Start of the sockets documentation make-over

Describe in detail all possible options to socket_create() without the need to refer to other documentation.


git-svn-id: https://svn.php.net/repository/phpdoc/en/trunk@106231 c90b9560-bf6c-de11-be94-00142212c4b1
This commit is contained in:
Jason Greene 2002-11-29 23:57:20 +00:00
parent ad0b4e7089
commit 4d84a12305

View file

@ -1,5 +1,5 @@
<?xml version="1.0" encoding="iso-8859-1"?>
<!-- $Revision: 1.3 $ -->
<!-- $Revision: 1.4 $ -->
<!-- splitted from ./en/functions/sockets.xml, last change in rev 1.4 -->
<refentry id="function.socket-create">
<refnamediv>
@ -14,47 +14,155 @@
<methodparam><type>int</type><parameter>type</parameter></methodparam>
<methodparam><type>int</type><parameter>protocol</parameter></methodparam>
</methodsynopsis>
&warn.experimental.func;
<para>
Creates a communication endpoint (a socket), and returns a socket
resource.
Creates and returns a socket resource, also referred to as an endpoint
of communication. A typical network connection is made up of 2 sockets, one
performing the role of the client, and another performing the role of the server.
</para>
<para>
The <parameter>domain</parameter> parameter sets the domain (protocol
family) to be used for communication. Currently,
<constant>AF_INET</constant> and <constant>AF_UNIX</constant> are
understood. <constant>AF_INET</constant> is typical used for internet
based communication. <constant>AF_UNIX</constant> uses pathnames to
identify sockets and can therefore only be used for local communication
(which is faster, on the other hand).
The <parameter>domain</parameter> parameter specifies the protocol
family to be used by the socket.
</para>
<table>
<title>Available address/protocol families</title>
<tgroup cols="2">
<thead>
<row>
<entry>Domain</entry>
<entry>Description</entry>
</row>
</thead>
<tbody>
<row>
<entry>AF_INET</entry>
<entry>
IPv4 Internet based protocols. TCP and UDP are common protocols of
this protocol family.
</entry>
</row>
<row>
<entry>AF_UNIX</entry>
<entry>
Local communication protocol family. High efficiency and low
overhead make it a great form of IPC (Interprocess Communication).
</entry>
</row>
</tbody>
</tgroup>
</table>
<para>
The <parameter>type</parameter> parameter selects the socket
type. This is one of <constant>SOCK_STREAM</constant>,
<constant>SOCK_DGRAM</constant>,
<constant>SOCK_SEQPACKET</constant>,
<constant>SOCK_RAW</constant>, <constant>SOCK_RDM</constant>, or
<constant>SOCK_PACKET</constant>. The two most common types are
<constant>SOCK_DGRAM</constant> for <literal>UDP</literal>
(connectionless) communication
and <constant>SOCK_STREAM</constant> for <literal>TCP</literal>
communication.
The <parameter>type</parameter> parameter selects the type of communication
to be used by the socket.
</para>
<table>
<title>Available socket types</title>
<tgroup cols="2">
<thead>
<row>
<entry>Type</entry>
<entry>Description</entry>
</row>
</thead>
<tbody>
<row>
<entry>SOCK_STREAM</entry>
<entry>
Provides sequenced, reliable, full-duplex, connection-based byte streams.
An out-of-band data transmission mechanism may be supported.
The TCP protocol is based on this socket type.
</entry>
</row>
<row>
<entry>SOCK_DGRAM</entry>
<entry>
Supports datagrams (connectionless, unreliable messages of a fixed maximum length).
The UDP protocol is based on this socket type.
</entry>
</row>
<row>
<entry>SOCK_SEQPACKET</entry>
<entry>
Provides a sequenced, reliable, two-way connection-based data transmission path for
datagrams of fixed maximum length; a consumer is required to read an
entire packet with each read call.
</entry>
</row>
<row>
<entry>SOCK_RAW</entry>
<entry>
Provides raw network protocol access. This special type of socket
can be used to manually construct any type of protocol. A common use
for this socket type is to perform ICMP requests (like ping,
traceroute, etc).
</entry>
</row>
<row>
<entry>SOCK_RDM</entry>
<entry>
Provides a reliable datagram layer that does not guarantee ordering.
This is most likely not implemented on your operating system.
</entry>
</row>
</tbody>
</tgroup>
</table>
<para>
<parameter>protocol</parameter> sets the protocol which is either
<constant>SOL_UDP</constant> or <constant>SOL_TCP</constant>.
The <parameter>protocol</parameter> parameter sets the specific
protocol within the specified <parameter>domain</parameter> to be used
when communicating on the returned socket. The proper value can be retrieved by
name by using <function>getprotobyname</function>. If
the desired protocol is TCP, or UDP the corresponding constants
<constant>SOL_TCP</constant>, and <constant>SOL_UDP</constant>
can also be used.
</para>
<table>
<title>Common protocols</title>
<tgroup cols="2">
<thead>
<row>
<entry>Name</entry>
<entry>Description</entry>
</row>
</thead>
<tbody>
<row>
<entry>icmp</entry>
<entry>
The Internet Control Message Protocol is used primarily by gateways
and hosts to report errors in datagram communication. The "ping"
command (present in most modern operating systems) is an example
application of ICMP.
</entry>
</row>
<row>
<entry>udp</entry>
<entry>
The User Datagram Protocol is a connectionless, unreliable,
protocol with fixed record lengths. Due to these aspects, UDP
requires a minimum amount of protocol overhead.
</entry>
</row>
<row>
<entry>tcp</entry>
<entry>
The Transmission Control Protocol is a reliable, connection based,
stream oriented, full duplex protocol. TCP guarantees that all data packets
will be received in the order in which they were sent. If any packet is somehow
lost during communication, TCP will automatically retransmit the packet until
the destination host acknowledges that packet. For reliability and performance
reasons, the TCP implementation itself decides the appropriate octet boundaries
of the underlying datagram communication layer. Therefore, TCP applications must
allow for the possibility of partial record transmission.
</entry>
</row>
</tbody>
</tgroup>
</table>
<para>
Returns a socket resource on success, or &false; on error. The actual
error code can be retrieved by calling
<function>socket_last_error</function>. This error code may be passed to
<function>socket_strerror</function> to get a textual explanation of the
error.
</para>
<para>
For more information on the usage of <function>socket_create</function>,
as well as on the meanings of the various parameters, see the
Unix man page socket (2).
<function>socket_create</function> Returns a socket resource on success, or &false;
on error. The actual error code can be retrieved by calling <function>socket_last_error</function>.
This error code may be passed to <function>socket_strerror</function> to get a textual
explanation of the error.
</para>
<note>
<para>