home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!dtix!darwin.sura.net!zaphod.mps.ohio-state.edu!cis.ohio-state.edu!rutgers!princeton!bow.Princeton.EDU!luochen
- From: luochen@bow.Princeton.EDU (Luoqi Chen)
- Newsgroups: comp.unix.programmer
- Subject: Re: Must Unix domain sockets have a pathname?
- Message-ID: <1992Sep15.231651.16329@Princeton.EDU>
- Date: 15 Sep 92 23:16:51 GMT
- References: <1992Sep15.060317.3750@news.uiowa.edu>
- Sender: news@Princeton.EDU (USENET News System)
- Organization: Princeton University
- Lines: 14
- Nntp-Posting-Host: bow.princeton.edu
-
- From article <1992Sep15.060317.3750@news.uiowa.edu>, by dsiebert@icaen.uiowa.edu (Doug Siebert):
- > Does a Unix domain socket (datagram, where sockets would be sending datagrams
- > of from 5 - 500 bytes each to each other between processes) have to be bound
- > to a name in the Unix filesystem? From the man pages I'd have to say "yes",
- > but I distinctly remember a few times seeing something that made me think that
- > unnamed sockets could exist. I know they can for sending to a named socket,
- > but I'd have to name the other end...
- >
- > Its a really trivial point, I know, but it has me curious.
- >
- socketpair() makes a pair of unamed sockets, currently it is only implemented
- for unix domain sockets.
-
- -lq
-