Path: utzoo!utgpu!jarvis.csri.toronto.edu!clyde.concordia.ca!uunet!samsung!zaphod.mps.ohio-state.edu!tut.cis.ohio-state.edu!ASC.SLB.COM!steve
From: steve@ASC.SLB.COM
Newsgroups: gnu.emacs.bug
Subject: Process management bug patch - context diff
Message-ID: <8912212225.AA01097.steve@boron.ASC.SLB.COM>
Date: 21 Dec 89 22:25:32 GMT
References: <8912151828.AA07601@sugar-bombs.ai.mit.edu>
Sender: daemon@tut.cis.ohio-state.edu
Distribution: gnu
Organization: GNUs Not Usenet
Lines: 42

Thanks for your comments.

The change you propose is incomplete. The current version of 
read_process_output (I mean the version in dist-18.54) returns 0
whenever the actual read system call returns a non-positive value.
That's why I changed it to return the result of the read call:

***************
*** 1562,1568 ****
  	nchars = nchars + 1;
      }
  
!   if (nchars <= 0) return 0;
  
    outstream = p->filter;
    if (!NULL (outstream))
--- 1569,1575 ----
  	nchars = nchars + 1;
      }
  
!   if (nchars <= 0) return nchars; /* 0; */
  
    outstream = p->filter;
    if (!NULL (outstream))

This shouldn't affect other calls to read_process_output, since tests are
on its result being positive or non-positive, not 0. The following greps
indicate this:

boron[steve]42> grep 'read_process_output'  /usr/local/emacs/src/*.c
/usr/local/emacs/src/process.c:	      if (n = read_process_output (proc, channel) > 0) 
/usr/local/emacs/src/process.c:read_process_output (proc, channel)
/usr/local/emacs/src/process.c:	while (read_process_output (proc, XFASTINT (p->infd)) > 0);

boron[steve]48> grep 'read_process_output' /usr/local/emacs/src/*.h
boron[steve]49> 


Without this change to read_process_output, the test for EWOULDBLOCK would
never succeed.

Thanks as always.