And now I need icons....

Well, the time has now come that I've been dreading. I need to figure out how to come up with some icons to represent each of Robo-Ninja's powerup abilities. I can probably find some free art resources for the general template of these icons, but I'll still need to do a bunch of graphics work to make the icons themselves. And come up with a consistent look between them.

This is going to get ugly.

So if anyone out there feels inspired to help me make some icons, let me know.

And sometimes things are easy.

Occasionally it turns out that my code design is actually decent, which means that on those rare occasions, adding things is easy.

Halfway through my cave level, I decided it needed boulders dropping from a hole in the ceiling that you had to dodge (by timing jumps when they go by). I didn't have any code in place yet for moving obstacles, although I had vague ideas about eventually including them, so luckily I had at least designed for the possibility.

Turned out, it was REALLY easy. It took about 15 minutes to find, download, and edit a Creative-Commons licensed boulder sprite. About 20 minutes to write the code for the boulder itself. Then about 15 minutes to write the code for initializing the boulders from the tile-editor level definition. BAM, the whole feature was working in less than an hour.

That felt good, after wasting so long reworking my collision code the other night.

Next I'll work on the 2nd cave room, which includes the first major ability powerup item. So I'll have to write the code to handle the UI for switching abilities. And I need to hurry and do that, because Aaron has been pestering me all day to hurry up and add a new ability. Well, when he's not running around the house jumping over things and pretending to be Robo-Ninja.

And this is why I do level design early.

So I started on a harder cave level, with some pretty hard and messy jumps and collisions.

And realized my collision code wasn't good enough. 

I had originally gone for an overly simple collision model, which I hoped would be good enough: check for collisions at each corner of Robo-Ninja, and deal with whichever is the "most important" collision.  Well, as Robo-Ninja is taller than a single 16x16 "block", I was running into instances where the head would go above the block without colliding, then the feet would collide with the block from beneath, and all sorts of brokenness would ensue.  This also got me into trouble when I put ramps too close to other collisions, as it would get confused about what is colliding where.

So tonight I overhauled and reworked the collision system. Which, once again, leads me to my previous statement: Ramps are weird. It only took me about 30 minutes to get the basic collisions working with my new model (checking multiple points around the perimeter of the bounding box, processing each point's collision separately).

Then another 2 hours to get ramps working. Now I'm tired. The good news is that the ramps work more smoothly than before. Previously, going down a ramp, you'd run too far ahead, your feet would leave the ground, then you'd bounce down a bit, then run too far ahead again, and so forth. Which was bad, because it made it really hard to jump while running down a ramp. (you had to time it perfectly when your feet hit the ground). Now, your feet lock to the ramp on the way down. 

So now I can continue on with my hard cave level. It took me about 25 tries to get through it while testing. My son was extremely excited about the existence of this cave level, but was rather unhappy that it was so hard.....

Working on some early level design

So I decided to put the actual programming on pause at this point, and do some early level-design. Mainly because I don't actually enjoy the details of level design very much. I really enjoy drawing a big rough map of the world, and planning out the overall gameplay. But the details of each little level? It's tedious.

Which is weird because when I was a kid that dreamed of making video games, I loved drawing out each level in minute detail on paper. Now that just bores me. Maybe because when you are 8 years old and drawing levels, you don't have to pay attention to making them the appropriate difficulty level, and interesting to look at, at the same time.

Anyway, so I've started on the first couple of "real" levels/rooms in the game. The first is pretty simple, without many things that can kill you. The second is full of death. And I'm realizing it's going to take a lot of work to make the challenge level right for this game. I'm not yet good enough at my own game to know which jumps and arrangements are going to be hard, and which are going to be easy, and which are just impossible to get through. There's one spot that I had to put in a checkpoint/save-point just for testing, so I could play it over and over again just to see if there was any way to get through it (yes, there is. It's not easy though).

So I've got a couple of possibly-finished levels (out of about 50). But that's why I started now. If I finished all the programming, and just had level design to do, I'd get so bored. So I need to pace myself: work on a few levels, then go back to coding. Work on some levels, then back to coding.

My current plan is to make the levels up until the first major ability item, then actually add the UI code for managing items. Then I need to write the code for the minimap. Which, I learned from Anguna, is a giant pain.  Hopefully I'll be smarter in how I write it this time around.

I also found yet another bug with my ramp code. Robo-Ninja can't jump while running down a ramp. Because I depend on the game's gravity to lower you down the ramp, it thinks you are in a freefall, so it won't let you jump. Oops.

Silly Mistakes

So I'm realizing that with this project, I don't have as interesting things to talk about as I did with Anguna. Anything that takes more than 30 minutes to implement in Robo-Ninja is usually because I made a silly mistake, not because I have to do cool stuff to make it work.

Tonight I was working on handling the back button on the Android build.  I wanted it to pop up a simple exit-confirmation type of menu.  LibGDX provides a nice handler for catching the back button (see the docs and this post on Stack Overflow), and it was pretty easy to pop up the menu. Quitting the game or canceling the menu were both no problem, but I kept having trouble whenever I'd use the "Main Menu" button.

No, this is not the final visual design for the menu. Yes, that button says Robo-Nina.  Oops.

I could get it to take me back to the Main Menu without issue, but the next time I'd hit the back button, the menu wouldn't appear. I spent a good 30 minutes stepping through with the debugger before I figured out my mistake:

When I started the game over, I instantiated a new UI/menu overlay object that controls all the on-screen popups (the Hud object). My back button handler, which I didn't re-instantiate, was holding on to a reference to the old Hud, which wasn't being displayed anywhere. So the old Hud, off in limbo-land, was trying it's hardest to show me a menu, it's too bad I wasn't actually DRAWING the old Hud.

LibGDX

Ok, I'm trying again to talk about LibGDX.

When I got it in my head that I was going to make a new game, I wanted to make my life easy, and actually have hope of getting it done in some reasonable amount of time, which meant using an existing game framework/library.

On Anguna, it was a lot of fun building everything from scratch, but I spent A LOT of time just doing the basics: animating sprites, building characters out of multiple sprites, getting backgrounds to scroll properly, etc.  On an device like the GBA, learning the weird tricks to do all that was a lot of fun. This time, on a modern smartphone platform, I just wanted to spend my time focusing on the game.

So I started evaluating different tools.  Unity and Marmalade (which used to be called Airplay, which I originally started using for the iPhone port of Anguna) are the two that seem to have the most use, but are also a bit expensive these days for a hobby project that might never get finished.

So then I stumbled upon LibGDX. It's a free open-source framework. You write your code in Java, and then it can be deployed to Android, Desktop, Html5, and iOS (using RoboVM). While Java has its warts, it's easier to be productive in than straight C or C++, which are often the alternatives. And doing my daily development on the JVM is really nice, as I've still never found a better platform for debugging code. Being able to hot-swap code during a debug session, then drop back to the stack frame and re-test makes some things SO much easier to get right.

Some other things I liked about LibGDX:

- It's open source, and has a pretty friendly and decent community. I like the idea of open source, particulary for hobby projects. But some open source communities and personalities are downright obnoxious. This one seems to be pretty cool. And they do a decent job of documentation.

- There's good support for existing tools (ie loaders for Tiled and other map editors), so I don't have to spend as long figuring out how to massage my assets to get them into the game.

- Like many of these other platforms, it is set up for 3d support, but has a pretty easy 2d api for people like me that just want to make a simple 2d game without worrying about 3d graphics math.

- Although most of the projects in gallery look pretty janky and homemade (like mine!), there are a few examples (Ingress) of "real" games that are using it. So at least it's mature enough for a real game.

So now that I've decided to use it for my project, I haven't really hit too many warts yet. That's a good sign. I ran into one place where the default Tiled map loader was pretty inefficient for large but sparse maps. Being open source, that was easy to fix. Overall, though, I've been pretty happy. It gets out of the way and lets me write my game.

Blogger app for Android (don't use it!)

Hmm, so I started typing a long post about LibGDX, using the Blogger android app. That was a mistake. While typing, I toggled over to a web browser to check something. When I came back, it had deleted my entire post. Turns out that wasn't a one-time thing, that's just how it works. I guess whoever at Google made the thing didn't read their own document about  the Android Lifecycle. Ah well.  Maybe I'll try again tomorrow.

Ramps are weird

So I've got most of the real core of Robo-Ninja complete. Not that that means I'm anywhere near complete. The basic core (map loading, characters moving and animating, collisions, etc) is the easy (and fun) part. The details are what takes all the time.

Anyway, after getting the basics of collisions working, I suddenly decided (for reasons I haven't really figured out) that I needed to add ramps. I wish I would have been smart enough to think of that earlier, as my design for collisions didn't really take ramps into account. And ramps are weird.

Overall, wall collisions are pretty easy, although slightly harder in a game like Robo-Ninja than they were in Anguna. In Anguna, you just check for collisions, and revert the character's position if he collided with a wall.  There were a few other gotchas, (type of wall, since you can sometimes walk over water. Or checking vertical collisions separate from horizontal so you could "slide" diagonally across a wall), but it was easy.  In Robo-Ninja, vertical collisions need to work similarly (walls above you will suddenly stop any upward velocity. Walls below will stop downward velocity). But horizontal collisions will make your guy suddenly switch directions.

Now ramps. Now we've got a collision that, being a diagonal ramp, is suddenly a vertical AND horizontal collision at the same time. And it should not only change your vertical velocity (stopping downward momentum), but also your position (as it needs to nudge you up vertically if you are moving left, but not send you jumping up in the air). And the tiny magic intersection tile between ramp tiles is really a normal tile. Which meant if Robo-Ninja moved a tiny bit too far forward in each frame, he'd collide with THAT tile instead of the actual ramp tile, and decide to turn around and run the other way.

I can't tell you how many times I kept bouncing off that highlighted tile until I  stopped being dumb and fixed my code.

Anyway, none of that should actually be hard. But it would have been a whole lot easier if I had decided I wanted ramps BEFORE implementing the rest of my collision code, so I would have actually designed with that in mind. Oh well. It works now.

Now I'm on to items/powerups, and the UI associated with them.