title unknown...

• 2012-10-08

  Indie Games, the documentary

  In my recently posts, as you should have noticed, I started to develop some games (the first one is survivor), so I’ve to confess that such field really caught my attention. Following this vibe, I’ve just watched a really cool doc called “Indie Games”, and I would like to briefly share my thoughts about it.

  The story goes behind the process to develop three well known indie games Super Meat Boy, Fez and Braid.

  This is a really interesting story where it’s shown how hard, intrisic and nice is to develop a independent game. Moreover it shows how nice is to have a sucessful game launched out and played by thousand of peoples (something that I’d like to experience someday).

  In short, it’s such a good documentary, the one that when I finished it my will was to just get myself in my room spending days and days developing some game.

  I would absolutely suggest to anyone who is interested in the subject to watch the trailer below and if you still fell interested why not to consider buying the movie? It’s available on the website and costs only $9.99.

• 2012-09-12

  Survivor, a C game developed in 48 hours

  Last week I participated in SPJam, a game developing marathon. About a hundred developers got together to develop any kind of game during 48 non-stop hours. I have to confess that was a really nice experience to be part of it, and I would like to share it through this post.

  Fortunately, we reached our goal that was to deliver a really good game by the end of the 48 hours. The game will be exposed in a public exhibition by the end of October, luckily we will have more time to improve the game.

  the game

  My team, named as PseudoGames, was composed by 3 members (me, zed9h and cairevs). For this competition we attempt to develop a game called Survivor, it’s a kind of survivor horror games with a 8 bit based graphics designed to be played as an arcade game (it’s also can be player in a pc).

  Basically the game is multiplayer (it’s possible to single play also) composed by 2 characters: a soldier and an engineer, both fighting against a horde of bloody zombies.

  In summary, the soldier is the striker one, he is really slow to run and to get items, but really good in shooting. In other hand the engineer is really fast and can build some barriers to delay the zombies (this feature is under development) and grab items faster than the soldier, however an engineer is not a good shooter.

  Here follow a simple gameplay of the first version of the game (this is a alpha version, it was the one who we deliver after the 48 hours), I will upload a more sophisticated version by the time we finish it.

  
  

  learn and have fun

  Our main goal in the marathon was not to develop the best game ever, instead of it we tried to spend our time in something where we could learn a lot and have some fun times. So Survivor was totally developed from scratch using C (yes, it’s fun) as the main language and SDL as the graph library. Actually we were the only team that didn’t use any kind of game engine, such unity.

  To be completely honest, it was my first attempt to create a full featured game using C (I had already developed a simple snake one) and we didn’t have time to implement unit testing or something like that.

  Given the internet connection was really poor, our development strategy was to use the unix manuals to get the SDL bindings and try to remember the most we could. I’ll not describe the entire development steps we used but I will try to list some libs and algorithms we used through the game:

  • SDL to controls and blitting
  • SDL_ttf to write text into the screen
  • SDL_mixer to music and sound effects
  • A* algorithm to handle pathfinding
  • try to decompose our project into some libraries

  In short, SPJam was a really nice experience, and I would like to invite everyone to check out the source code on Github.

  We are also implementing more features and bug fixes, so if you want to report some bug or share something fell free to drop your comment.

  the last advice

  My last advice is if you really want to learn a new language, I highly recommend you to develop a game using it. When you are developing a game you frequently have to explore all features the languages provides, and to be extremely cautious on performance issues.

• 2012-08-25

  Attempts to develop my first C Game

  Next week I’ll be part of SPJam, a Sao Paolo game marathon, and I’ve to confess that I’m really excited to it. The only problem is that my team chooses C as the main language to develop the game, and to be completely honest I’ve never developed a game in such language before.

  In order to be an useful member, I started to study C game development, and yep I have only a few weeks to learn it. So I want to share some steps I follow to create my first C game and to learn a few topics.

  Setup the environment

  Firstly, I started to setup my development environment. As IDE I always prefer to use vim, and as libs we will use only SDL - a multimedia low level library - which I recommend if you want to create 2D games in C (there is a option to extend it to 3D graphics also).

  The first attempt

  After the initial setup I started to develop some sample applications in order to put in practice what I had read. So my first attempt was to create a simple C app that draws pixel by pixel something into the screen. This is just a few topics I learnt:

  • The 8 bit RGBA is the default, so we have 2 bits for each color and alpha.
  • Bitshifting is a good choice when you want to change color in RGB
  • Function pointers are useful

  The second attempt

  My second attempt was to create a user interactive application, where it’s possible to use the keyboard to move a little square throught the screen. This actually was not such a big deal because the square is not a pixel by pixel drawing is just a SDL_LoadBMP call and keyboard events handling with SDL is preety simple.

  The Snake

  After the second attempt I decided to build a complete Snake game, with collision handling, feeding and movement. For me, the hardest part was not the collision one, actually was I would move the snake head and tail.

  To move the snake I started to build the head and after it the tail and save all pieces into an array, so the snake is just a array with a bunch of X and Y coordinated. To move the snake I have to iterate over the pieces and move the pieces in couples, I mean, get the piece[x] and put the x and y in piece[x+1], the problem with this approach is that you will always loose one piece[x+1] value (it was replaced by piece[x], you can use a temporary variable to deal with it), so for me was much more easy to start by the last piece of the tail to the head.

  The snake collision was easy because I have only a single collision point, that is the head, so every frame I check if the head hits any piece, food or the screen edges. Here follows some more advices:

  • When you are loading a bitmap always consider in your pixel calculus the x and y of it. (seems obvious, but I forgot sometimes)
  • Use SDL_Delay and SDL_GetTicks to control your fps, and have a more constant gameplay (thanks ZeD to this advice)
  • Separate your actions in functions.

  I have a lot of work to do in the game and mainly in the code, one of them is a refactoring to split the funcions in the corresponding files, other is to add text manipulation and other features in the game.

  Of course I cannot consider myself as a game developer expert, and I have a LOT to learn, but in other hand I think now I can be more useful in C game development than some weeks ago.

  If you want to check the code, you can visit the repo: https://github.com/pmariano/sdl-samples

• 2012-07-13

  Why I Avoid Using Java Syscalls

  Recently, I’ve been working on projects mainly related with video encoding/decoding, and in order to help me in those task I used the good designed called ffmpeg - which you probably have already heard about before. So I started do maintain a legacy java project which frequently uses syscalls to such tool, so I like to share my toughs and what I learnt.

  In short the context is: when someone upload a new video it’s necessary to retrieve its duration and other metas, so in order to get such information we have to use the well-known media library “ffmpeg”.

  It seems to be an easy task, right? Firstly, we’ve just to do a simple syscall with “ffmpeg -i myvideo.mov” using Runtime.getRuntime().exec(), then get the outputstream and parse the output with a simple regex to get it’s duration. No problem at all, peace of cake, the simplified code would be something like this:

  Process process = Runtime.getRuntime();
process.exec(“Mycommand”);
process.waitFor();
InputStream is = process.getErrorStream();
readTheInputStream(is);
//more and more code...


  Ok, so we have implemented the code and deployed into production. Howeve some months after, the application started to throw some errors, and when I decided to check some logs I got something like this:

  java.io.IOException: Cannot run program “ffmpeg”:
  java.io.IOException: error=12, Cannot allocate memory

  The fork !

  What??? My JVM was setup with -Xmx728MB, my server had 300MB free memory and nevertheless I’m getting an memory error when executing an ordinary syscall?

  After some searches on the internet I realized that the problem was caused by the execution lots of Runtime.getRuntime().exec().

  I check that ordinary command makes the JVM to execute a SO subprocess called fork(). Everytime time a fork() is called on a unix based OS according to Sun, a copy of the parent process(the process who called it) is copied to a new one, doubling the necessary java memory on JVM. This is a common behavior on a series of unix SO’s.

  In my case I’m talking about a web application running on a tomcat container (the application does lots of other tasks than just call ffmpeg), so my application was using a good amount of memory (around 200MB) and when a fork() is called the allocated memory doubled instantly, so I check that my simple ffmpeg -i command is using around 200MB of memory. So the total memory used to run my application and one syscall is 400MB.

  A simple solution

  Particularly, to solve my problem I used a library called Xuggler which don’t make use of any syscall. Instead of syscall it uses JNI with a C library (ffmpeg-x), so there is no problem with any fork() calls.

  In general, I would suggest to use native library instead of syscalls however if you have no available library and you don’t have to spend writing your one from scratch, I would recommend you to avoid use Java at all and try another language such as Perl or even ShellScripts.

• 2012-07-12

  Worth watching documentary.

  It’s good to see that some old tech truths lives nowadays.

• 2011-11-09

  Code that works is not good enough.

  There are no excuses to write bad code. Yes, simple like that. I mean, you should write good code everytime.

  First of all let me explain what I mean by “bad code”: bad code is every snippet (method, class, whathever) of code you write only worring about if it works or not - come’on developers, if you think that make the code works is the most difficult part of your system, I’m sorry to tell you that you are doing it wrong and I suggest you to do some code maintenance.

  So, what would be a good code? In my opinion, is the code you are proud of, code that when you wrote it you have learned something or had some design concerns and spend some time on it. Developer shouldn’t seek to write the best code ever, but must work to write code with minimum quality with the given time.

  It is known that most of applications needs or will need some new features, bugs or some performance improvements, I mean, every system needs some maintenance. So if you write bad code that just “works”, when you’ll have to do some maintenance on probably you will face some problems, such as:

  • Spend too much time on trying to understand the code you wrote (test cases included).
  • Eventually rewrite features because no one is able to understand it anymore.
  • Making your routine just boring.
  • Write more and more bad code (“the broken window“ theory)

  I’ve listed so many time the default excuse to wrote a bad code:

  I didn’t have time enough to write a good code so I did that sh*t.

  Shame on you guy, what I good developer would argue would be:

  I wrote the best code I could in the time which was given to me.

  Ok, so if you don’t have much time why you don’t try to automatize things? Create/use more libraries, read books about code quality to give you an idea on how to proceed in a fast way, use TDD to driven you design and speed up maintenance, use pair programming to have other ideas, read other one code, and so on.

  In short, don’t find excuses for your lack of code quality, the only excuse I could take is “Well, this is the best code I could wrote“, and for me this means that you want to learn and to improve your code, and it’s fine, it’s means that you did your best and if it’s not good, right, let’s refactoring.

  I wrote bad code, everyone wrotes. But is easy to realise when you code is not good and do some refactorings later and, when I’m now able to improve I always ask for some help ou different opinions.

  I work in a media provider, and yes (in contrast with my title), sometimes you have just one second to solve a production bug, so you are in a situation where no one (or just a few, blessed, amazing ninja developer, what I’m not) is able to implement a good code. So what I did is write the best code I could in this situation (even it’s a workaround - I hate this word) and right after the problem is stablished I do a refactoring, leaving no bad code to future maintenances.

  To finish my point, I would like to quote Robert C. Martin (a.k.a Uncle Bob) and his statement on this thread:

  The only way to go really fast is to write the best code you can possibly write, with the best design you can possibly give it.

  So get as far as you can go on improve code quality and at least write a code you are proud of.

• 2011-11-07

  RubyConf Brazil 2011 in a few paragraphs

  Last Thursday and Friday I had the opportunity to attend RubyConf Brazil 2011, a well-organized conference by Locaweb and Fabio Akita. On this post I’ll show up my feelings and thoughts about it using just a few paragraphs.

  First, let’s talk about the venue. The conference took place in a well-known “shopping mall” in São Paulo near to Av. Paulista, which means that most of attendees could use public transport such as subway to come to the conference.

  The decision of using a mall as an conference venue for me is always a good idea. You are a step closer to a wide variety of restaurants and, when things get boring, you can take a walk across the mall.

  One aspect that I think could be improved for next editions is the total space for people to walk and to talk in during coffee breaks (yeah, another point to the organizers, the food was excellent). I don’t think more space as a whole is necessary, just rearranging the food through all the place instead of centralizing it on the talk rooms exits. I saw lots of free spaces but people insisted on being next to the food (I guess, it’s an instinct).

  Now, let’s go to the talks I’ve attended. I won’t give a complete review about all of them, just a short opinion over my feelings and thoughts about those I liked most. Let’s start.

  Douglas Campos (@qmx) on JRuby on Steroids

  
  

  photo by @hannelita

  Great talk presenting how JRuby team is using the newest Java 7 features to improve JRuby. Douglas explained such a hard topic as JVM is in a simple and deep way. And to get things harder he was talking about Java platform stuffs in a ‘rubyists’ conference, so he could have faced some tomatoes or something like that anytime (just kidding, at all, there are few extremists there).

  José Valim (@josevalim) on Learn and Understand more about Ruby creating your own Programming Language

  

  photo by locaweb

  As José said, he gave the ‘long-titled’ talk on the conference. It was a really nice talk where  José explained in which ways building a new language could improve our development skills (really, really interesting). At the end, he gave us some kind of quiz where he whether is interesting or not adding some new features to Ruby and after his briefing, we voted on what he had suggested.

  Aaron Patterson (@tenderlove) on Who makes the best Churrasco? and Why I hate Lambdas

  
  

  photo by @hannelita

  As usual Aaron gave a outstanding talk presenting us with some rails (mainly ActiveRecord) internals. He explained things that we didn’t even think existed before, like peculiarities with threads and connection pooling with Rails. At the end he show us some features he would like to implement on further versions of rails, such as the new rails route resolver which uses graph algorithms instead of the current regex pattern matching . After this presentation I saw a lot of people (including me) realizing that there’s still a lot to study.

  The last RubyConfBr 2011 talk had Aaron as speaker too, replacing other speaker that couldn’t come. The talk was entitled “Why I hate Lambdas”, and, IMO was one of the most amazing and funny talks I’ve ever seen. The content was really simple: why he hated lambdas and what to use in order to replace it. This talk took just a few minutes (I guess about 15 minutes) after that he started talking about the ‘Worst ideas ever’, where he showed us a set of bad ideas he had to implement (or thought to implement) in order to learn or prove something. Believe me, he wrote a parser, called it Phuby, to execute php code, inside a ruby application. Using Phuby he was able to beat ‘dhh’ by creating a 2 minute blog instead of the usual 15 presented by rails community. What a great talk - I could learn that even the worst ideas we have should be implemented, because it the end - and at least  - it’s possible to learn something.

  There were other great talks such as the ones from Bruno Oliveira (@abstractj), Nick Sutterer (@apotonick), Lucas Húngaro (@lucashungaro), David A. Black (@david_a_black) and Nando Vieira (@fnando). If you want more details about the talks you can check eventials.com/rubyconfbr.

  Taking this into account it’s possible to consider RubyConf 2011 one of the greatest 2011 conferences, where I was able to learn, talk with old friends, exchange ideas and laugh a lot.

  In short, conferences, for most us developers should be more than just presentations. IMO, conferences are great opportunities to see how far you have been in studies and to take notes of the newest stuffs there’s out there, and the best conference is the one I leave with more questions and doubts than when I arrived, so now I’ve a lot things to read and study and I intend to post some of them here in my blog.