Date: Mon, 10 Dec 2001 17:16:23 +1030 From: Greg Lewis <glewis@eyesbeyond.com> To: Bill Huey <billh@gnuppy.monkey.org> Cc: freebsd-java@FreeBSD.ORG Subject: Re: jdk1.3.1p5 Message-ID: <20011210171623.D73467@misty.eyesbeyond.com> In-Reply-To: <20011210024138.GA3148@gnuppy>; from billh@gnuppy.monkey.org on Sun, Dec 09, 2001 at 06:41:38PM -0800 References: <20011210001702.10731.qmail@web14303.mail.yahoo.com> <20011210024138.GA3148@gnuppy>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, Dec 09, 2001 at 06:41:38PM -0800, Bill Huey wrote: > On Mon, Dec 10, 2001 at 12:17:02AM +0000, shanon loveridge wrote: > Yes, the native threaded version of the JVM will have the ability > to exploit stuff like the Posix AIO API when it maps itself to NIO. This > will allow it to have superior scalability over any kind of select()/poll() > model over many thousands of idle FDs. Hopefully, those two systems will > be conceptually compatible and the remapping of that IO event model will > be straight foreword under FreeBSD. With SMPng lock granularity replacement > and a new IO model I would expect it to be a very strong competitor to other > server-side systems in the Unix and NT kernel based arenas. It's not clear > if this group realizes the importance of this, but it's definitely one of > my implicit goals when I started working on native threading in this community > is to have world class IO event handling. All else aside, if there are no native threads there is no JDK 1.4 port. If people don't think native threads are important, they should take some time to think about whether they want a JDK 1.4 port or not... :-) Maybe that may get through where more technical arguments fail :). > The current FreeBSD port is missing the very important HotSpot compiler and > other things necessary for it to be performance competitive with other Unix > based JVMs, etc... HotSpot is very very important. Its the performance piece we're lacking to get FreeBSD accepted as a really solid alternative for serving Java based web apps and the like. I _think_ that native threads may also be a prerequisite for HotSpot... I also think the browser plugin is important, but thats because the people I work for have a Java applet based product and 'cos it yanks my chain when I'm browsing with Mozilla :). Big vote of thanks to Joe for the progress he is making in that area. > Also, as a minor note, *cough*, native threading is a prerequist for the > JCK (Java Certification Kit) tests so that the current JVM we're working on > can legally be released to the general public. It's a bit frustrating in > that it's not clear to me or that I'm not getting the necessary feedback from > folks in the community that indicate that native threading is important > both of these technically and legally reasons from the luke warm responses > I get from folks. I can't comment for anyone else, but I apologise if you feel that you're getting luke warm responses about native threading. I'm aware of how important it is and appreciate how you've run with the ball on this! My part in this is to obviously get your patches out in a new patchset so we can get some more eyes and testers on this. Unfortunately I've had to work most of the weekend and haven't gotten a complete handle on things :(. Patchset 6 is still under production... -- Greg Lewis Email : glewis@eyesbeyond.com Eyes Beyond Phone : (801) 796 6999 Information Technology Web : http://www.eyesbeyond.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-java" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20011210171623.D73467>