Slashdot Mirror
FreeBSD 5.3 on the Horizon
underpar writes "ZDNet UK is reporting that FreeBSD is nearing a code freeze. August 15th is the deadline which will be followed by the usual beta testing and a final release hoped for by October 1st. ZDNet interviewed the software engineer leading the release work, Scott Long, for the article. He says: 'The 5.3 release will be the first one where we see the real benefits of that. The multithreaded network stack will outperform everything we've done before, for running applications such as Apache or MySQL.' Status reports can be found on the FreeBSD website." I've been using the last technology release of FreeBSD for some time now, and am really looking forward to the 5.3 release, as well as the 5-STABLE branch that's rumored to follow soon after.
A patch I wrote for the CVS versions of Apache/APR to Add KQueue support has been added to the FreeBSD Port version of Apache.
Just make the port with "WITH_EXPERIMENTAL_PATCHES=1" and you can get a 10-25% boost in performance. (depends on your traffic patterns..)
Its a quick way to get more performance out of Apache on FreeBSD, without waiting for the 5-STABLE branch.
-Paul Querna
Using samba, if you share Fat32 and write to it from the network, you end up with corrupted files.
I hope it has been fixed, but I somehow doubt it since it's been around for at least 2 years (earliest bug report was on 4.6RC) so it exists in -stable as well.
A native port is already there! /usr/ports/java/jdk-1.4.2
p.s. If you want a prebuilt binary of jdk-1.4.2, then complain to Sun. They're the ones that prohibit the distribution of Java packages for BSD.
Don't blame me, I didn't vote for either of them!
"The JDK(TM) it produces is de facto compliant, but use in a production environment is still at your own risk."
But thanks for your words of encouragement. That "We don't want you" crack must make you feel really good.
Ok, first of all, linux compatability isn't any slower than running native binaries. Its just syscall translation, simple mapping of what linux syscalls the app is using corespond to what freebsd syscalls to do the same thing.
And you really can't blame FreeBSD for Sun having horrible license restrictions on java. If java were free it would already be ready for you. But because its not, there is a serious lack of people who are willing to sign away their life and ability to ever sue sun so that they can do the work of porting something they don't want anyways just for you.
See also: http://www.freebsd.org/java/dists/14.html
It's Sun compliant, just like 1.3. What do you want? An insurance policy? It's fscking open source software for chrissake.
For the record, we use native 1.4 on 15 fbsd 4.10 boxen (using tomcat/freemarker) for production enterprise systems with 4K+ users slammin' the boxes each day. No problems so far.
But I suggest you keep waiting. We'll be sure to send you a certificate or 'notice of native compliance' or something. Juuuust keep waiting...
RTFA ;)
It won't be in beta until mid-August. The final release is expected in October.
You just read the headline didn't you?
Bad Panda! No Bamboo for you! In matters of importance ACs will not be responded to. Want to say something critical,OK
ITYM ROFL! :-)
Yup, FreeBSD is fully userland ready, has been for, lets see, the last 6-8-ish years that I've been using it!
On the subject of RTFA, as the article says, 5.2.1-RELEASE is a little jumpy in some parts, and if you need solid stability, stick with 4.10-RELEASE for the time being.
I've used FreeBSD in production environments for years and years and years. Right now, I'm running 4.9-RELEASE and 4.10-RELEASE on production servers both at work and at home.
I'm tinkering with 5.2.1-RELEASE on a new Dell X300 laptop and a P4 desktop at the moment. They're both working pretty well, and surprisingly, I've got almost everything on the (very screwed up hardware-wise) X300 working! I have managed to break 5.2.1 several times, but it was mostly by doing really wacky things with the Project Evil code, upping and downing and kldloading and kldunloading different drivers on different interfaces with not enough kernel memory allocated for the bloated third party windows code!!!
Having said that, Project Evil is nothing short of a *GODSEND*, and Bill Paul is god! It's pretty amazing to be pinching windows NDIS drivers and compiling them into FreeBSD kernel modules - opens doors for all kinds of obscure hardware that couldn't be used before!
It's still too early for me to make any definitive comment on whether 5.x.x is good on desktops as yet, but if it's anything like the FreeBSDs that came before it, it will be nothing short of excellent when it hits -STABLE.
I find your ideas intriguing and I wish to subscribe to your newsletter.
Well, 10.3 had bits of FreeBSD 5 in it according to Apple's page for it.
And according to the tiger preview page it's based on FreeBSD 5.x - so that would probably be 5.2.1.
save the GNUs!
From the article
FreeBSD 5.2[3] will also introduce a software layer that lets Windows network drivers work with FreeBSD. This layer, dubbed Project Evil, means that wired and wireless network cards should be able to work with FreeBSD even if the manufacturers have not written any drivers for the operating system.
This is totally awesome! FreeBSD network drivers are very reliable, but hard to come by for very new devices (eg. wifi). I would totally use this feature even with some reliability sacrifice.
In Soviet Russia, articles before post read *you*!
One problem with the FreeBSD installer is that it's both an installer and a configuration tool with menues that does not remember previous settings that you have done.
The OpenBSD installer is just that : an installer. Post configuration is mostly done after installation.
No one needs to break an NDA; any portions of the 10.4 DP that would be relevant to FreeBSD are publicly available through version 8.0b1 of the Darwin project.
...and you require 5 nines reliability you are living in cloud cuckoo land if you think you can get what you are asking for gratis.
Stop being so cheap and shell out some dosh with Sun if you really need fanatical support and guaranteed operation. Although, if you're supporting that many people and it's as mission critical as you are saying then cost should be no problem, so I'm inclined to think you're just trolling on the whole Java issue.
I've stress tested my companies J2EE product on a FreeBSD box, pusing it to its limits and had zero problems, however, YMMV.
I am NaN
According to the status update released a couple days ago, makeworld now completes natively, IIRC. Go search google. :)
"Why do you consent to live in ignorance and fear?" - Bad Religion
Except in this case, he's 100% wrong. The threading issue was a problem with early versions of the port. It's now been fixed. If you just read what google gives you, and don't bother to check the mailing lists, you'll end up with stale complaints that are no longer a problem.
5-STABLE was scheduled for 5.2 originally, but they pushed it back. (this was like a year ago when they thought 5.2).
In fact, originally 5.3 was set for Late May-ish, early June, but 4.10 got in the way, 5.2.1 was still pretty recent, and 5.x still needed work.
I'm working on it. I think it's quite stable right now, but some features are still missing. I really hope to get it done before 5.3.
I use FreeBSD 4.x, FreeBSD 5.x, and MacOS X 10.3.x, and I can attest that FreeBSD 5.x material is in it already.
(Links shamelessly stolen from Kyro's post, and modified to point to Apple's US server)
> You're absolutely right. However there's a difference between the developer claiming "usable for most tasks. However [...]" and a developer claiming a version is a "Production" version.
And in all my experiences (I know, anecdotal evidence, but still, obne that is confirmed by many others who tried) 'usable for most tasks' in the FreeBSD world is a lot more usable then 'production ready' in the Windows world, and even in the Linux world.
> I find it rather funny that most of the responses claim that I haven't looked at the ports tree. On the contrary, I have kept very close tabs on the advancement of Java under FreeBSD. Back when our app ran under 1.1.8, we happily deployed FreeBSD. I will be first in line to deploy it again when the 1.4 JDK is bumped up to production-ready.
Have you actually tried it?
I have been running the FreeBSD native port for a long time, both server and client side, and it simply works, period.
There is an issue, that issue is Sun's source code licensing. That is however not a technical issue, and if you can build it, it is very usable for a production environment also.
I'm currently playing with the 1.5 jdk.
Just use package :
pkg_add -r net-snmp
or
portupgrade -NPP
It is good to be lazy ...
You know, 'portupgrade -aPP' is much faster, because it uses binary packages, as apt-get does.
Read this and this.
In short, to BSD users your argument is laughably silly and makes you look like an ignoramus.
I may be wrong, but I honestly think that most reasonable people will not interpret a cartoonish picture of a devil-like creature wearing sneakers as any indication of satan worship
I myself have walked into my church (I'm Catholic) wearing a FreeBSD daemon shirt.
The OS-X Kernel is Mach, yes, but then immediately loads the BSD kernel which pretty much runs monolithicly (very non-Mach). Below that level, you get in to a BSD userland, and the Carbon / Aqua layers. It's really not all *that* bizarre.