A question about saving dumps from a crash led to some explanations of the process.
The 2006Q3 pkgsrc release is coming up, and Joerg Sonnenberger has built a good number of the binary packages for DragonFly using that release. Please test and report issues.
As part of a larger discussion on users@, I wondered what can be done to reduce power usage. There’s different options available, but it doesn’t matter much unless you have multiple servers.
The tersely named chsh command is what changes a user’s shell: Gergo Szakal has the best description of several on how to use it.
BSDCertification.org has the results of their “Test Delivery Survey” available as a PDF. It “summarizes the results from a recent survey of potential testing candidates to determine their geographic locations, the price they are willing to pay to take an examination, and their thoughts on various methods for delivering IT certification exams.”
It’s mind-bogglingly complete like many of their reports, though I question the idea of surveying to see what price people will want – the only price anyone can agree on is ‘free’; nobody volunteers to pay more money, no matter how realistic the price.
Welcome to the newest committer: Victor Balada Diaz.
OnLAMP.com has a 3-page interview of Charles M. Hannum, recently known for describing NetBSD as ‘stagnant‘.  DragonFly gets a slight mention.
A conversation about NFs led Matthw Dillon to give a short description of locking under NFS for most any operating system: generally broken. DragonFly clusters ought to, in contrast, work.
After making some adjustments, Matthew Dillon created a new test program for his virtual memory page tables.
Are you a DragonFly user who can give a talk at NYCBSDCon, October 28th-29th? If so, contact Matthew Dillon.
As Oliver Fromme found, bsdstats.org has a script that reports on BSD usage. It’s easy enough to run on your own, though it could be added to the base DragonFly system. (which would no doubt affect the per-flavor BSD scores on the site.)
Want something to do? Bring dcron to feature-parity with vixiecron (check end of post), or shave some time off the DragonFly boot process.
Adrian Nida was looking for a ‘how-to’ for Postgres. Petr Janda supplied one.
If you’re unfamilar with Postgres, Postgres is to MySQL as BSD is to Linux.
UnixReview.com has a light week, this week: Exploring the RFID+ Certification and a review of the book CCNA Official Exam Certification Library.
To go with an earlier post about rc and launchd, Rahul Siddharthan described a similar tool from Ubuntu Linux: Upstart.
Strangely, one of the listed reasons for Upstart is that Apple’s launchd isn’t free enough in GPL terms, but it’d probably be easier (in licensing terms, thanks to the BSD license) to integrate launchd in DragonFly than the ‘more free’ Upstart.
Sepherosa Ziehau has a description of how to set up your DragonFly box as an 802.1x client, whether wired or unwired.
Rahul Siddharthan asked why BSD shutdowns happen so much faster than with Linux. Out of a number of responses, Oliver Fromme seems to have the most complete writeups.
While on the topic, I posted a note about Apple’s replacement for rc/init/inetd: launchd. It has been partially ported to FreeBSD, though it’s not in the base system.
Delete files that start with – by preceding the filename with two dashes. Someday, you’ll thank me and the 5 people that answered this question on users@.
Julio M. Merino Vidal is working on improving GNOME support in NetBSD and pkgsrc; this has some side benefit for DragonFly, since we use pkgsrc too.
Seen today on Slashdot: A longtime Debian developer is leaving the project, saying, among other things, that a more direct leadership structure, similar to Ubuntu, would prove more effective. Compare that to Charles Hannum’s “NetBSD is stagnating” message, where he also says a stronger leader for NetBSD would help.
This idea matches up with one of my favorite books: The Mythical Man-Month, where Frederick Brooks mentions that a software project should be led by an experienced worker, rather than by committee. It is also similar to the Linux kernel development model (though there’s plenty of other factors that affect it) and other things, like Perl’s pumpking.
On the other hand, there seems to be a cycle where a particular Linux distribution becomes ‘cool’ for about a year or two – Debian, or maybe Slackware, then Red Hat, then SuSe, then Mandrake, then Knoppix, and now Ubuntu. Yes, it’s an inexact timeline.