This is a big long nerd brain dump about behind-the scenes software stuff on this website. Even if you're not a web geek, there's a possibility you might find some of it interesting. But if not, you've been warned.
UPDATE: A couple of days later, I may have changed my mind about my apparent decision below. Find out more.
Since Blogger announced the shutdown of its venerable FTP publishing system a couple of months ago, I've been working to figure out what new system I'm going to use to publish my writing here. (I'm glad Blogger postponed the shutdown for an extra month, but of course that meant I simply procrastinated about it until now.)
I have about three weeks to make the change, and I've boiled it down to two options, neither of which is ideal, but both of which are serviceable:
They are two of the most popular and long-running blogging platforms on the Web. There are other possibilities, but I don't need all the complexity of Drupal, and don't find other options like ExpressionEngine all that compelling. Movable Type has always used the static files publishing model I prefer, while WordPress requires a plugin like Really Static to hack it into doing what I want.
However, Movable Type's day in the sun may be past. While some high-profile sites of people I know—such as John Gruber and Dave Shea—use it, its popularity seems to have been in general decline since the licensing controversies of version 3, way back in 2004. The current version 5 (MT5) is brand new, and an open source project, but I don't sense the same community vibrancy and wealth of third-party extensions WordPress has. Six Apart, the company that created Movable Type, also seems to have been focused on its other hosted blogging tools, TypePad and Vox, for years.
There's a fork of Movable Type called Melody too, which is cool. But forks like that tend to arise when the originating platform is losing air. Yes, I know there are lots of people who love it, but I just get the sense that general enthusiasm for Movable Type has faded—even in the vibe I feel after installing and playing around with MT5 last night. The basic software is great, mature, and solid. But when I want to muck around and extend it, the available resources are a little sparse and often out of date.
Just when Movable Type stumbled in 2004, Matt Mullenweg's WordPress—itself a fork of the awkwardly-named b2/cafelog blogging platform—was hitting its stride. I know Matt a bit, and have been using WordPress on other websites (most notably Inside Home Recording and Lip Gloss and Laptops) since 2006. I like it and recommend it to friends, despite its sometimes-sprawling nature, and its reliance on a dynamic, on-the-fly, database-driven publishing approach that I find somewhat brittle.
Indeed, that dynamic approach really was the only thing keeping me from switching to WordPress right away. I understand WordPress and how to tweak it, I like the wide range of themes and plugins available for it, most of my geek friends use it in some form or another, and the community is second to none. Recent versions are also very easy to upgrade in place, which is a big improvement over the way most installable blog platforms (WordPress and Movable Type included) have usually worked.
So when Matt Mullenweg's colleague, Victoria-based Lloyd Budd, pointed out the Really Static plugin to me, it looked like a perfect solution. It takes a regular WordPress blog and generates plain-old text files which otherwise continue to look and work pretty much just like the original WordPress pages. Nice.
Therefore, last night, after all that procrastinating and evaluating, I installed both WordPress 2.9.2 (with Really Static) and Movable Type 5.01 into test directories on my server. The WordPress install went smoothly, since I've done it before. Movable Type took a couple of tries, but I got it working without too much trouble. Disabling either software installation seems to leave the resulting static blog pages essentially intact, which, after all, is my key criterion in this whole production.
As I said above, I found Movable Type underwhelming. I really, really wanted to like it, because it would be interesting for me to learn how to work with and tweak a new publishing system. The default appearance theme is certainly nicer than WordPress's, and its more modular system of templates and styles is also a bit more elegant. Though not a key feature for me, it's much easier to publish multiple blogs with a single Movable Type installation (at least for now). Some tech-head friends I respect a lot think its Perl-based CGI architecture is inherently better than WordPress's collage of PHP scripts.
All that may be true, but MT5 looks, to me, like it's catching up to features and polish that WordPress has offered for at least two or three years. Searching for some alternative themes and styles, as well as fairly simple plugins (displaying my recent Twitter posts on my home page, for instance), didn't yield very many options, and most of the results I did get seemed to be talking about MT installs a version or two old.
Not that the WordPress option is perfect. The forest of PHP files that WP uses can make heavy customizing kind of a chore ("which of those 250 files was I supposed to edit again?"). The Really Static plugin remains a hack, though an effective one, so making it do what I want requires some duplication of template files, some advance planning of how I want to structure my blog and archives, careful pruning of HTML pages on the server if I choose to delete something, and the awareness that when WordPress releases a new version (like the upcoming 3.0, or even a service update like 2.9.3), I might have to wait to make sure Really Static plays nice with it.
So right now, I'm leaning heavily toward the WordPress/Really Static approach [but wait! see my April 14 post to find out if I changed my mind]. Any Movable Type advocates (or people with different suggestions) who want to convince me otherwise can email me or leave a comment here—at least until I disable comments on this post at the end of the month (see below).
One annoying thing about moving away from Blogger as my publishing system is that I'm going to have to lock down my blog archives. What I mean is, I'm not planning to import all my nearly 10 years of existing blog posts into the new system and republish them. Indeed, one of the advantages of the static-files approach I'm choosing is that I can just leave my old posts exactly as they are, whether from 2001 or 2009.
But since I'll no longer be able to update the pages from Blogger, that also means that no one will be able to post new comments to those posts—or, more accurately, if I leave things as they are, people can write comments, but they'll never show up on this website. So my plan, over the course of the next three weeks, is to disable new comments on my old posts, and gradually disable them on newer and newer ones until FTP publishing stops working at the end of the month. Then, if I time things right, I can seal off comments on my latest entries, tie up the old blog in a bow, fire up the by-then-ready new system, and be done.
I've already started. Want to comment on one of my posts from April 2007 or earlier? Sorry, you can't. Same for entries for my occasional Penmachine Podcast from last year or before. They were never much for comments either.
Yeah, it's a bit of an awkward transition, but after a decade of largely smooth sailing with Blogger, I can hardly expect anything else. I'll miss the simplicity and familiar orange-and-blue colour scheme of writing in Blogger, but I won't miss its bizarre system of labels, its strange way of handling podcast enclosures, and of course the consistent unreliability of FTP publishing in the first place. Besides, for the geek in me, making the change is sort of fun.
As long as I don't screw anything up too badly over the next few weeks, anyway.