This whole "not working" thing

Several people recommended that I should take an extended holiday while I have the chance.

But I'm not sure that I'm cut out for vacations. I get antsy without a place to be and I feel like I'm slacking (or I worry that other people will think I'm slacking).

So vacation or not, the time off is a good opportunity to do some housekeeping.

I signed up for a new VPS server with John Companies. I requested the Fedora Core 5 installation, which I had just recently used for the (upcoming) website. FC5 is fantastic, especially when paired with yum (or apt-rpm) and the atrpms and dries repositories.

Even though I happily run Ubuntu on the desktop, and couldn't imagine going back to any other desktop Linux distro now, I just feel more comfortable with RedHat on the server. Their model of building community with Fedora and selling support for their Enterprise Linux package is the way to go.

With the server set up, the next order of business was to start moving things on around. A few years of server-rot had set in and it was time to redeploy. In particular, I wanted to try to get as close to out-of-the-box as I could. My normal practice was to remove most of the pre-packaged solutions in the web application stack (apache http, php, mod_perl, tomcat, mysql, etc) and roll my own. This was a deeply ingrained habit after years of being just a step or two ahead of the distribution packaging curve. (And building one custom component in the stack often forces you to build them all...)

But things have settled down. I was able to use the stock rpms of everything from apache to wordpress to mediawiki to to memcached to eaccelerator. This greatly simplified the deployment process. In fact, I am documenting the steps as I go in hopes of creating a guide that will help both me and others who want to recreate the setup.

The old site kept used the domain to host multiple different applications. The new server splits the subdomains out by application. For example, the blog now lives at and the wiki now lives at The URLs are shorter and cleaner and it will be easier to maintain this way.

Granted, the flurry of 301 redirects will kill my rank in the search engines for a few months. But really, it is worth it.

The replacement "www" site will host a simple aggregated view. I will pull in bits of the blog feed, the wiki recent changes/new pages feed, my flickr feed, feed, etc. I am planning on using Planet as a starting point, but don't feel strongly one way or the other. I will post the code for whatever solution I come up with.

In making the changes I also realized that I want to rewrite the Spring (currently down) and AWS OpenSearch applications. Both applications wrap third-party APIs and translate them into OpenSearch feeds; I can likely combine them into a single application. Perhaps this is a good excuse to learn Ruby (sans Rails).

Here's a question: is there any interest in an open source, modular engine that makes it easy to publish OpenSearch feeds wrapped around existing APIs? And if so, what language would you want it written in? The target audience would presumably be customizing/extended it for their own needs, so it would have to be a language that enough people were comfortable with. (Perl, one of my native tongues, seems to have fallen out of favor...)

Between the OpenSearch "wrapper" engine, the new homepage, seeing launch (knock on wood), the two long-term projects I want to kick off (more on that tomorrow), and spending as much time meeting with people in the Bay Area as I can, it looks like I won't have time to worry about slacking off.