Saturday, August 5, 2017

Assert your assumptions.

A lot of software bugs are caused by the following scenario: The programmer assumed a certain situation and wrote code that would rely on this assumption. Very often, these mental assumptions are never made explicit, and live in the programmer's mind only. A standard weapon against these bugs in the arsenal of the coder is the assertion.

When working in teams I would often argue for the use of asserts in production code, to the astonishment of my colleagues. The most frequent objection to this would be the performance impact. However, the 98% of your code that is high level will never show a perceivable performance impact from asserts. The remaining 2% you just disable the asserts in non-debug builds. The inner loop of the physics solver, the inner loop of the collider, the inner loop of the path planner. Sure, leave them out there. For all other code? Ship your product with asserts! You'll be amazed at what gets triggered.

To take the controversy a small step further: I also firmly believe in making the asserts user-facing. Let's take Android apps for instance... instead of having Android report "Foo stopped working" I prefer to show a dialog with the assertion that failed, the file name and line nr. By doing this you can leverage your customer base to improve your app's stability. A certain user may detect the pattern: "Every time I do X after doing Y, I get a crash with message Z." Exactly the information the developer needs to fix the bug!

And just for good measure, (or is it overkill?) I took it further again: and sometimes will even show callstacks to users. For instance, a steam user is quickly able to screencapture and post it on steam, or email it to the dev. So much valuable info.

So last month, I was noticing crashes reported by Google Play on my Android app. And not always will users contact you with the crash report, so I even with pulling in the user for crash reports, the right information would not always reach me. One solution is to adopt a heavyweight framework to do these reports for me. Link to an analytics framework. But it's possible to make a very lightweight solution yourself.

I was interested in the assert message only. No need for user info, device info, anything. Just the assert message with source code line number. And this is very small data. Smaller than an UDP packet!

So I coded a minimalistic assert report server to run on Amazon's EC2. Just a few lines of Python code, where a socketserver.UDPServer prints out the datagrams it receives.

Client side, it is a little bit more code. Mine is in C, as I make NDK based Android Apps, but in Java it is even simpler. This slightly verbose C code will do the trick to open a socket and send a single datagram to my assert report server.

I am happy to report the assert report server is doing its job, and helps me to get on top of the crashes in my game. For future improvement, I should probably add an app version number to the assert message, so I will know if it is an old issue or not. Thank you for reading, and let me know what you think of assertions, and if you run assertions in your production code or not.

Sunday, June 18, 2017

Android Revisited.

Every year or so, I come back to Android development. And often, quite a bit has changed. On top of that, I will have forgotten gotchas, and fall in the same trap twice. Hence, I need to document my process better. So here's another Android report.

I managed to hold out on IDEs altogether. I never used Eclipse, and always did ndk-build and ant on the command line. This time around, I have been forced to use Android Studio, as the google play game services are only updated for Android Studio builds. Android Studio is built on top of the Gradle build system. Gradle is a Java-only thing. For C++ it depends on either ndk-build (couldn't make it work with Android Studio) or CMake. So CMake it is now. You have to write a lot of CMakeLists.txt files for your game, and its dependencies. It is a hassle.

You can only add a single CMakeLists.txt to your Android project. So dependencies need to be included from a top level file. To do this, use: add_subdirectory( srcdir builddir )

Settings in your Gradle files are used to overwrite settings in your AndroidManifest.xml file.

Google Play Games is a mess as always. The latest version will crash on launch for me, so I had to downgrade it in my app's Gradle file.

The 'DEX' gets too large, so you need to selectively use Google Play Games dependencies like this:

dependencies {
    compile 'com.google.android.gms:play-services-auth:10.0.1'
    compile 'com.google.android.gms:play-services-nearby:10.0.1'
    compile 'com.google.android.gms:play-services-games:10.0.1'
}

If you target SDK level 21 then the Android Soft Keys will be hiding your app's content. For proper sizing, switch to targetSdk 19 instead.

After changing dependencies in your app's Gradle file, you need to do a clean, otherwise conflicts can arise, like duplicate definitions.

To link against a pre-built library, e.g. the Google provided gpg-cpp-sdk library, use the following CMake syntax:

add_library( gpg
             STATIC
             IMPORTED
)
set_target_properties( gpg
  PROPERTIES IMPORTED_LOCATION
  $ENV{HOME}/src/gpg-cpp-sdk/android/lib/gnustl/${ANDROID_ABI}/libgpg.a )

There is something wrong with the arm64-v8a build of Google Play Games gpg.a static library.

Wednesday, April 19, 2017

Real Time Ray Tracing for games.

In 2015, I've been tinkering with a real-time photon mapper to do Global Illumination for simple voxel worlds. Recently, I thought there should be more I could do with this idea.

So my Real Time Ray Tracing experiments are back, but with a twist. The minimalistic voxel world was a little too minimalistic. And it couldn't be much more detailed if I were to keep a 60 FPS target. So something had to give.

This time, I have dropped the Global Illumination feature, and no longer compute indirect light. Instead, I just fire one primary ray, and one shadow ray for each pixel in the projection. Granted, this does not give you much advantage in quality over rasterizing. But it does give nice crisp shadows, that can be cast from a local light, casting shadows in all directions.

To have a little more detail than simple voxel models, I went with AABB primitives. To avoid confusion: Normally AABBs are used in ray tracing for spatial indexing. In my project, however, I use the AABBs are the geometric primitives. They are more efficient than voxels, as large blocks and planes only need a single AABB primitive.

I do use spatial indexing to cull primitives though, but for this I use a simple voxel grid. In each grid cell, I have an instance of a primitive model. These primitive models are packed up as SIMD friendly sets of 8 AABBs in SoA form. Using 8-way SIMD, a single set of 8 AABBs are intersection-tested against a ray, in just a few instructions.

Target platform is AVX2 equipped CPUs, and it's currently running on GNU/Linux and MacOS. It's running close to 60Hz on a 4-core Haswell, at the expense of loud fans running, unfortunately.

Game Play is yet to be determined. Post your ideas in the comments. Thanks!

Saturday, March 4, 2017

The Little Bike That Could

So, I released a game yesterday. You can grab a free copy of The Little Bike That Could over at the itch.io site.

I made this over the course of one month, using the same custom game engine that I use for all my games. I had tremendous fun creating it.

Bicycle dynamics are so fascinating. It's a lot more involved than it appears to be at first glance. As a matter of fact, scientists had a really hard time determining where a bicycle's stability comes from. It turns out it is not caused by gyroscopic effect of the wheels.

It's pretty amazing that humans can cycle at all, as the process is so much different than it appears. For starters... imagine your self riding in a straight line on a bicycle. You want to make a left turn. What do you do with the handlebar?

If you answered: for a left turn, you turn the handlebar left... No, that is not what happens, surprisingly! Instead, you turn the handlebar in the opposite direction (called counter steering.) So you turn them right, which causes the bike to fall (lean) to the left. Only now, do you turn the handle bars to the left, to stop the fall to the left. This maintains the left-lean, and causes a left turn.

To stop the left-lean, you briefly steer further left which brings the bike upright again, and thereby ending the turn.

How remarkable is that? To start a left turn, you first steer right. To end a left turn, you steer left. As a child, our brain somehow is trained on this behaviour, and for the rest of our lives, it enables us to ride and steer a bicycle.

So I made a scientifically valid physics simulation for a bicycle. The player will choose a lean angle (to left or right) and then the computer will turn the handlebar to reach a desired lean-angle, causing the turn. I also added independent controls for rear and front brakes. This enables rear-wheel slides and front wheel stoppies. I created 10 levels with different challenges.

It's hard to really explain the game play, and others do a much better job of it. So allow me to introduce Iwan who played my game and made an excellent video about it.

Wednesday, March 1, 2017

The alternate universe without a computer mouse.

Consider the de facto standard user interface for computers and computer software. It is what we call the Graphical User Interface, or GUI. Comprised of a mouse cursor and windows, sliders, buttons, scrollbars and such.

I remember seeing my first computer mouse on a trade show. I'm going to venture that it was a Macintosh mouse. And it seemed magical to me. All I knew was 8-bit micros from Sinclair, Commodore, Acorn and the like. And suddenly there was this element on the screen that somehow was moved in sync with the movement of a little device. I seemed from the future!

The computer mouse may have been popularized by Macintosh, but its way was paved by Xerox Parc.

Personally, I'm more at home with command-line interfaces, with text based commands. As I've put it earlier: How reproducible is a mouse click? I use the Visual Studio IDE to port my games to windows. But because I use it so intermittently, I tend to forget my work flow. Where do I click, in which sub-sub-sub-menu to solve this issue with the compilation? I spent too much time wrestling with the GUI, wishing I was back in the command line environment with Makefile, as I do under Linux. If I need to reproduce something, I can just examine, and re-run the script.

While taking a shower, a thought struck me: what if the GUI was never popularized, or even invented? What if the text based interface, on a command line, would have persisted from the 70s into 2016. Let's visit this alternate universe, where the computer mouse, and the GUI, never evolved.

What would this alternate universe look like? I'm going to make a bold statement here: I think it would have been more technologically advanced than our current society.

Without a computer mouse, and without the GUI, we would still have an Internet. And we would still be ordering pizzas over the Internet. Just not with clicking buttons, but by typing in a command:

$ order-food -restaurant=marios -cheese=gorgonzola -meat=proscuitto -size=m

...or something similar.

The real kicker here, is that all these internet-based services would be text based, thus so much easier to extend and compose into more complex services. Maybe in a script, you want to combine the pizza-order with another command send to your fridge to chill the wine. If there is a text based API for service A, and one for service B, anyone (not just programmers working for the corporations behind A and B) would be capable of layering new services on top of these. No source-code of the web-sites would be needed. Extending a GUI based application typically cannot be done with no access to the source code. Even if source code is available, it is pretty hard. Extending a command line based application is much easier, and does not require source code. We could invoke services in scripts, sequence them, layer them, schedule them, whatever we desire.

This means that in our alternate universe, software applications will be far easier to extend. Developers can build on top of what others have made. Being able to extend third party GUI applications tends to be impossible, or at the very least hard, and would require things like plug-in infrastructures, and SDKs. Of the gazillion services available on the Internet today, very few are customizable, extensible. In our universe we spend a lot more time reimplementing the 'full stack' over and over again, instead of building on top of what others have made.

To conclude, I think the invention of the computer mouse was a curse, that just looked like a blessing. Sure, it may make the life of creatives easier. Producing digital art is so much easier with it. But in terms of computer technology, I fear that we are poorer for it, not richer. The GUI killed the command line, and that may very well not have been a good thing.

Did the computer mouse doom the technological progress in our society?

Saturday, February 11, 2017

The Little Bike That Could.

I am announcing my latest indie title, which is currently under development: "The Little Bike That Could." Like many of my previous titles, it comes with an exceptional physics simulation. It's been fun implementing and tweaking it. Wheelies, Stoppies, Slides and even loopings on a half-pipe are possible.

It is currently implemented for use with a gamepad on linux. The initial release will probably be on the ich.io portal, and after that I may experiment with phone-tilt control. So far, I've designed and implemented 6 different game levels, the last of which is fiendishly difficult.

If you want to follow the progress of this game, or participate in early testing, join the Google+ community for it.

Tuesday, January 3, 2017

New windows install

I've reinstalled windows on my PC, going to Windows 10. Here's a list of things to do on a new install:

  • Expand the explorer 'ribbon' to set the 'View' properties on your local drive. This is needed if you want to see file extensions.
  • Installed Visual Studio 2015
  • Installed Sys Internals Process Monitor.
  • Installed git for windows, which comes with the great git bash, that lets you do UNIX on windows.
  • Set Google as Default Search.
  • Using a git shell, make a new RSA key using 'ssh-keygen.exe -t rsa' command. Paste the public key in GitHub.com site's settings.
  • Set privacy options. Settings -> Update & security -> Advanced Options -> Privacy Settings. Switch everything off here. Note that windows will sometimes set them all back 'on' after a windows update. Sneaky if you ask me.

Sunday, January 1, 2017

2016 totals

So, The Little Crane That Could is holding stable on Android, but dropping on other platforms. Here are the 2016 results (Number of free Downloads.)

2016 2015 2014 2013 2012 2011
iOS 416K 630K 1300K 3199K 3454K 1550K
Android 1515K 1525K 825K 1579K 1656K -
Mac 10K 20K 30K 53K 81K -
OUYA - 0K 4K 15K - -
Kindle 48K 52K 46K 95K - -
Rasp Pi - ? ? 6K - -