So Adrian notes that the Census Bureau keeps statistics in a handy searchable web form. And I'm nowhere to be found. And I know I filled out the form and mailed it back, even though it didn't promise that I still had a chance at winning the twenty million dollars.

According to the docs, I must not be on the list because my name is too unique. And the fact that I'm still alive must be more difficult to figure out than that.

And suddenly I wonder how people in the Witness Protection Program handle the census. I mean, do they fill in their fake names?

Looking at today's new bugs and finding nothing. It's 12:53 A.M., and I've already lost.

I bet you'd like to think that /usr/include/pam.h has something to do with the PAM library, from the "pam" package.

3:31pm nalin@rapier:~> rpm -qf /usr/include/pam.h
netpbm-devel-10.33-2

Awesome.

So, applications are actually supplying the data, but I got confused by the Python bindings, which isn't surprising, really.

I've just finished tagging oddjob 0.25 and am currently building packages for Fedora Extras. I screwed up on the FE tagging, so there's some release number inflation going on, but at least I can work on other things for a while.

I think I ruined V for Vendetta for some friends when I summed up Hugo Weaving's performance as V, the title character, like so:

• Covered from head to toe in a costume. Check.
• Fights bad guys. Check.
• Emotes by tilting his head and moving his arms. Check.
• Explains to the bad guys that he's about to beat them up. Check.
• Employs wire-fu. Check.
• Unnaturally strong. Check.

Yep, V's a Power Ranger. Too bad he didn't have a giant robot. That would have been awesome.

Spent part of last week and the large chunk of today fixing up oddjob's support for reloading its configuration without a full restart. The guts of it are working in CVS now, but I'm leery of memory leaks, so I won't tag a new release yet.

Part of why I'd put off doing this earlier was that the D-BUS daemon itself didn't support restarts. Or so I thought. I was toying around with a prototype for a graphical browser for introspection data (summary finding: few applications even bother to provide any) and was shocked (shocked, I tell you) to find that dbus-daemon provides a ReloadConfig method. The man page, however, will only say this:

SIGHUP will cause the D-BUS daemon to PARTIALLY reload its configuration file. Some configuration changes would require kicking all apps off the bus; so they will only take effect if you restart the daemon. Policy changes should take effect with SIGHUP.

which is quite maddening when you don't know if the part of the configuration which you absolutely need to have reloaded is part of the, er, part which actually gets reloaded.

Anyhow, one of the problems the browser ran into was that oddjobd wasn't providing data for the root object (/), and without a place to start, a browser would be dead in the water. So I fixed that, and now, provided your configuration's ACL allows it, your application gets a complete tree of objects. Woo-hoo!

LinuxWorld Expo hit Boston last week. I initially thought the turnout was pretty decent, only to later to be shown that I had been mistaken — it was in fact rather slow.

At least this year the conference organizers dispensed with all of the fill-out-the-form business from last year, which made it easier to attend a few of the technical breakout sessions. There were some real jewels there (Jeremy Moskowitz, Gerry Carter), but overall the conference schedule didn't exactly wow me.