jon * wrote:
>
> on 3/31/00 1:20 AM, Stefano Mazzocchi <[EMAIL PROTECTED]> wrote:
>
> > Hey, man, don't take it personally... probably I had some network
> > congestion between the 250 hosts that I have between me and tesla... :)
> > (with locus is much better, this is why i would like to move there)
>
> that is funny because our hosting provider for locus isn't very good (yes,
> we are switching). and Icarus (www.working-dogs.com) is off a T1 through BBN
> (ie: a very good provider).
>
> > damn... question: how long would that take? (not just because of jserv:
> > if the locus CVS goes down, man, I can't think of the consequences...)
>
> Well, Apache.org is one of the most widely mirrored code bases. ;-)
>
> > Great! Happy to see this project still alive... :)
>
> This project is still totally alive. But, it is not alive from a development
> standpoint, only a bug fix standpoint.
>
> > Hmmm, the fact that Jean-Luc didn't know what I was talking about
> > worries me: did you post those proposed changes on the list?
>
> I believe they were posted to jserv-dev by Jean-Frederic...???
>
> > FYI, guys, DataChannel architects sent me a bunch of changes to
> > mod_jserv that allowed for AJP socket recycling and they told us they
> > had a 400% performance increase over HelloWorldServlet... which doesn't
> > mean you get 400% performance increase on your servlets, of course, but
> > it surely doesn't hurt.
>
> The last patches that I saw regarding this stuff required changes to the
> core httpd and I really just didn't have the energy to go and push this
> through the core httpd group at the time. Part of my problem is that if we
> had to change the core httpd, then that would require us to be in sync with
> their release schedule and I really didn't want to tie us to that.
>
> -jon
I found it. It was submitted by John Milan if it is the same one. no
mention of DataChannel.
As far as I can remember, it required changes in the core (I was not
enthousiast with this)
and was more than buggy on Unix (was designed & tested on W32). By
example, it was not taking care of the multiprocess design of Apache (on
unix) (I hope I understood it well).
ideas were good, but easier to wait ajpv2* from tomcat, that we could
integrate back in JServ (if needed). For me the next protocol version
will be compatible with tomcat or won't be. (a pain in the ass for
ajpv12. my fault).
Jean-Luc
--
----------------------------------------------------------
To subscribe: [EMAIL PROTECTED]
To unsubscribe: [EMAIL PROTECTED]
Archives and Other: <http://java.apache.org/main/mail.html>
Problems?: [EMAIL PROTECTED]
