DragonFly 4.0.2 has been tagged. I’m building the release images now. If you’re already running 4.0.1 it’ll be easy enough to upgrade to; you will want to catch up to this commit fixing a quiet memory issue.
As part of another thread, Steve Petrie posted an in-depth description of how and where and why he’s using DragonFly. Worth looking at either for workflow tips or for just seeing the use case.
If you want to help I/O performance when DragonFly is virtualized, here’s a short checklist of what to work on. I haven’t noticed any problems – but I’m not taxing any of my VMs that heavily.
This page, Varialus et Anisoptera, set up by… I’m not sure of the real name but it’s ‘varialus’ on IRC – has a detailed description of the DragonFly install process and installation of MATE, plus extra notes. I always find these sorts of cheatsheets entertaining.
Robin Hahling wants feedback on where to go in DragonFly with rcrun(8), service(8), and similar commands. Follow the thread to see the various opinions.
Predrag Punosevac posted his writeup of using LDAP and DragonFly, which I’m noting here for the next person that needs LDAP authentication.
There’s an extended article about the DragonFly 4.0 release on linuxfr.org. You need to be able to read French to enjoy it fully, or perhaps through translation, but it goes into some good depth.
A fellow whom I’ve only seen named as Bill is working on what he calls ipfw2, though technically what’s already in DragonFly is ipfw2, since it’s the second version of ipfw. Either way, he has a project page up describing what he’s done so far, and what he plans.
If you look at your local DragonFly mirror, you’ll see ISO and IMG versions of DragonFly 4.0.0RC3. Please run, break, and report.
(Check the iso-images directory.)
Matthew Dillon had some followup commits that went in just after I tagged RC2 of DragonFly 4.0 last night, so I’ve tagged RC3. Tagging’s cheap, anyway.
I just tagged a second release candidate of DragonFly 4. Matthew Dillon’s recent reapctl() addtions – now called procctl() – just went in.
The release candidate for DragonFly 4.0 came out last week, and normally the release would happen after a week. There’s still a few people reporting an odd freeze, so until we can find a cause, we’ll continue to wait.
If you can read French, or can translate, here’s an article covering the use of DragonFly as a desktop system.
Your local mirror should have a copy of the release candidate for DragonFly 4.0.0 by now. Please try it out and report problems. Note that this is a x86_64 only version; there’s no i386 version though you may be able to manually build on i386.
It’s been possible to install and run clang on DragonFly for a long time, of course, and at least build world with it. However, John Marino is putting in significant work to make clang one of the system compilers, replacing the older gcc44 that’s in DragonFly now. (The newer gcc47 stays.) This won’t be part of the next release, but it should be available soon after.
I noted the last few things that should be committed before the DragonFly release. People have spoken up for most of them, but it wouldn’t hurt to try recent -master with the upmap/kpmap work that recently went in. Benchmarks wouldn’t be a bad idea, either.
There’s been so much work in DragonFly recently that makes a desktop easier (i915 support, dports, and so on), that I decided to resurrect an older Dell machine and use it as my desktop.
The Dell that I’m using is a leftover from someone else’s workplace; it’s 7 years old, and has “only” 4G of RAM and a Core 2 DuoE6600 CPU in it. It works, however.
Setting up DragonFly and installing xorg and so on is pretty straightforward. Using dports makes it crazy quick to add all the packages. I went for XFCE4 because I could. Starting X gave me some trouble at first; the default config couldn’t find the mouse and would eventually crash.
Running ‘X -configure’ created a xorg.conf file I could edit, and these lines in /etc/rc.conf gave me a working mouse:
moused_enable="YES" moused_type="auto" moused_port="/dev/ums0"
The crashing problem with my radeon-driven video card was fixed by turning off the acceleration – uncommenting this line in xorg.conf did it:
Option "NoAccel"
Video performance isn’t as nice as I would like it with acceleration, but this is an older machine anyway.
I couldn’t get sound working. Francois Tigeot has a branch of DragonFly that contains newer sound drivers brought over from FreeBSD, here:
git://leaf.dragonflybsd.org/~ftigeot/dragonfly.git (pcm_2014_september branch.)
It doesn’t support device cloning, so I can run Youtube videos and XMMS, but not audio from both at the same time. (for instance; not that you’d want to do this other than by accident)
I installed x11/webfonts, and web pages look a bit better after changing my default font preferences.
And… that’s about it. It’s a working desktop. Digging up a half-height video card that has working acceleration is a next step, but I can’t imagine that’ll be expensive. I wish I had done this a long time ago.
I’m doing this little extra feature because I ran into several news items over the past week or so that made me say “what the hell?” out loud to my monitor.
Fedora To Get a New Partition Manager. All? Almost all? Linux distributions use gparted, which is open source and can be updated. Why not add to that? Also, it’s yet another preannouncement about how this new replacement tool will work – it’s not functional yet.
Text streams should be the fallback interface in Unix. Every 2 or 3 years someone gets this idea in some form – somehow it doesn’t overcome 40+ years of text usage.
Revisiting How We Put Together Linux Systems. Nobody can find fault with ideas like easier package management and signing. (Though maybe having the same upgrade mechanism for base + 3rd party software isn’t a good idea) However. this answer, coming from part of the group behind systemd, ties all software installation into having a btrfs volume – even requiring a virtual btrfs volume if there isn’t one installed. Incompatible software versions are dealt with by turning /usr into a sort of container. That kills any sort of need to interoperate with other software. And of course it assumes there is no Unix but Linux. (via)
Grump grump grump.
If you are on DragonFly, using pf, using altq, and using fairq to control usage, there’s a latency bug that Matthew Dillon recently fixed. He’s posted an announcement and committed fixes to master and 3.8, so it’s only an upgrade away.
You should perform a full world and kernel install if on master.
Several people (including me) have been getting bit by a problem: when performing an installworld with a changed kernel, the vn kernel module is loaded, but it was built by the previous kernel and may cause problems when it doesn’t match up.
To fix that, vn is now built in, instead of being a separate module. The rescue initrd (which is what is being mounted when it has this problem) is now installed via a ‘make rescue‘ command that can wait until a successful installworld and reboot.