Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cwjcc!tut.cis.ohio-state.edu!att!chinet!les
From: les@chinet.chi.il.us (Leslie Mikesell)
Newsgroups: comp.unix.questions
Subject: Re: Sysv pipes/FIFO's & blocking
Summary: I meant when the writer hasn't closed its end
Message-ID: <9362@chinet.chi.il.us>
Date: 25 Aug 89 18:25:40 GMT
References: <9348@chinet.chi.il.us>
Reply-To: les@chinet.chi.il.us (Leslie Mikesell)
Organization: Chinet - Public Access Unix
Lines: 36

In article <9348@chinet.chi.il.us> les@chinet.chi.il.us (Leslie Mikesell) writes:
>What is supposed to happen under sysV if a pipe or FIFO contains
>3 characters and a read() requests 4 characters?  What about fread()?
>Is it documented anywhere?


I've received some mail replies that assumed that the writer was finished
and had closed the writing end of the pipe, in which case both read()
and fread() return the 3 available characters.  I wanted to know about
the case where the reader catches up, but the writer isn't finished.

On SysVr3.2 (AT&T 6386), read() will return the 3 available characters,
fread() will wait until 4 characters are available or EOF.

This surprised me a little:
-----------------
#/bin/sh
READ=500
while :
do
   echo "1" 
done |
while :
do
   dd bs=$READ count=1 of=/dev/null
done
-------------------
It gives a mixture of:
  1+0 records in
  1+0 records out
and
  0+1 records in
  0+1 records out
with the latter form increasing with the size of READ.  Why?

Les Mikesell