Matthew Dillon commited changes to make the next release require pkgsrc.
Simon ‘corecode’ Schubert has put together a Mercurial repo of the DragonFly source code, for those who like their SCMs decentralized.
This doesn’t directly have anything to do with DragonFly, but it’s interesting: Sun is offering an Ultra 20 workstation free with a 3-year subscription to the services they offer. Those services cost $360 a year (which is how it’s billed, not monhly like they say), so it’s about $1K for the computer.
It’s certainly neat; while you could assemble a similar machine for close to a third of the price from off-the-shelf parts, it wouldn’t have the support, or run nearly as well. In terms of units, Sun is the second-largest Unix vendor around (here’s the first), and the biggest when it comes to server systems, as far as I know. If I worked with more Solaris machines, or more with Java, it would be an attractive offer. (First seen on The Howling Void.)
I took my own advice and made a little experiment: a cheesy status page for Joerg Sonnenberger’s pkgsrc binary site. The page lists the update time, in descending order, for all the packages on the site. (and links to them too.) Could it be better? Yep.
The default “Waiting for SCSI devices to settle” delay has been reduced to 5 seconds.
Jeremy C. Reed is writing an article, and he’s taking a survey of DragonFly users.
I have a number of small posts to link, and I’m placing them in one article: First, Joerg Sonnenberger mentions that he is updating his pkgsrc binary collection on an ad-hoc basis, though not everything is there yet. (Tracking updates is an afternoon project that would be enourmously beneficial, if anyone wants to try it…)
Also, Bob Bagwill contributed his own description of How Cabling Should Be. Last, Hiten Pandya commented on the idea of automatic crash reports, much like the “Feedback Agents” that upload crash information on the Windows platform.
Joerg Sonneberger wants everyone to try out pkgsrc if possible; apparently large projects like KDE and Gnome are building, mostly.
UnixReview.com this week has two book reviews: Network Administrators Survival Guide and High Order Perl (I saw the book author, Mark Jason Dominus, at a conference a while back, come to think of it), an article about IPC called “Networking’s Easier than Programmers Realize“, and an article about forensic CDs, which is a cooler way to say “rescue CD”.
Let’s say you want to debug a crash, but it was caused by a separate kernel module? Hiten Pandya has a way to use asf(8) to get at the module-specific data.
Questions about multiprocessor machines and routing ability led to this post from Matthew Dillon, who described the bottlenecks (and how they will be eliminated).
Just to follow up on earlier threads: the first part of the multiprocessor-safe network interrupt code has gone in.
Sergey Glushchenko asked a question about how the LWKT scheduler functions, and Matthew Dillon wrote up a rather detailed answer.
Do you use wireless? Specifically, the iwi, ipw, wi, or ndis drivers? Do you need WPA encryption? You need Andrew Atrens’ large patch.
Simon ‘corecode’ Schubert plans to commit this patch before the next release if he can get at least one person using one of each of the drivers listed above to test. That means before December 15th, so time’s a-wasting! Andrew Atrens has already been using this patch in production.
BSDCan 2006 is looking for proposals for (technical) papers, for presentation at their next event in May, 2006.
Matthew Dillon’s posted his first patch that can make network interrupts multiprocessor-safe. If you don’t want to run bleeding-edge code, it’s worth reading for the explanation.
‘walt’ wrote up a rather nice description about how debugging works, or at least how it can work.
UnixReview.com has several new articles: “Making a Dashboard Widget for Systems Administration Purposes” (for you Mac/BSD users), “Common Network Protocols“, focusing on Perl, and “John & Ed’s Scripting Screwups“.