Mailing-List: contact dev-help@couchdb.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@couchdb.apache.org
Received-SPF: pass (athena.apache.org: local policy)
Subject: Re: COUCHDB-864 and 1.0.x
From: Klaus Trainer <klaus.trainer@web.de>
To: dev@couchdb.apache.org
In-Reply-To: <AANLkTi=MifAga8xne+CRR3Qw7dQKGCGVeKwy4fCFgL-X@mail.gmail.com>
References: <AANLkTi=MifAga8xne+CRR3Qw7dQKGCGVeKwy4fCFgL-X@mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"
Date: Sun, 22 Aug 2010 19:16:00 +0200
Message-ID: <1282497360.14749.13.camel@devil>
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Sender: klaus.trainer@web.de

+1
As soon as there's an adequate test.
I'm quite curious about that test ;).

-Klaus


On Sun, 2010-08-22 at 14:34 +0100, Robert Newson wrote:
> All,
> 
> I committed a one-line fix (suggested by Adam) to trunk which now
> allows multipart/related PUT's to succeed without mochiweb closing the
> connection. At base, mochiweb uses a process dictionary variable to
> keep track of whether it ought to close the connection or not. Because
> we use the request from a different process, part of that handling is
> skipped, causing mochiweb to close a connection when it shouldn't. The
> fix (or, less charitably, *hack*) fixes that issue.
> 
> I've struggled to write an etap test that demonstrates the original
> problem (the bug was discovered when doing performance testing from a
> node.js-based test program) and I feel I owe one. I would like this
> patch to land on 1.0.x where I think a confirming test is mandatory.
> 
> I'll continue trying to write the test and will put it to trunk once I
> have it. In the meantime, I would like to hear from others about
> whether this should go to 1.0.x? It allows efficient and atomic
> insertion of documents with attachments, which is an important feature
> for me, though to date it is only used by the replicator.
> 
> B.