Tower Defender Post-mortem

Posted by (twitter: @GBGames)
August 21st, 2008 5:06 am

My 2nd Ludum Dare didn’t go as well as my first. While I managed to get Tower Defender submitted, it can’t be called a game so much as a tech demo.

What Went Right:

  • Simple game mechanics still work.

    Like my LD#11 Minimalist entry, I wanted to use simple mouse-movement-only controls. I feel that mousing over your units to make them attack made sense, and while I only had archers available in the end, it seemed to work. It’s too bad there wasn’t more of a game built around the mechanic, but I intend to flesh it out after LD.

  • I had an office door I could close.

    My cats are incredibly reliable. If I am doing anything that looks like productivity, they will insist on sitting on my lap, resting on my arms, and otherwise preventing me from working. Being able to close the door on them helped keep me focused on game development. Towards the end I got lax about keeping the door closed, but the cats left me to work for the most part.

  • Using Test-Driven Development

    Test-Driven Development, or TDD, is great for designing your code. Also, since code changes often, you can feel confident that your changes won’t break functionality since your tests will tell you if they did break. More than once, I was surprised that a seemingly innocuous change resulted in failing tests, so I was able to keep the game working at all times. I know that I wouldn’t have caught one specific crash problem right away, and it might have resulted in a non-working game for hours, preventing me from submitting anything. Since I found those problems sooner, even in code that wasn’t directly being tested, I felt that using TDD was the right thing to do.

What Went Wrong:

  • Learning Test-Driven Development while using it.

    I know quite a few people would disagree with the use of TDD during Ludum Dare, but I think what burned me was my inexperience with implementing it. I spent too much time trying to figure out how to apply it to rewriting code that I already had written. My first bunch of tests were helpful, but all I ended up with at the end was a slightly smaller Game class with a separate Timer class, and it seemed that if I applied TDD to the entire project I would barely have an SDL window by the end. While my normal projects might benefit from test-driven design, my LD game needed to get finished in 48 hours, so I had to alternate between writing tests first and skipping tests. I’m sure once I get some TDD experience, I’ll be much faster and know when it is in appropriate to write tests. For LD#12, it was a learning experience.

  • I still didn’t have a good handle on SDL

    Last LD, I noted that I hadn’t practiced using SDL much, and right before LD#12 started, I realized that I still hadn’t done so. I never had to render animated sprites in SDL before, and I skipped it in favor of static images moving around, but not before spending precious time learning what I would need to do it. Again, there was too much wrestling with technology instead of game development, and this time it prevented me from finishing my game.

  • Working long hours really does screw with your productivity

    It’s common in the programming world to find people working Twelves, especially in the game development industry. Crunch times are intuitive. If a project needs to get done in a week, and there are two weeks of work to be done, then have everyone work longer each day. Well, it is common knowledge, even if that knowledge isn’t applied, that working longer hours doesn’t translate into greater productivity.

    I experienced these issues firsthand with the 2nd day of LD#12. I realized I had worked about 12 hours straight by the end, and I was making sillier and sillier mistakes. Sometimes my tests would save me, but since I didn’t write tests for a good portion of my code, I had to figure out what I did wrong most of the time. Bugs were finding their ways into my code a lot easier, and debugging was painful. When I did LD#11, I got plenty of sleep and took frequent breaks, and ended up with a finished game. I wonder if I could have done LD#12 better if I took a few more decent breaks during that 12 hour stretch.

  • I didn’t get game play until the very last minute.

    I knew that getting game play up as quickly as possible was important, especially in a timed competition, and yet I believe I struggled so much with the technology that the game didn’t start to form until I had minutes left to package it up and submit it. I think if I had used a few more hours in a productive way, I could have made something enjoyable.

What I Learned:

  • I still have a lot to learn.

    It’s weird when you feel confident going into a competition like this and then hit a wall due to your own lack of knowledge. I was depending on TDD, SDL, and common game programming concepts such as OnMouseOver, but I didn’t have much experience with them before this competition started. I like using LD as a learning experience, but next time I’ll focus on learning only one tech or tool for LD at a time.

  • Test-Driven game development is awesome.

    Yes, the learning curve slowed my productivity down, but I already saw many benefits from using a test-first design for my coding. I could see that my code base was going to be much better for it, particularly in terms of my ability to make cross-platform games, but I had to stop applying it due to time constraints. I was already trying to incorporate TDD into my main development before LD, but now I see that it’s going to provide better benefits than I originally thought.

  • I need to work on my pacing for LD.

    It seems most of my productive work happens during the 2nd half of Ludum Dare, and it makes me wonder what happened during the first 24 hours. I saw that more than a few people had working prototypes up and running within a matter of hours, and I want to make sure my future LD entries are in a playable state as early as possible, too.

Once again, 48 hours resulted in a bunch of code and experience I didn’t have before the weekend started. Even though my submission can’t really be called a game, it has potential, and I had a lot of fun working on it. The next LD is in December. A few months should give me time to develop my skill and technology base.


Leave a Reply

You must be logged in to post a comment.

[cache: storing page]