Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!swrinde!mips!sgi!shinobu!odin!sgihub!dragon!sam.wpd.sgi.com!pj
From: pj@sam.wpd.sgi.com (Paul Jackson)
Newsgroups: comp.sys.sgi
Subject: Re: Problem with pipe/fork
Message-ID: <1991Apr16.191706.3898@dragon.wpd.sgi.com>
Date: 16 Apr 91 19:17:06 GMT
References: <1991Apr16.150557.2344@jsp.umontreal.ca>
Sender: news@dragon.wpd.sgi.com (CNews Account)
Reply-To: pj@sgi.com
Organization: Silicon Graphics, Research & Development
Lines: 29

In article <1991Apr16.150557.2344@jsp.umontreal.ca>, gaumondp@JSP.UMontreal.CA (Gaumond
Pierre) writes:
[[ Sample program presented that has parent writing pipe to child.
   He asks why child doesn't notice EOF on the pipe after the parent closes it. ]]

|>   int fdi[2];
|>   
|>   pipe(fdi);
|>   if (fork()!=0) { /* parent */
	[[ Loops writing to fdi[1], then closes fdi[1]. ]]
|>   } else { /* child */
	[[ Loops reading from fdi[0] until eof - but eof never seen. ]]
|>   }
|> }

In the child, before the loop until eof, add the line:

	close(fdi[1]);

The fork resulted in both the parent and child having
fdi[1] open for writing, and the child will see eof
on fdi[0] only when all available writers to the pipe
have closed the write end.

-- 

				I won't rest till it's the best ...
				Software Production Engineer
				Paul Jackson (pj@wpd.sgi.com), x1373