Xref: utzoo comp.os.mach:519 comp.sys.next:7600
Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!mit-eddie!uw-beaver!milton!wiml
From: wiml@milton.u.washington.edu (William Lewis)
Newsgroups: comp.os.mach,comp.sys.next
Subject: Re: Named pipes and Next
Message-ID: <7394@milton.u.washington.edu>
Date: 12 Sep 90 00:45:17 GMT
References: <ERAN.90Sep11153920@adelbert12.stanford.edu>
Distribution: comp
Organization: University of Washington, Seattle
Lines: 24

In article <ERAN.90Sep11153920@adelbert12.stanford.edu> eran@adelbert12.stanford.edu (eran yehudai) writes:
>I am writing a utility that uses named pipes (created by 'mknod p
>name'). It works fine on suns, decs and even IBM aix. Only today I
>found out the NeXT (I'm not sure about Mach in general) does not
>support them, and I am looking for a simple alternative.
>
>Is there a *simple* way of creating an effective pipe between two
>independent processes? The Speaker Listener mechanism sounds about
>right, but I'd rather not go into objective C.
>Any ideas?

   Well, you could use Unix-domain sockets; they seem sort of
named-pipe-like (IMHO). If you want to be Mach or 
NeXT-specific, you could use Mach ports and the name-server;
unless you use NXStreams though you have to invest a little learning
about "Mach Interface Generator" and details of mach ports (which
I myself still haven't done =8) ). Since this is comp.os.mach + 
comp.sys.next I guess you're not worried about portability off of
a NeXT. Check allocate_port() and NXOpenPort() in the manuals. 
The nameserver is under netname_check_in() and netname_look_up().

-- 
wiml@milton.acs.washington.edu       Seattle, Washington  | No sig under
(William Lewis)   |  47 41' 15" N   122 42' 58" W  |||||||| construction