Re: another quirk (fwd)

[Howard Fear <hs...@pooh.pageplus.com>]

Rob Hartill writes:
> I'd say it's a showstopper since it craps all over the CLF by lying about
> the number of bytes sent. Some people out there make a living out of selling
> bandwidth by the megabyte shipped.

I've included a patch which seems to work here, but I'm not sure about
its greater impact.  I believe that this avoids sending headers (and
re-initializing the byte count) for nested includes.  I'm not sure,
however, whether send_parsed_file ever gets called as a sub-request
(as opposed to a main request) except when it calls itself recursively.

It shouldn't when redirected right?  And we all know that it can't
be called from a cgi right?  Are there any other cases I'm missing?

For nested calls, I shouldn't need the headers the second time around.
When they're sent, they're simply dropped anyway.  There might
be some error conditions that get ignored I guess, but it seems to
continue to process the obvious ones (file not found).

--
Howard Fear      email1: howard_fear@pageplus.com
                 email2: howard_fear@redcape.com
                 http://www.pageplus.com/~hsf/


*** ../../apache_19961107020012/src/mod_include.c	Sun Nov  3 19:00:15 1996
--- mod_include.c	Sat Nov 16 23:14:38 1996
***************
*** 1679,1685 ****
      }
      
      hard_timeout ("send", r);
!     send_http_header(r);
  
      if (r->header_only) {
          kill_timeout (r);
--- 1679,1687 ----
      }
      
      hard_timeout ("send", r);
!     if (!r->main) {
!         send_http_header(r);
!     }
  
      if (r->header_only) {
          kill_timeout (r);