Submitted by DragonFly BSD 1.4 has been released. “The two biggest user-visible changes in this release are a major revamping of libc, ctype, and wchar support, as well as changes made in the kernel which require us to bump the major rev for all of our shared libraries, and the introduction of PKGSRC to manage third party applications. DragonFly no longer supports the FreeBSD PORTS system.” Get it here.
I’m very glad to see the DFBSD project moving ahead.
I presume that work still needs to be done before it is declared production ready. How much of a leap forward towards that goal is this release?
I see a lot of bugs and reworking has been done but in a strange kind of way, that tells me little about the status of the project. I found this on the release page:
“From a stability perspective we believe this is our best release to date!”
It’s amazing to see the amount of progress that has been made in DragonFly the last few months given the much smaller number of developers as compared to the other BSD’s. I am very impressed with the quality of the code that has gone into DFly and the way that the team goes out of their way to squash ‘buggs’ before more new code is added to the source tree… I am really looking forward to all the cool new features that Matt has planned for the 1.5 release in the next 3-4 months as well!
> It’s amazing to see the amount of progress that has been made in DragonFly the last few months given the much smaller number of developers as compared to the other BSD’s.
Well, it is normal: the more developers you have working on a project, the more synchronization/communication between them you have to do (a dozen of chief cooks in a small kitchen can not work together efficiently).
DragonFly developers are organized as a small “medical emergency team” (one surgeon giving instructions and doing the very hard stuff –Matt Dillon– and a few skilled assistants around him, doing atomic chunks of work), so progress is extremely fast while keeping the patien^H^H^H^H^H^Hsystem quite stable.
“I presume that work still needs to be done before it is declared production ready.”
Depends on the situation. General desktop use? Yeah. It still needs work. But it had been every bit as stable for me as FreeBSD 4.x had been, and of course, has newer features and drivers to boot. Is it ready for big MP machines? Not yet, but they are getting very close. I’d be suprised if DF doesn’t scale as well as Linux or FreeBSD in a years time.
“I see a lot of bugs and reworking has been done but in a strange kind of way, that tells me little about the status of the project.”
Uhuh. Bugs happen. They get dealt with. Same everywhere.
“From a stability perspective we believe this is our best release to date!”
Show me a project that doesn’t have the same tag line. Seriously.
“It’s amazing to see the amount of progress that has been made in DragonFly the last few months given the much smaller number of developers as compared to the other BSD’s.”
More than anything, that is what stands out for me as well. They are doing what others are doing, but with far less money and people power.
They are doing what others are doing
No, they don’t
I think they do something better – they are experimenting with new concepts, bringing a fresh air to unix-land. At least that’s my impression of their work. I don’t use it, and probably won’t in the near future (I’m happy with Free~, and I’m not experienced enough to use it I think) – but I think it is one of the most exciting and promising projects today. It also has its distinct identity amond the BSDs. I know this is oversimplification, but there is some truth to the OpenBSD = security, NetBSD = portability, FreeBSD = general purpose. Dragonfly adds clustering to the fray.
Today clustering is linux + 3rd party cruft. I have a (very) brief experience with clustering. We tried to build a 4 node cluster with the clustering version of Mandrake 3 years ago. Clustering Mandrake meant MDK 9.1 with “enterprise” kernel (exactly the same you would find on stock mandrake CDs), the usual stuff (that is available in ports for FreeBSD) and Mandrake scripts to help you set up the cluster. There are of course tweaks you can do to the linux/bsd kernel to perform better in a clustering environment, but DragonFly would be built for that task from the ground up. If the userland tools to help setting up “secure anonymous system clustering over the internet” will be as friendly as most BSD tools (it is much easier to learn BSD unix than linux – just compare setting up a good firewall with pf vs. linux’s iptables) – we have something that is brand new and unique. And despite their focus on clustering, from what I’ve seen, they don’t neglect the “general purpose” aspect of the OS – so I think Dragonfly will be usable when it’s ready as a simple desktop OS as well with the added ability to perform tasks much faster by connecting to friendly anonymous nodes and sharing cpu cycles. Sounds incredible? It is, and I think this is Matt’s vision.
I also think that this is and should be the future – we share our thoughts, aspects of our lives (think of the blogosphere), our goodwill (think of helping completely unknown persons on forums) and the fruits of our labour (think of Free Software developers) over the Internet. This is gift economy at its best (and incidentally, I’m writing about this in my dissertation). So why not share our more “material” resources? I think most people don’t even realize the potential of the Internet, and how it can profoundly change our life (and not just our, but humanity’s in general) and change the way we think of each other (over a few generations of course – if nothing bad happens). I think Matt understands this … and dragonfly will be great.
Just think of the possibilities – when it solidifies to be as stable as FreeBSD 6.x is today, it can serve as a basis for initiatives like PC-BSD. Call it PC-Dragonfly: a userfriendly OS that anyone can use and install. Many users than will hang out on PC-Dragonfly forums, they get to know each other, etc. What if PC-Dragonfly had an architecture that would let very easy sharing of resources between the forum members. How fast would I be able to encode that h.264 movie
))? Think of something similar to peer-to-peer networks, but sharing your cpu cycles. Yes, I know it can be done today, there are lots of clusters, some of them can be rented to do various tasks, but if dragonfly’s goals can be achieved (“secure anonymous system clustering over the internet”) it may bring that power to spontaneously forming communities.
OK, I that might have been a little bit of a stretch I know
It might even sound silly – but it is a possibility. This is not a short term project, I very much doubt that dragonfly will be “finished” in the upcoming 1-2 years, and there is so much that can happen till then. We’ll see.
“No, they don’t
I think they do something better ”
I meant general features wise, not architecturally. I do agree with nearly everything you’ve said tho.
…DragonFlyBSD will succeed. I am loving FreeBSD 6.0 right now (in the guise of PC-BSD, altho’ I’ve done all manual updates, basically rebuilt the system). I am so pleased with its performance! My understanding is that DFly BSD has the potential to be an even better performer. Looking forward to that.