Fw: [PATCH] Allow output buffer size to be set for HTTP stream

[Scott Nichol <sn...@computer.org>]

I posted this code, which I find very useful, over a month ago.  Wouter and
I exchanged a couple of messages regarding this.  Ultimately, he said he was
looking at re-writing the transport layer altogether.  If that re-write will
be included in the 2.2 code, great!  If not, please re-consider inclusion of
this small change.  My preference is that with the 2.2 release, I will no
longer have to use a custom built soap.jar.

Thanks.

Scott Nichol

----- Original Message -----
From: "Scott Nichol" <sn...@computer.org>
To: <so...@xml.apache.org>
Sent: Thursday, April 12, 2001 12:54 PM
Subject: [PATCH] Allow output buffer size to be set for HTTP stream


> The attached file contains changes (diff -u format) to HTTPUtils.java and
> SOAPHTTPConnection.java to allow the output buffer size to be set for the
HTTP stream.
> The current code does not explicitly set the buffer size.  For all JDKs I
have, this
> means the output buffer will be 512 bytes.  When large requests are
posted, the buffer
> is flushed many times, creating less than optimal network I/O.  By
allowing the buffer
> size to be specified at the transport level, its size can be optimized on
a per-call
> basis if necessary.
>
> Scott Nichol
>
>

Re: Fw: [PATCH] Allow output buffer size to be set for HTTP stream

[Wouter Cloetens <wo...@mind.be>]

I apologise for not having gotten around to this yet. I've built up quite a
to-do list from suggestions, reports and patches from the fora, but it's
becoming painfully clear that I just don't have the time and energy anymore.
:-(
That's after changing from a job in which I was mostly very busy, to a job in which
my *brain* is actually really busy to the point of exhaustion. ;-)

So it's clear that my big plans for 2.2, which include a redesigned transport layer,
JMS, transport-independent session management, as well as the addition of some
missing SOAPContext/provider functionality, won't make it in that revision...
I'll try to get around doing your patch and some MIME/attachment documentation
though.

bfn, Wouter

On Fri, May 18, 2001 at 12:30:12PM -0400, Scott Nichol wrote:
> I posted this code, which I find very useful, over a month ago.  Wouter and
> I exchanged a couple of messages regarding this.  Ultimately, he said he was
> looking at re-writing the transport layer altogether.  If that re-write will
> be included in the 2.2 code, great!  If not, please re-consider inclusion of
> this small change.  My preference is that with the 2.2 release, I will no
> longer have to use a custom built soap.jar.

RE: [PATCH] Allow output buffer size to be set for HTTP stream

["Matthew J. Duftler" <du...@watson.ibm.com>]

Hi Scott,

Ok, your changes have been committed. Please take a look and make sure
everything's as you intended.

Thanks for the contribution,
-Matt

> -----Original Message-----
> From: Scott Nichol [mailto:snichol@computer.org]
> Sent: Friday, May 18, 2001 12:30 PM
> To: soap-dev@xml.apache.org
> Subject: Fw: [PATCH] Allow output buffer size to be set for HTTP stream
>
>
> I posted this code, which I find very useful, over a month ago.
> Wouter and
> I exchanged a couple of messages regarding this.  Ultimately, he
> said he was
> looking at re-writing the transport layer altogether.  If that
> re-write will
> be included in the 2.2 code, great!  If not, please re-consider
> inclusion of
> this small change.  My preference is that with the 2.2 release, I will no
> longer have to use a custom built soap.jar.
>
> Thanks.
>
> Scott Nichol
>
> ----- Original Message -----
> From: "Scott Nichol" <sn...@computer.org>
> To: <so...@xml.apache.org>
> Sent: Thursday, April 12, 2001 12:54 PM
> Subject: [PATCH] Allow output buffer size to be set for HTTP stream
>
>
> > The attached file contains changes (diff -u format) to
> HTTPUtils.java and
> > SOAPHTTPConnection.java to allow the output buffer size to be
> set for the
> HTTP stream.
> > The current code does not explicitly set the buffer size.  For
> all JDKs I
> have, this
> > means the output buffer will be 512 bytes.  When large requests are
> posted, the buffer
> > is flushed many times, creating less than optimal network I/O.  By
> allowing the buffer
> > size to be specified at the transport level, its size can be
> optimized on
> a per-call
> > basis if necessary.
> >
> > Scott Nichol
> >
> >
>