Bug 16673

Summary: mod_include problems with multiple include virtual requests and cgi-wrapper
Product: Apache httpd-2 Reporter: Jelmer Jellema <apche_bug>
Component: mod_includeAssignee: Apache HTTPD Bugs Mailing List <bugs>
Status: CLOSED DUPLICATE    
Severity: major    
Priority: P3    
Version: 2.0.44   
Target Milestone: ---   
Hardware: PC   
OS: Linux   
Attachments: Raw session using cgi-script handler (output_cgi-scripthandler.txt)
Raw session using a wrapper (AddHandler/Action) (output_wrapped.txt)

Description Jelmer Jellema 2003-02-02 12:42:00 UTC
This is a follow-up on bug 16498, in which I presented a problem in 
mod_include in version 2.0.40: when using "include virtual=" elements in a ssi 
page AND using a wrapped cgi-handler (AddHandler + action), mod_include 
returned a wrong Content-length header.

In 2.0.44 this is fixed by not returning a Content-length header at all for 
ssi pages. This works for some browsers (i.e. Netscape 4.08), but not for 
others (i.e. IE 6) because of another bug:

When using a wrapped cgi handler (I now refer to bug 16498), mod_include will 
enter 2 0-length lines after the output of an "include virtual=" element. (see 
the 2 attachments output_cgi-scripthandler.txt and output_wrapped.txt for the 
difference).

IE thinks, after the first 0 line the file is done and closes connection. 
Maybe this is not according to protocol (I don't know the ins and outs), but 
it's certainly a problem with a lot of visitors, who all use IE 6. Better not 
to return these lines.

Please see the attachments of bug 16498 for the setup, and the to outputs 
enclosed here.
Comment 1 Jelmer Jellema 2003-02-02 12:44:12 UTC
Created attachment 4677 [details]
Raw session using cgi-script handler (output_cgi-scripthandler.txt)
Comment 2 Jelmer Jellema 2003-02-02 12:45:55 UTC
Created attachment 4678 [details]
Raw session using a wrapper (AddHandler/Action) (output_wrapped.txt)
Comment 3 Jeff Trawick 2003-02-24 04:07:19 UTC
The description and the incorrect output you showed matches another
PR I fixed earlier today: PR 15423.

Please verify that the fix for PR 15423 (small patch attached to that PR)
resolves the problem.  That fix is now in 2.1-dev and has been proposed for
merge into the stable tree, which is 2.0.45-dev.

Thanks!


*** This bug has been marked as a duplicate of 15423 ***