Asynchronous networking for C

I like having 0 warnings in a build....

I'm pretty confident with the _CRT_SECURE_NO_WARNINGS guard, which allows for specification of that directive outside, without creating a warning. I'm not sure if the second change can be more elegant.

Tested on Windows and Mac.

Have to call WSAAddressToStringA (non-Unicode version) manually, otherwise Visual Studio redefines it to call the wide version, causing a buffer overrun and a corrupted address string.

Kit is a project manager for C, and with a few simple changes, dyad gains compatibility as an includable modules.

These changes do not effect the codebase - they just do some restructuring to meet Kit conventions.

I'm not 100% about other platforms, but this gets rid of some MSVC 64-bit warnings. A quick google suggests that I'm casting these vars to what the function expects, which is happening anyway. This contribution only removes warnings -- the original writer of the code should judge whether the intent matches the implementation. If this code was always correct, this PR is solidifying that position. If this code was always wrong, I guess I'm prompting for a fix.

Listening sockets are created with the AF_UNSPEC family flag, this causes an IPv6 socket to be returned on Windows systems supporting them. As Windows has completely separate networking stacks for IPv4 and IPv6, the end result is that such sockets won't accept IPv4 connections, including local connections via loopback (127.0.0.1, or localhost which maps to the same).

The workaround I came up with was to pass in AF_INET for the socket family in dyad_listenEx():

  /* Get addrinfo */
  memset(&hints, 0, sizeof(hints));
  hints.ai_family = AF_INET;
  hints.ai_socktype = SOCK_STREAM;
  hints.ai_flags = AI_PASSIVE;
  sprintf(buf, "%d", port);
  err = getaddrinfo(host, buf, &hints, &ai);

However, this leads to the opposite problem: IPv6 clients can't connect. Would there a way to support both protocols for listening sockets?

Seems like httpserv and httpserv2 examples did not works properly after successful compilation with cygwin/mingw gcc, at least in my windows environment.

Thehttpget works as expected, but httpserv output only a server listening on port 8000 and then have no ractions on external requests (brandmauer switched off). I checked different ports, additional logs in events listeners - seems like only onListen catched.

There some compiled binaries for test. VirusTotal complains a bit about binaries compiled with mingw-gcc, so I put only cygwin ones, results similar.

system: W7 x64

cygwin-gcc:

Using built-in specs.
COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=/usr/lib/gcc/i686-pc-cygwin/4.8.2/lto-wrapper.exe
Target: i686-pc-cygwin
Configured with: /cygdrive/i/szsz/tmpp/gcc4/gcc-4.8.2-2/src/gcc-4.8.2/configure --srcdir=/cygdrive/i/szsz/tmpp/gcc4/gcc-4.8.2-2/src/gcc-4.8.
2 --prefix=/usr --exec-prefix=/usr --bindir=/usr/bin --sbindir=/usr/sbin --libexecdir=/usr/libexec --datadir=/usr/share --localstatedir=/var
 --sysconfdir=/etc --libdir=/usr/lib --datarootdir=/usr/share --docdir=/usr/share/doc/gcc --htmldir=/usr/share/doc/gcc/html -C --build=i686-
pc-cygwin --host=i686-pc-cygwin --target=i686-pc-cygwin --without-libiconv-prefix --without-libintl-prefix --enable-shared --enable-shared-l
ibgcc --enable-static --enable-version-specific-runtime-libs --enable-bootstrap --disable-__cxa_atexit --with-dwarf2 --with-arch=i686 --with
-tune=generic --disable-sjlj-exceptions --enable-languages=ada,c,c++,fortran,java,lto,objc,obj-c++ --enable-graphite --enable-threads=posix
--enable-libatomic --enable-libgomp --disable-libitm --enable-libquadmath --enable-libquadmath-support --enable-libssp --enable-libada --ena
ble-libjava --enable-libgcj-sublibs --disable-java-awt --disable-symvers --with-ecj-jar=/usr/share/java/ecj.jar --with-gnu-ld --with-gnu-as
--with-cloog-include=/usr/include/cloog-isl --without-libiconv-prefix --without-libintl-prefix --with-system-zlib --libexecdir=/usr/lib
Thread model: posix
gcc version 4.8.2 (GCC)

I hope this particular problem of my environment. Under linux everything works like a charm.

static dyad_Event createEvent(int type) { dyad_Event e; memset(&e, 0, sizeof(e)); e.type = type; return e; }

'e' has been freed when this func returned any memory cordump problem ?

Sorry for this simple question, but I do not understand how to reconnect after a failure. I tried a few combinations of dyad_end / dyad_update but it always fails. (see example here here ) The only way that seems to work is to create a new stream and reset the callback functions. Is this the correct method?

`

Hey, I am trying to convert this to c++ for a project I am on. When i change the dyad.c file to dyad.cpp the conversion requires me to cast all of the reallocs instantly.

For instance: data = (char)dyad_realloc(_data, *capacity * memsz); instead of: *data = dyad_realloc(_data, *capacity * memsz);

This causes the connection for the ircbot to fail and return a bad request. Do you know of a way to fix this?

...to #defines

Hey, I experienced a crash on Windows 7 because of these magic numbers. According to this page: http://pubs.opengroup.org/onlinepubs/009695399/functions/inet_ntop.html inet_ntop's behavior can be bad if the size parameter is less than INET_ADDRSTRLEN for ipv4, INET6_ADDRSTRLEN for ipv6. So I just stuck those macros in.

I've tested on Windows 7, iOS 7, Mac OSX, and Android 4.4.2 at least. Not that you should trust me!

I'm not sure if you had a reason to use magic numbers instead of macros, maybe they are missing on some setups. I can do some #ifdef logic if that's the case though.

Thanks for this wonderful library, by the way!

Best, Dan

Just a nitpick, dyad_setSocketNonBlocking does this:

fcntl(stream->sockfd, F_SETFL, opt ? O_NONBLOCK : ~O_NONBLOCK);

Which blows away any other flags. It should be: flags = fcntl(stream->sockfd, F_GETFL); fcntl(stream->sockfd, F_SETFL, opt ? (flags | O_NONBLOCK) : (flags & ~O_NONBLOCK));

When calling "dyad_listen(serv, )", it does not accept local connections. As a workaround, I have used "dyad_listenEx(serv, "0.0.0.0", , 511)".

• Made client streams work on Windows (AF_UNSPEC => AF_INET)
• Made the source code compilable in C++ mode
• Added fast TCP loopback path for Windows (for Windows Server 2012, Windows 8 and higher)

I am trying to send a message each 100 ms but client accepts only one message per second. I checked my code. There is no sleep code around. What could be the reason?

Thank you.

Related issue: https://github.com/rxi/dyad/issues/25

Hi!

This PR brings the method dyad_unix_connect and a new example.

The method allows to connect over Unix domain socket. When it runs over Windows platform, it raises an error event. Also, I updated the API doc and added an example to help.

Regards.