The following reply was made to PR mod_cgi/2894; it has been noted by GNATS.
From: Gerd Knops <[EMAIL PROTECTED]>
To: Marc Slemko <[EMAIL PROTECTED]>, Craig Miskell <[EMAIL PROTECTED]>
Cc: [EMAIL PROTECTED]
Subject: Re: mod_cgi/2894: cgi triggers premature EOF to be sent to client
Date: Mon, 7 Sep 98 21:10:14 -0500
Hello,
I received an EMail from another person (Craig Miskell
<[EMAIL PROTECTED]>) who is able to reproduce the problem on the
same platform. I also did some more tests and the problem is not related to
the 'sleep', any code that burns a little time can trickle the problem. Even
the test script added below triggers the problem on occasion, the error log
then reports
Signal: TERM
Signal: PIPE
Signal: PIPE
Signal: PIPE
Craig suggested that the problem seems to occur at a 4096 byte boundary, and
also that only perl scripts seem to trigger it. Could it be some odd
combination of Apache/OPENSTEP 4.2/perl (5.004_02 in my case) that causes the
problem?
As a side node: I have mod_perl compiled in, and when I run an identical
script handled by mod_perl the problem goes away.
If you could provide any pointers as to where to start looking in the code
for the cause I'd gladly give it a try.
Thanks
Gerd
---------------cut---------------
#!/usr/local/bin/perl -w
foreach (keys %SIG)
{
$SIG{$_}=\&tell;
}
$|=1;
print "Content-type: text/html\n\n<HTML>\n</HEAD>\n<BODY>\n";
$t="The quick brown fox jumps over the lazy dog<BR>\n";
for($i=0;$i<200;$i++)
{
if($i%10==0)
{
print "$i-s<BR>\n";
}
print $t;
}
print "END<BR>\n</BODY>\n";
sub tell
{
print STDERR "Signal: ",shift,"\n";
}
