Image: openclipart.org |
Managed network devices, like the ones I usually write about for work, usually have a way to message the mothership about various issues. Examples include:
- Hi, I’m online.
- The power’s out here. I’m running on my battery.
- Here’s some stats from the last connection.
- One of my components just failed!
The messages aren’t that chatty, of course, and they often include some variable data. Some are more urgent than others, and might require some action by the network operators.
I had separate topics describing each message, and they came out of the conversion tool as concept topics—a lot more generic than I wanted. As I was trying to get everything done at once, I didn’t give it too much thought. Since the messages were reference material, they would be fine as references. I split them into sections (format, severity, cause, action), and moved on.
DITA to the rescue? Um… nope.
Later on, I came back to the messages. “There has to be a better way,” I thought. After all, the sections could get out of order, or end up with different titles—there’s all sorts of ways to be inconsistent with reference topics. My next thought was, “Hey, DITA has hundreds of elements, and its prime purpose is software documentation. There's probably an entire message domain waiting for me.”
In reality, there are three message-related elements in the entire ocean of DITA, and two of them are inline (<msgph> and <msgnum>). The third is <msgblock>, for tagging message output.
Ah, the joys of information architecture. Creating a message domain from scratch was a possibility, but would likely be a hard sell to the co-workers.
We’re in trouble(shooting) now
I gave a moment to the idea of using troubleshooting topics—then it hit me. A message has a condition (the message itself), a cause (why it was logged), and a solution (what to do about it). That’s exactly the structure of a troubleshooting topic!
The only sticky point was where to document the message format, and I quickly decided that was part of the condition. I used @outputclass="message" to tag the topics, and to have the transform use Format: instead of Condition: for the condition part. I converted a few to troubleshooting topics, and it worked as well as it seemed it would.
On to the next thing
Then yesterday, I got a meeting invite with an attachment, a follow-up to a discussion a few of us had last week. One of the groups in our far-flung department uses InDesign to produce actual printed deliverables (how quaint!). The fun part is, the page size is about 4 inches square—so it’s not a matter of tweaking our transform plugin; we need a whole new one.
But when I started looking at it, the structure almost leaped off the screen, despite a couple of misplaced pages. Each chapter contained a single task, and each step used one page for substeps and graphics. Having that revelation made the call go a lot faster and more smoothly, because it was one of those things that are obvious once you see it. I just happened to be the first one to see it.
So I did a conversion dance, involving lots of pixie dust: PDF → Word, then Pandoc converted that to Markdown. After some serious cleanup (and moving misplaced content where it belonged), I used a couple of scripts to break the Markdown file into topics and create a bookmap. DITA-OT gobbled up the bookmap and Markdown topics, and spit out DITA topics. Thus, I had a pilot book we can use as test data for the transform.
The InDesign users also have a couple more formats; one is close enough to a regular book that we’ll have them use the standard transform. The other is a folded four-panel sheet… that one is going to be interesting. I’m going to have to resist the temptation of blowing off documentation work for glorious coding.
Stay writing… until I geek again.