Mailing-List: contact apache-bugdb-help@apache.org; run by ezmlm
Precedence: bulk
Reply-To: apache-bugdb@apache.org
Date: Wed, 25 Jul 2001 00:25:33 +0900 (JST)
Message-Id: <200107241525.f6OFPXb07531@masamune.bio.is.tohoku.ac.jp>
From: <kabe@sra-tohoku.co.jp>
To: stoddard@apache.org
Cc: apache-bugdb@apache.org, kabe@sra-tohoku.co.jp
Subject: Re: general/8037: [PATCH] HEAD response header repeated twice for
 files over 32 kilobytes
In-Reply-To: Your message of "24 Jul 2001 14:24:44 -0000".
	<20010724142444.4710.qmail@apache.org>

>> Thanks for the bug report.  I believe the right fix is to uninstall the header filter the first time through on HEAD requests.  (right before we return OK). Will be fixed in the next release. Thanks again!
>> Bill

Thanks, I checked the CVS. (httpd-2.0/modules/http/http_protocol.c 1.330)

But I guess this could break in a long run;
what if some generator sent down a brigade like
	1.Big file bucket
	2.Big some bucket
	3.EOS

Bucket 1 will kick the partial write of the CONTENT_LENGTH, passes it to
HEADER and HEADER detaches.
Then Bucket 2 is processed by CONTENT_LENGTH, partial write, pass it down;
this time CORE gets it, which doesn't care r->header_only and just 
spits out the content of the bucket to network(bad).

I know the above situation won't occur in current Apache2, but some
third-party handlers/generators could assemble the brigade.
Any thoughts?
(I knew goto bad and if(){}ing was possible but indent change makes patch big)
--
kabe