Mailing-List: contact dev-help@apr.apache.org; run by ezmlm
Precedence: bulk
Received-SPF: pass (athena.apache.org: domain of steven.nairn@googlemail.com
 designates 209.85.198.189 as permitted sender)
DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=googlemail.com; s=beta;
        h=received:message-id:date:from:to:subject:mime-version:content-type;
        b=B6zaSLOfF6KE9yQuZwDXnt3LcBvdgz6/2m89kfehXqzm7biB0hI0zNB+hMGdSXGkEdfYMisDfNWPYxOpnZ5z1ZqCMqmnz8rHDE/RpQQWR/DawnARb+PpOelCRPwQyNRlH6Z776xRuAHkGibLGDX0crceC8pdZl7SOWADO64gm2w=
Message-ID: <b6adab440708310308j58d4b052t7489ee8871596be9@mail.gmail.com>
Date: Fri, 31 Aug 2007 11:08:50 +0100
From: "Steven Nairn" <steven.nairn@googlemail.com>
To: dev@apr.apache.org
Subject: Re: Am I hitting the list?
MIME-Version: 1.0
Content-Type: multipart/mixed;
	boundary="----=_Part_1570_29200990.1188554930117"

------=_Part_1570_29200990.1188554930117
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

Mladen Turk wrote:
> OK. The mail doesn't hit the httpd dev list as well.
> Looks like I really found something that no one wishes to hear :)
>
> However here is the link to the body that ASF is rejecting:
>
> http://jbecome.net/dload/aprtest.txt

The "testfile" failure on win32 is in test_gets_buffered() because
read_with_timeout() (in file_io/win32/readwrite.c) does not correctly
report end of file when the file is opened with the APR_FOPEN_XTHREAD
flag.

On Windows the APR_FOPEN_XTHREAD flag means that the file is opened
with FILE_FLAG_OVERLAPPED and is read by ReadFile() with an OVERLAPPED
structure. In this case when end of file is reached ReadFile() returns
false and GetLastError() returns ERROR_HANDLE_EOF (error 38, which APR
reports as 720038). See MSDN docs on ReadFile:
http://msdn2.microsoft.com/EN-US/library/aa365467.aspx

Without the "overlapped" stuff when the end of file is reached
ReadFile() returns true but sets the number of bytes read to zero.

The fix is to check for ERROR_HANDLE_EOF in read_with_timeout()
(attached patch).

The test_gets_buffered() test in testfile.c is new in 1.2.9 so the
problem probably existed in 1.2.8 but wasn't tested.

Cheers,
Steve.

------=_Part_1570_29200990.1188554930117
Content-Type: text/plain; name="patch.txt"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="patch.txt"; filename="patch.txt"
X-Attachment-Id: f_f60iu1g9

LS0tIGZpbGVfaW8vd2luMzIvcmVhZHdyaXRlLmMtb3JpZwlUaHUgQXVnICAzIDE3OjE2OjQxIDIw
MDYKKysrIGZpbGVfaW8vd2luMzIvcmVhZHdyaXRlLmMJRnJpIEF1ZyAzMSAxMDo1NTowOCAyMDA3
CkBAIC0xMjIsNiArMTIyLDkgQEAKICAgICAgICAgaWYgKHJ2ID09IEFQUl9GUk9NX09TX0VSUk9S
KEVSUk9SX0JST0tFTl9QSVBFKSkgewogICAgICAgICAgICAgLyogQXNzdW1lIEVSUk9SX0JST0tF
Tl9QSVBFIHNpZ25hbHMgYW4gRU9GIHJlYWRpbmcgZnJvbSBhIHBpcGUgKi8KICAgICAgICAgICAg
IHJ2ID0gQVBSX0VPRjsKKyAgICAgICAgfSBlbHNlIGlmIChydiA9PSBBUFJfRlJPTV9PU19FUlJP
UihFUlJPUl9IQU5ETEVfRU9GKSkgeworICAgICAgICAgICAgLyogRGlkIHdlIGhpdCBFT0YgcmVh
ZGluZyBmcm9tIHRoZSBoYW5kbGU/ICovCisgICAgICAgICAgICBydiA9IEFQUl9FT0Y7CiAgICAg
ICAgIH0KICAgICB9CiAgICAgCg==
------=_Part_1570_29200990.1188554930117--