr, 0);
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

ai->ai_addr can be used here.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Right, I guess it would be more explicit to use watcher since it's also used for all the following calls.. but using ai->ai_addr I can avoid the cast and it just looks better.

👍

@@ -90,14 +71,38 @@ int socket_watcher_start(SocketWatcher *watcher, int backlog, socket_cb cb)
int result;

if (watcher->stream->type == UV_TCP) {
result = uv_tcp_bind(&watcher->uv.tcp.handle,
(const struct sockaddr *)&watcher->uv.tcp.addr, 0);
struct addrinfo *ai = watcher->uv.tcp.addrinfo;
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If ai is NULL then result is not initialized.

Copy link
Member Author

@mhinz mhinz May 8, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Oops. I'll fix that.

.oO(Actually it shouldn't happen unless getaddrinfo() can return 0 and the resulting pointer to the addrinfo structures is NULL nevertheless.)

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Since libuv is expected to return negative integers, I now initialise it with -EINVAL.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@mhinz This should be UV_EINVAL without -, not -EINVAL. According to C99 only EDOM, EILSEQ and ERANGE must be defined, no EINVAL in this list.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Also on Windows UV_EINVAL and -EINVAL are going to be different AFAIK. And you return UV errors in other places. Also UV_EACCESS and not -EACCESS below, UV_ENOENT and not -ENOENT.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

👍

&(int){sizeof(struct sockaddr_storage)});
uint16_t port = (ai->ai_family == AF_INET)
? ((struct sockaddr_in *)&watcher->uv.tcp.addr)->sin_port
: ((struct sockaddr_in6 *)&watcher->uv.tcp.addr)->sin6_port;
Copy link
Contributor

@oni-link oni-link May 8, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

These numbers are encoded in network byte order, what we want is host byte order, so we need to call ntohs() on sin_port/sin6_port.
ntohs() is called later.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

ntohs() is used when port is used. I wanted to avoid wrapping this already huge assignment and using an extra assignment à la port = ntohs(port).

@mhinz mhinz force-pushed the listen/localhost branch from d49243b to 9472c0c Compare May 9, 2017 08:57
// number, a free random port number will be assigned. sin_port will
// contain 0 in this case, unless uv_tcp_getsockname() is used first.
uv_tcp_getsockname(&watcher->uv.tcp.handle,
(struct sockaddr *)&watcher->uv.tcp.addr,
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why not use a local struct sockaddr_storage x here? The content of x is not used in this call. Member watcher->uv.tcp.addr could be removed.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

👍

: ((struct sockaddr_in6 *)&watcher->uv.tcp.addr)->sin6_port;
// v:servername uses the string from watcher->addr
snprintf(watcher->addr, sizeof(watcher->addr), "%s:%" PRIu16,
watcher->addr, ntohs(port));
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hostname is not showing for me. Could be that we reading and writing from the same buffer.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hmm. Okay, I'll make a copy of watcher->addr first.

struct addrinfo *ai = watcher->uv.tcp.addrinfo;

for (; ai; ai = ai->ai_next) {
memcpy(&watcher->uv.tcp.addr, ai->ai_addr, ai->ai_addrlen);
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This could be removed if we use a local variable, see below.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

👍

struct addrinfo *ai = watcher->uv.tcp.addrinfo;

for (; ai; ai = ai->ai_next) {
struct sockaddr_storage sas = *(struct sockaddr_storage *)ai->ai_addr;
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This should be moved into the if (result == 0) {...} block without initialization (uv_tcp_getsockname() will fill it). Dereferencing ai->ai_addr can go wrong if the structure is smaller than struct sockaddr_storage.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Dereferencing ai->ai_addr can go wrong if the structure is smaller than struct sockaddr_storage.

How would I work around that? A union with sockaddr_in and sockaddr_in6 and switching on ai_family?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Just put struct sockaddr_storage sas; before uv_tcp_getsockname(). We get the socket name from the socket descriptor, not from the content of sas.

// v:servername uses the string from watcher->addr
char addr[sizeof(watcher->addr)];
strcpy(addr, watcher->addr);
snprintf(watcher->addr, sizeof(addr), "%s:%" PRIu16, addr, ntohs(port));
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Could also be done with a strlen():

size_t len = strlen(watcher->addr);
snprintf(watcher->addr +len, sizeof(watcher->addr)-len, ":%" PRIu16, ntohs(port));

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

👍

void socket_watcher_init(Loop *loop, SocketWatcher *watcher,
const char *endpoint, void *data)
void socket_watcher_init(Loop *loop, SocketWatcher *watcher, const char *endpoint,
void *data)
FUNC_ATTR_NONNULL_ARG(1) FUNC_ATTR_NONNULL_ARG(2) FUNC_ATTR_NONNULL_ARG(3)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We can use FUNC_ATTR_NONNULL_ARG(1,2,3) here.

void socket_watcher_init(Loop *loop, SocketWatcher *watcher,
const char *endpoint, void *data)
void socket_watcher_init(Loop *loop, SocketWatcher *watcher, const char *endpoint,
void *data)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Are we missing a watcher->data = data?

Copy link
Member Author

@mhinz mhinz May 9, 2017

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can we just remove data from the arguments? It's always called with NULL and only in server_start(). Maybe the original author planned ahead back then.

if (uv_ip4_addr(ip, port, &watcher->uv.tcp.addr)) {
// Invalid address, treat as named pipe or unix socket
tcp = false;
char *addr = xstrdup(endpoint);
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We could copy endpoint directly into watcher->addr and work with that string:

diff --git a/src/nvim/event/socket.c b/src/nvim/event/socket.c
index 48d4c3716..c254b8077 100644
--- a/src/nvim/event/socket.c
+++ b/src/nvim/event/socket.c
@@ -25,12 +25,12 @@ void socket_watcher_init(Loop *loop, SocketWatcher *watcher, const char *endpoin
                          void *data)
   FUNC_ATTR_NONNULL_ARG(1) FUNC_ATTR_NONNULL_ARG(2) FUNC_ATTR_NONNULL_ARG(3)
 {
-  char *addr = xstrdup(endpoint);
+  xstrlcpy(watcher->addr, endpoint, sizeof(watcher->addr));
+  char *addr = watcher->addr;
   char *host_end = strrchr(addr, ':');
 
   if (host_end && addr != host_end) {
     *host_end = '\0';
-    xstrlcpy(watcher->addr, addr, sizeof(watcher->addr));
 
     uv_getaddrinfo_t request;
     int retval = uv_getaddrinfo(&loop->uv, &request, NULL, addr, host_end+1,
@@ -40,6 +40,7 @@ void socket_watcher_init(Loop *loop, SocketWatcher *watcher, const char *endpoin
                                 });
     if (retval != 0) {
       // Failed to look up address.
+      *host_end = ':';
       goto do_pipe;
     }
     watcher->uv.tcp.addrinfo = request.addrinfo;
@@ -48,7 +49,6 @@ void socket_watcher_init(Loop *loop, SocketWatcher *watcher, const char *endpoin
     watcher->stream = (uv_stream_t *)&watcher->uv.tcp.handle;
   } else {
 do_pipe:
-    xstrlcpy(watcher->addr, endpoint, sizeof(watcher->addr));
     uv_pipe_init(&loop->uv, &watcher->uv.pipe.handle, 0);
     watcher->stream = (uv_stream_t *)&watcher->uv.pipe.handle;
   }
@@ -57,8 +57,6 @@ do_pipe:
   watcher->cb = NULL;
   watcher->close_cb = NULL;
   watcher->events = NULL;
-
-  xfree(addr);
 }
 
 int socket_watcher_start(SocketWatcher *watcher, int backlog, socket_cb cb)

}
result = uv_listen(watcher->stream, backlog, connection_cb);
if (result == 0) {
struct sockaddr_storage sas = *(struct sockaddr_storage *)ai->ai_addr;
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should be struct sockaddr_storage sas;.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Oh. Ooooh. Sorry, now I get what you meant earlier. I'll fix that.

@mhinz mhinz force-pushed the listen/localhost branch from 86dd7af to be8f20f Compare May 9, 2017 15:33
@oni-link
Copy link
Contributor

oni-link commented May 9, 2017

With :echo serverstart("127.0.0.1:12345678901") we have pipes in serverlist()
that kind of look like IP addresses. Is that okay?

With :echo serverstart("::1:1234567") we have tcp addresses in serverlist() that have
a different socket address (::1:54919). Probably a result of how getaddrinfo() transforms our port string into a number (1234567 % 2^16 = 54919).

@mhinz
Copy link
Member Author

mhinz commented May 9, 2017

With :echo serverstart("::1:1234567") we have tcp addresses in serverlist() that have
a different socket address (::1:54919). Probably a result of how getaddrinfo() transforms our port string into a number (1234567 % 2^16 = 54919).

:echo serverstart("::1:1234567") creates a Unix socket for me, as result of falling back when getaddrinfo() failed.

@oni-link
Copy link
Contributor

oni-link commented May 9, 2017

@mhinz, it creates a tcp socket for me. uv_getaddrinfo() does not fail. Using nc ::1 54919 -6 and sending some text, I get an answer (nc ::1 1234567 -6 also works for me).

@mhinz
Copy link
Member Author

mhinz commented May 9, 2017

@oni-link Hopefully it works as expected now.

EDIT: Hmm, strtol() returns 0 if it fails.. it would be nice to support host:0 as synonym for host:, thus getting a random port assigned.

errno gets set to EINVAL, but according to my manpage that's not portable.

I replaced strtol() by getdigits_safe().

@mhinz mhinz force-pushed the listen/localhost branch 2 times, most recently from 3e93f3c to ca39453 Compare May 9, 2017 20:21
@oni-link
Copy link
Contributor

oni-link commented May 9, 2017

@mhinz, port range check works.

host:0 and host: both choose a random port on my system.
With 0.0.0.0: and 0.0.0.0:0 the socket will listen on any address and picks a random port.
An empty string seems to result in a pipe with no name. Is this useful?

@mhinz
Copy link
Member Author

mhinz commented May 9, 2017

You mean ' ' instead of a real empty string '', right? The latter throws an exception for me.

It would be weird, but at least on Unix it should be no problem. I'd blame the user in that case. :> And current master acts the same, so I'd say it's fine for now.

@mhinz mhinz force-pushed the listen/localhost branch from ad60f05 to b832986 Compare May 21, 2017 23:09
@jamessan
Copy link
Member

jamessan commented May 22, 2017

The Linux QB failure is

11:42:41,907 WARN  - /home/quickbuild/buildagent/workspace/root/neovim/pull-requests-automated/src/nvim/event/socket.c: In function ‘socket_watcher_start’:
11:42:41,907 WARN  - /home/quickbuild/buildagent/workspace/root/neovim/pull-requests-automated/src/nvim/event/socket.c:102:11: error: conversion to ‘uint16_t’ from ‘int’ may alter its value [-Werror=conversion]

which doesn't make sense because both sin6_port and sin_port should be the same type -- in_port_t. Everywhere I've checked, that is a typedef to uint16_t. However, this is happening on Ubuntu 12.04, which is EOL, so I can't see what version of glibc it was using.

The FreeBSD QB failure is

    funcs.serverstart('127.0.0.1:')  -- assign random port
    assert(string.match(funcs.serverlist()[1], '127.0.0.1:%d+'))

on the assert:

19:17:33,940 INFO  - # test/functional/eval/server_spec.lua @ 61
19:17:33,940 INFO  - # Failure message: test/functional/eval/server_spec.lua:70: bad argument #1 to 'match' (string expected, got nil)
19:17:33,940 INFO  - # stack traceback:
19:17:33,940 INFO  - #     test/functional/eval/server_spec.lua:70: in function <test/functional/eval/server_spec.lua:61>

It'd be interesting to see what serverlist() is actually returning.

everything after it the port. If the port is empty or `0`,
a random port will be assigned.

If no address is given, it is equivalent to: >
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

is this true also on windows (pipes have a disjoint namespace from regular files)?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I have no idea. I trust the person who wrote it in the first place.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I wouldn't be so trusting :P

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For Windows the special name format \\.\pipe\nvim-<PID>-<COUNTER> is used.

uv_getaddrinfo_t request;

int retval = uv_getaddrinfo(&loop->uv, &request, NULL, addr, port,
&(struct addrinfo){
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

why not declare hints as a separate var?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Well, there's no need to keep it, so I used a throwaway compound literal.

@mhinz mhinz force-pushed the listen/localhost branch from b832986 to 156e6f2 Compare May 22, 2017 21:38
@mhinz
Copy link
Member Author

mhinz commented May 22, 2017

It'd be interesting to see what serverlist() is actually returning.

Hmm, if serverlist()[1] is nil, then serverlist() probably returned {}, right?

So serverstart('127.0.0.1:') failed whereas serverstart('127.0.0.1:0') succeeded?

@bfredl
Copy link
Member

bfredl commented May 24, 2017

Wouldn't it be useful if serverstart("localhost:") returned the actual used address, i e "localhost:46565" ?

bfredl pushed a commit to bfredl/neovim that referenced this pull request May 25, 2017
Server: use uv_getaddrinfo() for $NVIM_LISTEN_ADDRESS

This change implicitly adds IPv6 support.

If the address contains ":", we try to use a TCP socket instead of a Unix domain
socket. Everything in front of the last occurrence of ":" is the hostname and
everything after it the port.

If the hostname lookup fails, we fall back to using a Unix domain socket.

If the port is empty ("localhost:"), a random port will be assigned.

Examples:

  NVIM_LISTEN_ADDRESS=localhost:12345 -> TCP (IPv4 or IPv6), port: 12345
  NVIM_LISTEN_ADDRESS=localhost:      -> TCP (IPv4 or IPv6), port: random (> 1024)
  NVIM_LISTEN_ADDRESS=localhost:0     -> TCP (IPv4 or IPv6), port: random (> 1024)
  NVIM_LISTEN_ADDRESS=localhost       -> Unix domain socket "localhost" in current dir

Server tests: use helpers.command()

Server tests: endpoint parsing in serverstart()

Server: don't fall back to Unix sockets

Doc: explain the format for serverstart()
@mhinz
Copy link
Member Author

mhinz commented May 27, 2017

@bfredl Absolutely. In a follow-up PR. :-)

@jamessan In that case I'll just ignore the Linux QB issue.

What to do about the FreeBSD QB issue? Should I just put logging into the test until we can figure it out? Is anyone using FreeBSD by any chance?

@bfredl
Copy link
Member

bfredl commented May 27, 2017

Why? You already calculate the correct value for serverlist, why not return it already?

When using serverstart("ip.ad.d.r:") to listen on a random port, we need
to abide by getaddrinfo()'s API and pass in a NULL service, rather than
an empty string.

When given an empty string, getaddrinfo() is free to search for a
service by the given name (since the string isn't a number) which will
fail.  At least FreeBSD does perform this lookup.
@jamessan
Copy link
Member

Hmm, while testing the FreeBSD fix I ran into this:

==5748==ERROR: LeakSanitizer: detected memory leaks

Direct leak of 2 byte(s) in 1 object(s) allocated from:
    #0 0x4ecdc8 in malloc (/home/jamessan/src/github.com/neovim/build/bin/nvim+0x4ecdc8)
    #1 0xf828e4 in try_malloc /home/jamessan/src/github.com/neovim/src/nvim/memory.c:87:15
    #2 0xf82aa4 in xmalloc /home/jamessan/src/github.com/neovim/src/nvim/memory.c:121:15
    #3 0xf82f43 in xmallocz /home/jamessan/src/github.com/neovim/src/nvim/memory.c:196:15
    #4 0xf83038 in xmemdupz /home/jamessan/src/github.com/neovim/src/nvim/memory.c:214:17
    #5 0xf83e92 in xstrdup /home/jamessan/src/github.com/neovim/src/nvim/memory.c:429:10
    #6 0x175865d in vim_strsave /home/jamessan/src/github.com/neovim/src/nvim/strings.c:59:20
    #7 0xb2bdd9 in do_cmdline /home/jamessan/src/github.com/neovim/src/nvim/ex_docmd.c:553:22
    #8 0xb32b65 in do_cmdline_cmd /home/jamessan/src/github.com/neovim/src/nvim/ex_docmd.c:279:10
    #9 0xe6f0f1 in exe_pre_commands /home/jamessan/src/github.com/neovim/src/nvim/main.c:1607:7
    #10 0xe60ee8 in main /home/jamessan/src/github.com/neovim/src/nvim/main.c:335:3
    #11 0x7f74cce1a2b0 in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x202b0)

SUMMARY: AddressSanitizer: 2 byte(s) leaked in 1 allocation(s).
+++ exited with 1 +++

by running ./build/bin/nvim -u NONE -i NONE -n --cmd 'call serverstart("127.0.0.1:")' --cmd q. I tried a few other two "pre_command" command lines, but none of those leaked. I can reproduce it on master using serverstart()...

@jamessan
Copy link
Member

In that case I'll just ignore the Linux QB issue.

Unfortunately, that would mean QB is red for everyone until we get off of that Ubuntu release. @jszakmeister is there any chance the QB Linux systems could be upgraded to at least Ubuntu 14.04 (Trusty)?

In the meantime, I've added a cast which should silence this.

What to do about the FreeBSD QB issue?

I've pushed a fix for that.

Wouldn't it be useful if serverstart("localhost:") returned the actual used address, i e "localhost:46565"?

Agreed. I've also pushed a commit for this.

@@ -45,6 +45,12 @@ int socket_watcher_init(Loop *loop, SocketWatcher *watcher,
return UV_EINVAL;
}

if (*port == NUL) {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should this also be used if iport is 0 or is "0" always an acceptable port description?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

A port that looks like a number is used directly, otherwise a lookup is performed. We should be fine here.

src/nvim/eval.c Outdated
// "localhost:" will now have a port), return the final value to the user.
size_t n;
char **addrs = server_address_list(&n);
address = addrs[n - 1];
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why use address instead of rettv->vval.v_string?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Fixed.

@jszakmeister
Copy link
Contributor

@jszakmeister is there any chance the QB Linux systems could be upgraded to at least Ubuntu 14.04 (Trusty)?

@jamessan Yes I can. The real question, which has never been answered though, is what is our support policy? And I guess one of the additional headaches here is "what do we want to ensure we run against" vs. "what do we want to build against with no warnings"?

Let me know what you guys want here (and it's generally best to do this via email directly to me--notifications from GitHub have a tendency to get lost in the noise).

jamessan added 2 commits May 28, 2017 07:14
In the process of setting up the socket watcher, the address may be
changed (e.g., adding the OS-selected port).
Although in_port_t is a typedef for uint16_t, GCC in Ubuntu 12.04
complains about potential loss of data due to converting int to
uint16_t.  Since we know this isn't possible, silence the warning to
avoid breaking QB until it gets upgraded to a newer Ubuntu.
@jamessan jamessan force-pushed the listen/localhost branch from 872fab2 to 62d020a Compare May 28, 2017 11:17
@jamessan jamessan merged commit 9cc185d into neovim:master May 28, 2017
@mhinz mhinz deleted the listen/localhost branch May 31, 2017 11:56
justinmk added a commit to justinmk/neovim that referenced this pull request Nov 8, 2017
FEATURES:
0e873a3 Lua(Jit) built-in neovim#4411
5b32bce Windows: `:terminal` neovim#7007
7b0ceb3 UI/API: externalize cmdline neovim#7173
b67f58b UI/API: externalize wildmenu neovim#7454
b23aa1c UI: 'winhighlight' neovim#6597
17531ed UI: command-line coloring (`:help input()-highlight`) neovim#6364
244a1f9 API: execute lua directly from the remote api neovim#6704
45626de API: `get_keymap()` neovim#6236
db99982 API: `nvim_get_hl_by_name()`, `nvim_get_hl_by_id()` neovim#7082
dc68538 menu_get() function neovim#6322
9db42d4 :cquit : take an error code argument neovim#7336
9cc185d job-control: serverstart(): support ipv6 neovim#6680
1b7a9bf job-control: sockopen() neovim#6594
6efe84a clipboard: fallback to tmux clipboard neovim#6894
6016ac2 clipboard: customize clipboard with `g:clipboard` neovim#6030
3a86dd5 ruby: override ruby host via `g:ruby_host_prog` neovim#6841
16cce1a debug: $NVIM_LOG_FILE neovim#6827
0cba3da `:checkhealth` built-in, validates $VIMRUNTIME neovim#7399

FIXES:
105d680 TUI: more terminals, improve scroll/resize neovim#6816
cb912a3 :terminal : handle F1-F12, other keys neovim#7241
619838f inccommand: improve performance neovim#6949
04b3c32 inccommand: Fix matches for zero-width neovim#7487
60b1e8a inccommand: multiline, other fixes neovim#7315
f1f7f3b inccommand: Ignore leading modifiers in the command neovim#6967
1551f71 inccommand: fix 'gdefault' lockup neovim#7262
6338199 API: bufhl: support creating new groups neovim#7414
541dde3 API: allow K_EVENT during operator-pending
8c732f7 terminal: adjust for 'number' neovim#7440
5bec946 UI: preserve wildmenu during jobs/events neovim#7110
c349083 UI: disable 'lazyredraw' during ui_refresh. neovim#6259
51808a2 send FocusGained/FocusLost event instead of pseudokey neovim#7221
133f8bc shada: preserve unnamed register on restart neovim#4700
1b70a1d shada: avoid assertion on corrupt shada file neovim#6958
9f534f3 mksession: Restore tab-local working directory neovim#6859
de1084f fix buf_write() crash neovim#7140
7f76986 syntax: register 'Normal' highlight group neovim#6973
6e7a8c3 RPC: close channel if stream was closed neovim#7081
85f3084 clipboard: disallow recursion; show hint only once neovim#7203
8d1ccb6 clipboard: performance, avoid weird edge-cases neovim#7193
01487d4 'titleold' neovim#7358
01e53a5 Windows: better path-handling, separator (slash) hygiene neovim#7349
0f2873c Windows: multibyte startup arguments neovim#7060

CHANGES:
9ff0cc7 :terminal : start in normal-mode neovim#6808
032b088 lower priority of 'cursorcolumn', 'colorcolumn' neovim#7364
2a3bcd1 RPC: Don't delay notifications when request is pending neovim#6544
023f67c :terminal : Do not change 'number', 'relativenumber' neovim#6796
1ef2d76 socket.c: Disable Nagle's algorithm on TCP sockets neovim#6915
6720fe2 help: `K` tries Vim help instead of manpage neovim#3104
7068370 help, man.vim: change "outline" map to `gO` neovim#7405
@justinmk justinmk mentioned this pull request Nov 8, 2017
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

7 participants