Actually, there were a lot of personal firsts here: complete automated deployment and provisioning, complete test coverage, really good OO style, AJAX…

I’m really happy with how quickly it’s grown to thousands of gamers and I’ve heard lots of success stories of people setting up and growing their groups.

It hasn’t had a lot of improvement since its launch (aside from one badly needed graphic redeign — hm, it could probably use another), but there’s a lot on the way for fall 2009.

More info in blog posts.

1. I write really slowly, it took me 30-90 minutes each
2. I do graphics slowly, it took me 30-90 minutes each
3. Finding decent icons was painfully tedious, 15-45 minutes each

On top of the giant time sink, it’s really hard for a site to make any money by aggregating free content. All I can do is paste some ads around and their quality and relevance was uniformly low.

So I was doing lots of things I was bad at and not succeeding in attracting attention or revenue. At least I had the sense to kill this one young.

I could imagine restarting the site with a copywriter, a program to put together the icon set images, and user submissions of icons… but I still can’t see any way for it to earn back the time and money that’d take.

Because they didn’t exist in Ruby when I started, I also ended up writing a job queue and key-value store. Now that there are robust solutions out there, I’d really like to replace my homegrown solutions with more reliable dependencies.

I tinkered with this project most weekends from June of 2007 to August of 2008 and it’s still getting occasional tinkering, there’s always another edge case… It currently holds 8 GB of messages, so major bugfixes take a while to propagate through it. Ah well. It’s nice to have a project that has OMG MILLIONS OF ITEMS.

More info in blog posts, including a writeup of the user interface I’m quite proud of.

I was one of the game’s three active administrators from 2001 to 2005 and built the website (with help files, online player list, forums, code editor, world maps) and worked on whatever subsystem happens to need love. I also did a lot of work on reading and writing, the Mafia party game and the newbie school where players customize their characters and learn how to play. I helped train a half-dozen new wizards (programmers).

The mud has had bad luck with hardware failures, so the it’s been generally offline since mid-2009.

It never got very far, but it was a nice exercise in design and planning. I know it’s not much, but I felt like leaving up a little note in his memory.

In the stealth-based gameplay, you are a hacker in a network of computers and routing that’s both autogenerated and player-built. You advance by building up and protecting your own network and hacking through other networks.

Features:

• Area and room world, live-extended and -edited
• Persistence for players and the world
• Player stats: HP, Vitality, Strength, Stealth, and Perception
• Simple combat with attack, defend, hide, and regen
• Social commands like say, emote, and who
• Toggleable ANSI color with escape codes
• Commands can be recoded and reloaded live

During development the gameplay was actually about players building areas for each other to explore and conquer, but I couldn’t figure out a nice way to somehow balance an area’s payout vs. maintenance. Unfortunately, this code was lost over time.

In case you have a morbid sense of curiosity, you can view it on GitHub.

It was a significant project for me because it was the site I first did a good job of object orientation and first used unit testing. I’m also really happy with the graphic design, though I can’t claim most of it.

I was too immature to finish it despite the distractions in my life. I still think it’s a pretty neat idea and wouldn’t mind seeing it live. If you’re reading this and loving the idea, get in touch.

• WashingtonPost.com
• Post Politcs and Blogs
• Post: The Moby Quotient
• Post: Politics Glossary
• Choose Your Candidate
• Washington Post Update (if you only read one)
• Giving Notice
• Done at the Post

While working on those web apps, I fell in love with web coding. I’d been programming for ten years at that point and immediately took to the web’s design space. Six months later, I knew I had to go to college to pick up the theory to match my practical experience.

IBM runs the computers for Sears: all the cash registers are dumb terminals and there’s an IBM rack in the back of the store somewhere with the brains that handles transactions and end-of-day accounting. My team (roughly 10 at any given time) talked a tech in the store through unplugging old equipment and rebooting servers while we updated IBM’s database to look for them on the frame relay network instead of the SNA.

Of course, while we did this, the cash registers didn’t work. So we worked 4 AM to noon, starting with east coast stores and working west with the time change, to make sure we always had plenty of time to get the stores working before they opened. I think only twice in the 10,000 stores did we ever run late, and neither time by more than a few minutes.

I know this sounds like a terribly dull and dead-end job, but it was great: I had a great team, decent pay for my low experience, and I could see that we were making steady progress towards well-defined goals.

Despite the unpopularity Hostway had a contract to provide reliable service and I learned a lot about doing so as I maintained RegistryPro. I normalized the database schema, decomposed maintenance tasks, rebuilt unreliable jobs to act idempotently, and grew the backend from a single machine to a cluster. I automated as many jobs and reports as possible and then built monitoring into that automation to eliminate the need for human attention in day-to-day operations. In the fourteen months from April 2005 to June 2006 I removed 110 kloc (43% of the codebase), increased performance and reliability by two orders of magnitude, and similarly reduced support needs from two full-time engineers to about 5 hours a week for one. I can’t say I enjoyed working on RegistryPro, but I learned volumes about engineering services.

In the spring of 2006, when RegistryPro no longer needed the attention of a full-time developer, I worked on a web frontend for an internal application to monitor Hostway’s thousands of globally distributed servers. I learned AJAX and Django to build a website that alerted our sysadmins to downtime in real-time, presented historical outage data, and resolved trouble tickets. I had help with the interface design from Stu Young. Not long after I left Hostway the entire backend team quit and the project was abandoned.

Like my job with the student government, none of this is online anymore.

I maintained CambrianHouse.com (a PHP site eventually using the Zend Framework) by automating deployments (and shortening them from 60 minutes to 3), cleaning up broken HTML and CSS, adding JavaScript features, moving database access from views to functions and ORM, and adding lots of since-mostly-removed functionality in user profiles, content management, videos, voting, forums, and much more. I also did all of the programming behind IdeaForge, the subsite for developing crowdsourced business plans.

I entirely built the since-closed IdeaWarz.com, the selection engine of Cambrian House’s crowdsourcing of business plans. I took the original PSD comps and transformed them into a highly-interactive site that integrated closely with the main CambrianHouse.com (including cross-domain logins). It had a complex graphic design (which did not survive its transition into archive.org), user-submitted content, voting, and commenting.

I also helped out on various projects including Prezzle, NorseForge, Gwabs, Follow the White Laptop, Robin Hood Fund, others that never saw the light of day.

More info available in my blog posts.

The first Barking Stapler project is NearbyGamers, a “I made because I want to use it” service for tabletop gamers to find people to play games with. It’s been steadily growing a happy collection of gamers since its creation.

Icon O’Clock was a blog about icons, but it didn’t really take off and is now inactive.

ListLibrary is a highly-readable mailing list archive, another “scratching an itch” project.

Students: if you get the chance to work for your school’s government, jump on it. You’ll get to learn a lot about how your university works and improve it for your fellow students. It’s a great experience.

The list quietly died February 25, 2006, but a Mud-Dev2 has been organized to continue its mission with most of the original participants and some new faces.

Presentations tend to be great at this group.

This is an excellent technical group for any programmer — spend a couple minutes picking up Python and hurry to a meeting. There a number of very talented people in the group and interesting technical and theoretical topics are raised regularly.

In February 2006 I organized a social gathering so loop workers could pre-funk (on coffee) in the couple hours between work and meetings. This may be catching on as a regular occurrence.

I presented at the June 2006 meeting about my unicode-art project. There were only a dozen people as the meeting was announced late, but I walked through my motivation, code, and results to great response.

In January 2004 I gave a presentation on unit testing using PHPUnit and in February 2004 I gave a presentation on source control with cvs/svn. I have a hunch I’ve talked one other time, but I can’t recall.

ChiPHPug is a very friendly group, especially for newbies. We get a lot, I suppose because PHP generally attracts a lot. But there’s something for folks at all skill levels.