David Tweed posted a short but interesting anecdote of his real-world experiences dealing with a large number of files, to follow up with a recent discussion on handling large directories with Hammer.
Will Backman has another podcast up; this one being 38 minutes of FreeBSD Core Team interviews from the just-concluded MeetBSD event.
Will Backman visited iXsystems recently, and he has a 8 minute podcast with the details up as BSDTalk 163.
Matthew Dillon is adding versioning support to Hammer; it’ll support in-place version upgrading. The gory details of his current plan are available, with an interesting tidbit: Hammer directory lookups remain the same speed even with 2 billion files in a directory, while UFS will be O(N^3) speed after several hundred thousand.
I have a number of things to link which probably can all go together:
- Useful (Stupid) BlackBerry Tricks, to go with the previous ones for Unix, Vi, Emacs, and Regexes.
- Dru Lavigne brings work of the November OSBR issue: “Health and Life Sciences“, along with something else I didn’t know the BSD Fund was supporting: Events.
- BoingBoing is having an Obfusticated Code contest; I seem to have heard of this sort of thing before.
- Two links from sjg on EFNet #dragonflybsd: tarsnap and Sun’s hybrid storage plans. (PDF)
Since DragonFly is switching to git instead of CVS, something handy is ‘eg’, or Easy Git. It’s a wrapper around git that makes the transition from CVS easier, or so it says. (via _hasso_ on EFNet #dragonflybsd) The linked page lists some alternate programs that are also designed to make git acclimation easier.
The change from CVS to git will be happening this week, with git being moved in and mercurial added in a mirrored form, so both will be available. Expect some wierdness on the commits@ mailing list.
Some random links I’ve had built up:
- Dru Lavigne has links explaining cross-platform zip differences
- Microsoft is getting in on the idea of an App Store, same as Apple and Google. I want to point out that you can draw a straight line between the BSD world’s ports/package systems and this idea…
- pcc is seeking funding. (via)
Fresh from the howling void: (Useful) Stupid Unix Tricks and (Useful) Stupid Vim Tricks. Caveat Emptor.
Update: Emacs too.
It looks like there’s already BSD-specific patches for OpenJDK according to commenter Samh; anyone want to see how well these would work for DragonFly?
Also, does someone want to work on interrupt routing for DragonFly? It’s tough and necessary work, but there are enough people that need it that there’s a potential bounty of $400 or more.
Stephane Russell pointed at work bringing the OpenJDK to the BSDs – anyone want to help out?
Thomas Nikolajsen turned the slides from Matt Dillon’s NYCBSDCon talk about Hammer into a PDF.
Marc G. Fournier reported hitting 25,000 BSD systems checking in to bsdstats.org; most of that is PC-BSD, where the bsdstats client is on by default. (It’s present in DragonFly but not on by default.) Some cross-posted acrimony followed, thankfully not from DragonFly users.
OpenBSD 4.4 is out, and OnLAMP (as usual) has a developer interview to match. They touch on a number of products that are also used in DragonFly, like the sensors framework and pf. (via)
Hey, there’s a part 2 to the @Play coverage of the the devnull NetHack Tounament! (Part 1 mentioned here if you missed it.)
Since it was mentioned on the mailing lists, I’ll mention it here: instructions on building a vkernel are in the Handbook. I may not have linked that before.
The latest @Play column on GameSetWatch talks about something I didn’t know existed: a NetHack tournament. Given NetHack’s difficulty, the scores it describes are insane.
I don’t recall if I mentioned this before, but the Google Summer of Code software (the part that Google manages) is now an open source project, for anyone to participate in. If and when DragonFly participates next year, this application is how it will be managed.
The discussion over Git vs. Mercurial continues; Jeffrey Hsu has even volunteered himself to maintain and synchronize the two repositories. He also pointed out that there is precedent for this already: the git-using Linux kernel work has a Mercurial mirror.