This month, I started developing a new Atlantis game, code-named Helios. I am hoping that other developers will join my effort eventually, so I am spending more time than usual on project management and documentation. Today sees the end of my very first sprint, and I am pretty stoked about how much I got done.

Project planning

In agile development, the project gets broken down to sprints — intervals of time that are easy to plan for. In my case, I am planning to do a print per month, because it forces me to prepare a ton of design work, and because my time is easier to predict for a month than a week. I think I can commit to 5 full days of work during a month while still having a job.

For each of the sprints, I pick a few user stories, break them down to bite-sized tasks which can be finished in less than a day, and assign as many of these to the sprint as I think are possible to finish in a month. Since the tasks are small and the design for their user story is complete and written down, new contributors should be able to pick up any one of them that they feel comfortable working on. For this purpose, I keep my task board on Trello updated.

Even as a single-team developer, this helps me focus on my work. There is satisfaction in seeing how many tasks I have moved, and I never wonder what needs to be done next, so I don’t work on stuff that isn’t important.

What got done

I have planned an aggressive list of features for Helios, and I wrote Design rationale and a task breakdown for all of them. This gives me material for all my future sprints until the start of the game.

The core feature I worked on for first was the most difficult one. I believe in addressing risk head-on, so I worked on multi-hex movement for ships. Talking to the atlantisdev folks helped me narrow the scope to only ships, which don’t have to worry about guards, terrain modifiers, and other complications to the source. I also decided to delay the new action system I have planned, and in the end it turned out to actually be pretty straightforward.

I got a lot of other small and large things shipped. My other focus was making more things configurable on a per-game basis, by moving them into the json configuration file:

• players start with 10 men and $2000, which are configurable values.
• the width/height of the map is configurable, but stays fixed after world generation, and does not need to be recalculated every time the data file is loaded.
• Regions have a unique id, and the neighbor-graph of all regions is stored in the data file, which makes coordinates a merely cosmetic attribute of a region. They are no longer needed to rebuild the region::connect array.
• All regions are stored in a hash table, so they are fast to look up if you know their unique ID.
• Ship classes are configurable, with speed, cost, and capacity. The game is no longer limited to only three types of ship, although the Helios game will probably still have a longboat, clipper, and galleon.
• I killed the original world generator, leaving only the code that was used in Atlas. This will eventually be modified, too. It looks like the next sprint is going to have world generation as one of its themes.
• Terrain types are configurable. For Helios, all production in a region has been reduced by 80%.
• The code compiles on stock Ubuntu 13.04 again.
• I compiled and familiarized myself with the ALH code.

What did and didn’t work

The atlantisdev group is a great sounding board for ideas. I am not a game designer who creates things in a vacuum, and talking about my plans with other people helps me a lot.

I am still looking for other developers to help me. Two people have stepped up, and I have added them as collaborators to the Trello board, but none of them has started on any task work.

Communication with other developers is mixed. I started an #atlantis IRC channel on irc.lunarnet.org, but only one person has been showing up regularly. Hopefully, this is going to get better. Please, if you are interested in Atlantis developments, even just as a lurker, I want to talk to you.

Investigating ALH helped me get some ideas about world coordinates sorted out, realizing that the code didn’t need them, and as long as the client can draw a map, neither do the players. I am probably going to let players choose how they want them laid out on a per-faction basis, with a default preference for the ALH layout.