Joerg Sonnenberger pointed me at a recent post on the NetBSD tech-kern@ mailing list: Andrew Doran did some comparisons of MySQL’s sysbench on a multi-CPU system, with different operating systems. It unfortunately doesn’t include DragonFly, as DragonFly apparently would not boot on that system, but I’m a sucker for graphs.
It also shows generally better performance for NetBSD recently than for a Linux 2.6 kernel. This is interesting in part because MySQL performance on BSD has historically been worse than on Linux.
OnLAMP has another of its rare BSD articles up; this time on installing Subversion on BSD, with all the ‘bells and whistles”.
Spotted by Hasso Tepper: The Software Freedom Law Center has a new article up titled “Maintaining Permissive-Licensed Files in a GPL-Licensed Project: Guidelines for Developers“, which is another way of saying “How to treat BSD-style licenses right”.
Hasso Tepper wants to get rid of the pcidevs and usbdevs files, as the effort of maintaining them appears to outweigh the benefits.Â So far, most people agree.
KernelTrap now has web archives of the mailing lists for DragonFly, along with a number of other projects. The interface looks nice, and allows you to track by author, too.
Hasso Tepper has taken the OpenBSD sensors framework, as ported to FreeBSD by a Summer of Code project, and converted it for DragonFly along with a number of drivers.
Peter Avalos has updated libarchive and also calendar.Â Thanks, Peter.
Seen via Gizmodo and other places: 1 gigabyte of storage, then and now.
Noah Yan has committed more of his AMD64 work to DragonFly; check the README for details on how to experiment with it.
Hubert Feyrer has done a very nice job of collating all the online material from the various presentations, with data from Axel Gruner, that happened at EuroBSDCon 2007.
I recently completed a bulk build of pkgsrc using Joerg Sonnenberger’s pbulk tool; there has been discussion of using these packages as part of a mirroring system (poke around the thread for more.)
OpenBSD Journal has an interesting article up that talks about the life cycle of a bug, as seen by an OpenBSD user.Â I call it interesting because it gives a good summary of a bug-squishing process from a ‘user’ perspective.
The freeze period (where only bugfixes are committed) for the next quarterly release of pkgsrc starts tomorrow.Â Interestingly, this next quarter’s release marks 10 years of pkgsrc.
Strange as it is to use the words “C compiler” and “excitement” in the same sentence, there’s been a lot of excitement about PCC, the Portable C Compiler, as a faster replacement for GCC. (Previous story here)
There’s a web page for it, and a mailing list,
though no mail archive I can find associated with it archived at MARC. (Thanks, Anonymous). The web page has a link to an old PostScript document detailing the original PCC design – there’s something about old Unix manuals from Bell Labs that makes them fun to read.
And of course, there’s always the inevitable Wikipedia page.
In BSDTalk 129, Will Backman talks to Dru Lavigne about BSD Certification and her new job at The Open Source Business Resource.
Matthew Dillon, as part of a larger discussion, chimed in with some sensible descriptions of licensing and how it applies to the recent OpenBSD/Linux kerfuffle.
Hasso Tepper added a whole pile of uftdi(4) drivers. Why? Apparently it’s cheaper to buy from FTDI than it is to buy a vendor ID from usb.org.
Noah Yan posted how to apply his recent patch for building an AMD64 kernel.Â Be warned; it does not create a full usable system – yet.
There’s an effort to make the ‘official’ pkgsrc logo happen; previous discussion was described here. It looks like the simple version is the candidate; there’s a fun, alternate version that unfortunately won’t reproduce well.
pcc has been added to NetBSD (via pkgsrc) and OpenBSD, and Steve Mynott has been messing with it on DragonFly. It doesn’t work as a replacement for GCC, but it looks promising. There are other alternatives in progress, too.
This week, BSDTalk talks about sysjail, the Open/NetBSD version of FreeBSD ‘jail’, with Michael Dexter. (Yes, I realize that’s an oversimplification.)
Sepherosa Ziehau has a patch that makes it possible to assign polling(4) to specific CPUs.
Update: There’s a new version of that patch.
If enabling ACPI means that some of the devices attached to your computer can’t be found, YONETANI Tomokazu has a patch that may fix it.
Sascha Wildner has added two new man pages: kernconf(5), for explaining kernel options, and firmware(9), for the process of loading firmware images into the kernel.
Chris Turner is also a new DragonFly developer with commit access.Â Welcome, Chris.
Hasso Tepper has a patch that appears to fix net-snmp; it can be downloaded for someone who needs SNMP now, and it should hopefully be integrated into pkgsrc soon.
DragonFly’s newest developer with commit access: Noah Yan, already known to be working on the AMD64 version of DragonFly.Â Welcome, Noah.
Hasso Tepper has brought in extensive changes to agp(4), from FreeBSD.Â For a full list of the manyÂ new supported devices, puzzle through the man page diff.
There’s a good number of commits to DragonFly that I don’t mention on the Digest because they are relatively small, or not necessarily part of a larger plan. However, I’ll take a minute to mention the work by Sascha Wildner; he has kept the man pages in DragonFly up-to-date almost single-handedly, and done an excellent job. How good? See this excerpt from the IRC channel #dragonflybsd on EFNet:
(17:17:57) corecode: wow linux man pages are unreadable
(17:18:00) corecode: incomprehensive
(18:11:52) _hasso_: corecode: linux doesn’t have swildner ;P
YONETANI Tomokazu reports that DragonFly will boot on his Core 2 Duo MacBook Pro, though Daniel Tralamazza reports it won’t boot on his 1st generation MacBook Pro. I didn’t know this was possible…
Sepherosa Ziehau has made changes to the nfe(4) driver that, among other things, allow a card with that chipset to transmit data at full line rate.
If you haven’t already, you should make sure your NFS mounts can be put into the background.Â Sooner or later, it’ll save you a lot of waiting.
Sepherosa Ziehau has added ‘in-progress’ support for a number of Broadcom networking chipsets.Â Check the commit message for features, credits, and so on.Â Thanks, Sephe!
Peter Avalos has updated OpenSSH to 4.7p1.
Dave Hayes posted his scheme for upgrading OpenSSL on a DragonFly 1.8 system to the latest version.Â This is useful if you haven’t yet moved to 1.10.1, and want to avoid recent OpenSSL security issues.
Peter Avalos has added the option to not compile GCC3; this will shave a few minutes off a buildworld, and not hurt anything if you prefer GCC4.1.Â GCC3.x isn’t going away yet, however.
Linux Weekly News is reporting that AMD is planning to move away from binary driver support to an open-source driver, though the majority of the work will still originate from outside the company.Â That makes better 3D support on DragonFly at least possible.Â (via aggelos on EFNet #dragonflybsd)
Peter Avalos has updated libarchive to version 2.2.7
BSDStats.org is reporting month-on-month increases in the number of reporting clients for almost every BSD flavor; while it’s not the most scientific method of reporting usage numbers, it’s the most scientific one I know of.Â Â It’d be nice to see some more DragonFly hosts in there. (hint hint)
DragonFly has switched to IPFW2.Â Anyone want to tackle an upgrade to pf, too?
Undeadly.org has more coverage of the ongoing license issues.Â It is currently halfway between legally serious and Internet drama.
As a followup on the relicensing issue, Theo De Raadt wrote a description of the issues, with his central point: you can’t modify a license (i.e. remove a BSD license) without the agreement of the author. (via Undeadly)
Edit: Changed title for a better description.