Received: (qmail 3389 invoked by uid 2012); 25 Mar 1999 16:12:41 -0000
Message-Id: <19990325161241.3388.qmail@hyperreal.org>
Date: 25 Mar 1999 16:12:41 -0000
From: Jacques Distler <distler@golem.ph.utexas.edu>
Reply-To: distler@golem.ph.utexas.edu
To: apbugs@hyperreal.org
Subject: Instructions for simulating RefererIgnore incorrect
X-Send-Pr-Version: 3.2

>Number:         4129
>Category:       mod_log-any
>Synopsis:       Instructions for simulating RefererIgnore incorrect
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    apache
>State:          feedback
>Class:          doc-bug
>Submitter-Id:   apache
>Arrival-Date:   Thu Mar 25 08:20:03 PST 1999
>Last-Modified:  Fri Apr 30 15:20:00 PDT 1999
>Originator:     distler@golem.ph.utexas.edu
>Organization:
>Release:        1.3.6
>Environment:
NextStep3.3 (hppa), with NeXT's cc compiler.
1> uname -a
NEXTSTEP golem Rose2L  7100LC 
>Description:
<http://www.apache.org/docs/mod/mod_log_config.html> says:

	" Beginning with Apache 1.3.5, the mod_log_config module has also
	subsumed the RefererIgnore functionality from _mod_log_referer_. The
	effect of RefererIgnore can be achieved by combinations of _SetEnvIf_ 
	directives and conditional CustomLog definitions.

and <http://www.apache.org/docs/mod/mod_setenvif.html> gives the example

		SetEnvIf Request_URI "\.(gif)|(jpg)|(xbm)$" object_is_image
		SetEnvIf Referer www\.mydomain\.com intra_site_referral
  

	The first will set the envariable object_is_image if the request was
	for an image file, and the second sets intra_site_referral if the 
	referring page was somewhere on the www.mydomain.com Web site.

This does NOT, however, simulate the behaviour of the RefererIgnore directive in
mod_log_referer.
>How-To-Repeat:

>Fix:
The "correct" construction, which reproduces the behaviour of

	RefererIgnore www.mydomain.com

is

	SetEnvIf Referer "^http://www\.mydomain\.com|www\.mydomain\.com" intra_site_referral
>Audit-Trail:

From: Jacques Distler <distler@golem.ph.utexas.edu>
To: apbugs@hyperreal.org, apache-bugdb@apache.org
Cc:  Subject: Re: mod_log-any/4129: Instructions for simulating RefererIgnore
	incorrect
Date: Thu, 25 Mar 1999 10:31:21 -0600

 Whoops!
 
 To *really* simulate the RefererIgnore directive, you also need to drop  
 blank http_referer fields. So make that:
 
 	SetEnvIf Referer "^http://www\.mydomain\.com|www\.mydomain\.com|^$"  
 intra_site_referral
 
State-Changed-From-To: open-feedback
State-Changed-By: coar
State-Changed-When: Thu Apr 29 12:43:08 PDT 1999
State-Changed-Why:

I'll grant your point about the null Referer value, but why
do you think the full URI needs to be specified?  The
mod_log_referer module just checks to see if the RefereIgnore
string appears anywhere in the Referer field value.  It
doesn't check only the host portion of the Referer URI.

Comment-Added-By: coar
Comment-Added-When: Thu Apr 29 12:44:44 PDT 1999
Comment-Added:

Sorry, you're not saying the full URI is required, just
that the test text be anchored at the beginning of the
string.  Which is still not a requirement of mod_log_referer.


From: Jacques Distler <distler@golem.ph.utexas.edu>
To: coar@apache.org
Cc: <apbugs@apache.org>
Subject: Re: mod_log-any/4129: Instructions for simulating RefererIgnore
	incorrect
Date: Fri, 30 Apr 1999 17:11:58 -0500

 I'm not sure what we are quibbling about.
 
 The NCSA http-referer log, with the RefererIgnore directive, ignored blank  
 HTTP_REFERER fields and HTTP_REFERER fields which are URIs from the named web  
 site.
 
 I expect that there are various different regular expressions one could  
 write down which would accomplish this in mod_log_referer. However, if the  
 documentation states that
 
 	mod_log_referer and
 
 		SetEnvIf Referer REGEXP
 
 	simulate the NCSA behaviour
 
 then they ought to do that. The fact that mod_log_referer logs null  
 HTTP_REFERER fields is a departure from the behaviour of NCSA http-referer  
 log. That should be noted in the docs.
 
 Perhaps my modification of your sample REGEXP only muddies the issue.
 
>Unformatted:
[In order for any reply to be added to the PR database, ]
[you need to include <apbugs@Apache.Org> in the Cc line ]
[and leave the subject line UNCHANGED.  This is not done]
[automatically because of the potential for mail loops. ]
[If you do not include this Cc, your reply may be ig-   ]
[nored unless you are responding to an explicit request ]
[from a developer.                                      ]
[Reply only with text; DO NOT SEND ATTACHMENTS!         ]