LISA Progress Report

I'm at LISA again, in Dallas this year (which makes my fourth trip to Texas in the last few months). I ran a configuration management BoF on Tuesday night and a Puppet BoF last night. Both of them were well-attended, and we maintained good discussion through almost the entire two hours in both cases without having to resort to presentations, which I'm very happy about.

LISA itself doesn't seem to have changed all that much over the years. In fact, someone complained to me that USENIX should be prescribing antidepressants before the config-mgmt BoF because it hasn't changed since he was last there four years ago (he did allow that Puppet was a significant change, fortunately). I unfortunately agree -- we're still discussing what the term means and how we should move forward.

This year, the biggest source of discussion in the BoF was whether and how to support ad-hoc, manual administration in an automated world. Most people in the room who spoke up wanted their automation tools to support it, but to me it's like asking compiler writers to support writing in assembly when necessary. My perspective is that assembly is a separate problem; if you want to write assembly, then do so, but don't expect my compiler to know or care that you're doing so, and certainly don't expect my compiler to extract semantics from the assembly that you wrote.

I know I've barely been blogging, and I'm going to try to fix that. I've been moving painfully slowly on this latest release, and I'm embarrassed enough at the lack of progress that I haven't wanted to publicize that, but I realize that that's a big mistake; I should instead be advertising what I'm doing and what problems I'm having (and solving), so people can clearly see what's being accomplished, even if it isn't in the form of a release.

So, hopefully, I'll start blogging more, including discussion of what's going on development-wise.

I'm also planning on blogging the test Puppet scripts that I write - these are simple scripts that I use to verify behaviour manually. For instance, here is the code currently at ~/bin/test.pp:

class yayness {
    $testing = funtest
}

class other {
    include yayness
    $value = $yayness::testing
    notify { "my value is '$value'": }
}

include other

The purpose of this was to test that the include method was no longer doing lazy evaluation.

I've also just joined Dopplr, which helps people who travel track their friends' travel, hopefully helping them to meet up in the various destinations. I use TripIt to organize my travel, but Dopplr is really a different kind of service and is relatively complementary -- TripIt is useful for keeping track of all of the details I need when traveling (confirmation codes, times, etc.), while Dopplr can hopefully allow me to meet up with friends who are in town while I'm traveling.

Thu, 15 Nov 2007 | Tags: travel, lisa, lisa2007, usenix, dallas, conference, puppet, lisa07