Dan, Does the patch you posted always call setTcpNoDelay(true)? I wonder whether it is better to give the developer the option of whether to do this or not, e.g. by adding methods to SOAPHTTPConnection to get/set whether to disable Nagle, then adding another overloaded post method to HTTPUtils that takes a parameter as to whether to disable Nagle. The default would be not to disable, so that everyone who does not explicitly disable will get the same behavior as they do currently.
Just a thought. Scott Nichol ----- Original Message ----- From: "Dan Davis" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Sent: Wednesday, May 22, 2002 2:44 AM Subject: Re: Vote: Apache SOAP 2.3 Release Schedule > > +0 > > I wanted to make sure a change I made to HTTPUtils.java is included. > I'm presenting a paper that includes this change. In HTTPUtils.java, > call setTcpNoDelay (true) once you've got the socket. This disables > the Nagle algorithm on the client side and speeds up each individual > client substantially -- 200 ms to 20 ms on a LAN, no effect if both > client and server are on the same host. > > I posted this change earlier, but never made a bug report, so I wasn't > sure the change would make it in. > > I'm sorry I've taken so long to reply to vote, but I'm on vacation and > finally got access to the email where I receive soap-dev. > > I've tested this change in both applet and standalone, as I'd been > concerned that setting socket options might violate security restrictions > in some browsers/JVM's. I know Sun's is okay, and suspect that all > Java 2 JVM's will be fine, but I haven't tried it with a good number > of JVMs. > > ****************************** > I'm participating this summer in a 150 mile bike-a-thon to end Multiple > Sclerosis. If you would like to make a donation to help, please contact me > or visit this URL: > https://www.nationalmssociety.org/pledge/pledge.asp?participantid=61823 > to pledge on-line. Thanks! > ****************************** > > >
