Mailing-List: contact dev-help@hc.apache.org; run by ezmlm
Precedence: bulk
Reply-To: "HttpComponents Project" <dev@hc.apache.org>
Date: Thu, 6 Feb 2014 14:44:11 +0000 (UTC)
From: "Dmitry Potapov (JIRA)" <jira@apache.org>
To: dev@hc.apache.org
Message-ID: <JIRA.12693402.1391623778279.48834.1391697851161@arcas>
In-Reply-To: <JIRA.12693402.1391623778279@arcas>
References: <JIRA.12693402.1391623778279@arcas>
Subject: [jira] [Commented] (HTTPCORE-372) ChunkedInputStream and
 ChunkDecoder allows closing chunk absence
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable


    [ https://issues.apache.org/jira/browse/HTTPCORE-372?page=3Dcom.atlassi=
an.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=3D13=
893390#comment-13893390 ]=20

Dmitry Potapov commented on HTTPCORE-372:
-----------------------------------------

Oleg,

I'm not sure I've got you correct.
ChunkDecoder lines 180-181 become unreachable,  that's why whole block on l=
ines 179-182 was removed. New block at lines 90-100 follows the same logic =
as in past for case when endOfStream =3D=3D false; for endOfStream =3D=3D t=
rue case I fill cr and lf with zeroes to have exception thrown at line 102,=
 so, no new throws needed to indicate protocol violation.

So, I see the only one change, which can be introduced here: remove lines 9=
4, 95, 124 and 125 which sets state to COMPLETED and completed flag to true=
.

If you have any other considerations concerning these changes, please, let =
me know

> ChunkedInputStream and ChunkDecoder allows closing chunk absence
> ----------------------------------------------------------------
>
>                 Key: HTTPCORE-372
>                 URL: https://issues.apache.org/jira/browse/HTTPCORE-372
>             Project: HttpComponents HttpCore
>          Issue Type: Improvement
>          Components: HttpCore, HttpCore NIO
>            Reporter: Dmitry Potapov
>             Fix For: 4.4-alpha1
>
>         Attachments: expect-closing-chunk.patch
>
>
> I've found that both these classes treats chunked streams without closing=
 chunk as complete streams without any errors. This contradicts RFC2616 =C2=
=A73.6.1
> Consider the following scenario:
> 1. HttpClient sends request to Server
> 2. Server accepts request and send response headers with Transfer-Encodin=
g: chunked
> 3. Server sends first chunk completely and dies with segfault before send=
ing the next chunk
> 4. HttpClient receives response headers and creates ChunkedInputStream
> 5. ChunkedInputStream reads this first chunk and encounters eof
> Expected result:
> IOException thrown indicating incomplete response
> Actual result:
> No exception is thrown, -1 is returned on the next read and nobody knows =
that response is incomplete.


--
This message was sent by Atlassian JIRA
(v6.1.5#6160)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
For additional commands, e-mail: dev-help@hc.apache.org