Talk:Network socket/Archive 1
This is an archive of past discussions about Network socket. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page. |
Archive 1 |
Suggested merge
I suggest that Ip socket be merged into this article. It appears to be a stub covering the same subject. I plan to complete this merge by 2006-09-08 if there is no objection and/or a consensus is reached. -- Ratarsed 11:46, 29 August 2006 (UTC)
If this article is about IP sockets then the comment about SNA should be moved elsewhere. SNA only runs over DLSw on the Internet as it as its own lower level protocol. In addition there is not a clear definition here of a socket. Does it include the API? The queue? If there are multiple definitions they should be clearly identified and listed. The higher level definition of "socket" is in error by including "an end-point of a bi-directional communication link in the Berkeley sockets API". Pointing to a parent entity is not a valid definition. User:jbottoms76
Request
Could someone go over the process an application might take in creating a socket -I'm interested in how port numbers are assigned in particular User:DragonSlayre12:14pm, 9 September 2007 (New Zealand time). —Preceding unsigned comment added by 124.197.57.56 (talk) 00:16, 9 September 2007 (UTC)
Missing references: The article does not refer to any API standard. There are some de facto standards for Berkely Sockets and WinSock.
Citations to RCF/STD are needed. For Example, RFC3493, R. Gilligan et al., “Basic Socket Interface Extensions for IPv6,” February 2003, http://www.rfc-editor.org/ discusses difference between v.4 and v.6 versions of the library. —Preceding unsigned comment added by 67.173.186.180 (talk) 04:22, 7 February 2010 (UTC)
Merge proposal for Datagram socket into Network socket
The discussion for the proposed merger of Datagram socket into this article (Network socket) is at: Talk:Datagram socket#Missing scope distinction; Merge destination re-proposed. Sparkie82 (t•c) 17:46, 1 January 2016 (UTC)
- I Support this proposal. I have put up merge banners on the two articles. ~Kvng (talk) 21:37, 24 August 2017 (UTC)
- I went to perform this merge and discovered that Stream socket and Raw socket were also merge candidates. I have put up additional banners and widened the proposal. I will wait for more input or more time before proceeding. ~Kvng (talk) 16:32, 24 October 2017 (UTC)
Done ~Kvng (talk) 16:29, 11 December 2017 (UTC)
Socket without address??
The explanation says : "Sockets need not have an address (for example for only sending data)" Can someone explain how that works?
Surely, to build a connection, you have to have a source socket (address and port) and a destination socket (address and port). If you have no source address, how do your packets reach the network (which interface should they be sent through)?
Should it say "Sockets need not have an address explicitly set (for example for only sending data will use the IP address of the interface used for sending and be automatically assigned a high port number)" With perhaps a link to "high port numbers" -> https://en.wikipedia.org/wiki/Ephemeral_port Or am I missing something? — Preceding unsigned comment added by Max.allan.surevine (talk • contribs) 07:26, 3 August 2016 (UTC)
You can create an UDP socket, not bind() it, and use sendto() to send data to example.com:4711. Then the source address will be chosen based on the route to example.com. Probably that's what the author had in mind. With TCP client sockets, you don't have to bind() either, but the source address will be calculated in the same way when you connect() to example.com. — The "for example for only sending data" seems confused, though. In my UDP example, example.com:4711 can reply just fine. JöG (talk) 06:30, 27 January 2018 (UTC)