Many moons ago I used to be pretty proficient at 3D programming.  Not really for work but more side projects, fun, to learn.  I wouldn’t say I’m artistic, hell, I can barely draw a stick-man.  I do know what looks good though thematically, a side effect of 25 years of UI and software design I suppose.

I haven’t kept up with things though, which is part of why it’s always difficult for me to understand people who say “I’m bored”.  There are simply so many things to learn, to spend time tinkering with and have fun playing with that I never have enough time.  I suppose I’m used to the old days, the pre-internet era where you had to learn everything the hard way, going over endless books, chatting on BBSes, finding examples in magazines and so on that I find it extremely easy to find side projects to play with on the internet and relatively quick and easy to become proficient at those new toys and techniques.  Not saying I’m an expert in everything I mess around with but I usually get enough knowledge to have fun and feel like I learned something quickly.

Anyway, back on topic… In one of my numerous daily meetings I found out that there are a number of projects that plan on using the three.js system for 3D rendering.  While listening to the presenter discuss their techniques I loaded up the information on three.js and saw they have a ton of pretty cool demos as well as a very cool example page that shows off the myriad of features the engine supports.  I was impressed.  How is all of this possible inside a browser window?  When did that happen?  I always found web pages rather .. bland.. when compared to the demo/intro scene of yester-year.  This ability to render 3d objects and effects in a browser window seemed like fun.  I hate the logo you see at the top of the screen and wanted to try and replicate it but animated, using this three.js system.

Well, after experimenting and reviewing their examples I had a bunch of things working and was excited to play more but hit a roadblock.  Well not a roadblock stopping me from work but something that I’d fight with every step.  Javascript.  Javascript is the back end language you use to leverage your three.js content.  You might say “well Will, you write java code every day, you have been for more than 15 years, why would you consider Javascript a roadblock, surely you can apply some of your daily practice and be proficient at javascript instantly”.  Well sir, you’d be wrong.  I suppose after 25 years of professional development and countless years learning in grade school I’ve been ruined by structured programming languages.  Javascript is the wild wild west, the rules are so loose that you can chop your feet off and not even know it until you start running.  I found it very difficult to organize my ideas into logical pieces then have them communicate with the code effectively.  I’m very sure it’s an experience thing but I like having organized ideas that are totally encapsulated.  Where the concepts do not contaminate each other except for what you wish to expose.. With Javascript you can do anything anywhere, making the code, in my opinion, messy.  If I try and code up nice tightly composed ideas I find with my idea of keeping the ideas separate and easy to use.

I can work around javascript, its just a matter of relaxing the years abuse of writing small cohesive pieces of code and being more relaxed with my designs.  Sadly though the WordPress CMS backend I’m using for this website it’s exceedingly difficult to embed javascript elements without re-writing much of the theme you’re using.  I wanted to change the logo above to make it animated but realize that I’d have to rewrite the theme I’m using to accommodate for doing that and I’m lazy so I’ll just post my example to a supplemental web page instead you can visit here:

http://www.pixelsystems.net/proto/logo.php

It’s not really a 1-to-1 translation of the graphic logo animated but just me learning how to load 3D fonts, implement tweening, how to programatically fill a cat-mull spline polygon and so on.

What is impressive, after all my complaining is the cost of entry to start learning and having fun with javascript and three.js.  The cost of entry is nothing, beyond the time required to read and experiment.  I was always impressed what you got for free with Java (free language, free IDE, tons of source code), with Javascript it’s even easier, there are tons of good editors out there for writing code, many of which interface with libraries like three.js and offer code completion, running right out of the IDE and other niceties that we take for granted now.  It’s very easy to write code, save it and run it instantly.  You can debug it instantly in your browser so very quick access to the data structures, their contents, console logging and so on, with nothing else to install and set up.

I’ll likely add this stuff to the endless parade of “play more with when I have free time” items on my plate.

Just a little update, my last post was detailing how I’ve renewed my Minix Box, however I’ve since then essentially retired it.  It was still usable (and I may still use it at a different location) I did come into an inexpensive laptop that works flawlessly for driving my TV.  Since it has an SSD it starts significantly faster and with the extra horsepower doesn’t chug like the aging Minix.

I have started a few new side projects worth noting, I’ll discuss the one that’s taken up most of my time recently.  In the past I picked up a few inexpensive WIFI low light webcams.

For the better part of 2 years I ran them with some commercial software called WebCamXP which allowed me to configure a wide range of motion sensing security for the cams, off-site storage of security footage, notification and so on.  It worked really well, hence me using it for 2 years but… It required a dedicated Windows PC to JUST sit and monitor and detect camera footage.  It’s not like I don’t have spare PCs lying around but it did use a significant amount of processing power to do complex frame comparisons to provide accurate motion detection.  The cameras also, like most things that are inexpensive, were not terribly reliable, they would often need resetting.  Most likely due to the sheer amount of traffic they were generating (5-12 FPS of a 720×540 feed running 24/7 on wifi).

For the last 6 months or so I sort of just disabled the system and the cams sat idle.  Until I came across a inexpensive deal on DDR around a Chinese home monitoring system, “Xiaomi Mijia 6 in 1 Smart Home Security Kit“.

It arrived without issue from China and thankfully works pretty well out of the box.  The initial gateway configuration was a bit strange considering the gateway ONLY speaks Chinese, the prompts however aren’t that important and it integrated with my WIFI fine and was able to detect all the supplemental devices just fine.  The Kit I got came with the gateway (which acts as the hub for the sub devices), a motion sensor, door/window sensor, remote switch, Chinese specific AC WIFI plug and temperature humidity sensor.  The nice thing is that it’s ridiculously easy to add new devices and the devices that came with the package were easily discovered by the gateway and instantly added to the Android app without issue.

I used the MIHOME kit for a few months, happy with how it was working.  The sub devices run on coin batteries, can be placed pretty much anywhere in the home and work flawlessly.

One day I must have just been killing time and happened upon OpenHAB2.  The premise is that it is a self contained package that runs on most systems (Win/Linux/RaspberryPI) which brings together all your internet type devices into one easy to use and manage location.  Since I am running a Linux machine at home to manage my media and other services it made sense to install OpenHAB and see if I could get it to drive the MIHOME as well as the webcams and other devices.  After some installation and configuration I had OpenHAB talking to the MIHOME kit, my garage door opener, Orvibo wifi socket, webcams, cell phones etc.  The system allows me to write some pretty complex rules like:

“If no-one is home and I detect the garage door open then send Will an Android notification message saying….The Garage Door is open, did you leave it open?”

or

“If I detect motion from the MIHOME motion sensor on the main floor and no one is home, turn on the WIFI socket and lamp, start recording the main floor webcam and send myself an Android notification message that motion was detected”

I’m still experimenting with it, I do wish it had a better remote viewing type application that was easier to use than configuring a VPN or other security type layers.  I like that it uses significantly less power, less bandwidth with a lot more flexibility on what activities the devices are available to perform.  I do like that OpenHAB is developed in Java, something I’m familiar with so making changes or investigating improvements is pretty easy, though the build system is quite complex.

I’ve since ordered a few more motion and door sensors for the MIHOME kit which will easily integrate with OpenHAB to provide better coverage for my home.

I really haven’t had a lot of time to spend on new things, I’ve had a lot of personal distractions over the past 2-3 years that have really limited my time and interest in learning new things.

I have however spent a little time looking at HTML5 game engines.  Not that I would have time to actually write a game but I always like reading about and experimenting with graphics and HTML5 seems quite suited to that.  I only experimented with a few game engines, I looked at quite a few but the two that I found the easiest to use and understand were http://www.kiwijs.org/ and http://www.pixijs.com/.

I spent the most time using pixijs simply because there were enough tutorials and web resources to get a better understanding how the pieces tied together.  It’s one thing to write your own engine that you think other people can use, it’s another to actually show people how to use it rather than assume they want to spend hours pulling it apart and actually understanding how things fit together.

Eventually I would like to get back using pixijs and modify the pixelsystems.net logo a the top of this page to make it animated, perhaps some simple 2D effect from the good old Amiga demo/intro days.  I’m just surprised how far browsers have gone where you can write a legitimately good looking and fun game in a web based language instead of C++.

I finally released SudoTape to the market.  The last stretch was around testing and getting some of the final artwork done.

I initially listed it in the photography section of the market but this morning I revised it and put it in the tools section.  You don’t really need to have a camera on your device in order to use it so it makes more sense to put it in the tools section.  I don’t know if the market automatically filters stuff out so even if you don’t have a camera and the app doesn’t really need a camera I’m not sure if you’d see it as available to download or not.  Not that it matters though, I presume most devices come with at least one camera now anyway.

I had problems creating a signed APK, I never really read anything about it and Google’s information online is sketchy at best, at whatever documentation I found was related to Eclipse so it wasn’t much help since I use IntelliJ.  Luckily IntelliJ’s created signed APK wizard wasn’t too tricky so I uploaded the app and some art/screenshots last night.

Going to see if anyone downloads it to see how much I want to support it.  I’d like to add some other features like saving the files to DropBox, perhaps unit conversions on the define markers options dialog, etc.  If there is little interest I’ll hold off adding more features and think of a new project.

You can find SudoTape on the market here:  https://market.android.com/details?id=net.pixelsystems.sudotape

You can read SudoTape’s basic documentation here: http://pixelsystems.net/?page_id=66

I finally figured out how to use GitHub.  It didn’t work well at all with IntelliJ, mostly because the learning curve is pretty vertical regarding to setting up permissions and connecting github with your machine (pgp keys, etc).

You can read about my first github project here and all the trouble I had getting camera images from the camera functionality on your Android phone into my Android app.

My project is coming along nicely.  I learned the hard way that the Android emulator is quite different than an actual device so when I got my phone back running my project on it pointed out a number of issues that I hadn’t thought of because they weren’t exposed with the emulator.

Here are some issues I tackled with recently:

• If you have multiple <intent-filter sections in yourAndroidManifest.xml defining:

<action android:name=”android.intent.action.MAIN” />

<category android:name=”android.intent.category.LAUNCHER” />

then each of those sections will create icons for your application with no other identification saying which is which.. Can be a bit confusing on which icon to press to run what main mechanism

• If you are launching an custom intent/activity from preferences DO NOT specify data= in your preferences.xml unless you are set to handle it, otherwise you will get a “cannot find activity” exception.  This took me several hours to discover, a lot of head scratching and swearing because I got lazy and never deleted the data= part that I copied from Google’s example code.
• The bundle as the argument to onCreate() for your activity is actually very useful and tied to onSaveInstanceState(), this I supposed is common sense but no example I read actually showed them being used together so some digging exposed to me how to use it, very handy.
• PreferenceManager.getDefaultSharedPreferences() is super easy to use, beats property file loading/saving which I use at work a lot.
• My app is targeting Android 2.1 so I couldn’t use Camera.getNumberOfCameras().  I was using Camera.open() to find out if the device has a camera but not calling Camera.release() which caused some really strange exceptions later on in my code
• this information on screens proved very important to several UI problems I was having (multi-touch), should be fundamental reading for anyone learning Android programming.

I started writing code a long time ago, so I went through the days where you had to worry about how much memory you had available to you and how you could get your data into your application without causing out of memory issues.  With all the windowing environments and new object oriented languages it’s easy to forget “the good ole days”.

I was reminded this week when I was working on an app I’m writing for Android.  The software I want to write deals with images saved from the device’s camera (or hopefully other images on the device that the user may have imported,etc).  Android has a lot of great functionality around reading and importing bitmaps but I underestimated how well those bitmaps are kept in memory.  I was running out of memory when I wanted to save my manipulated bitmap based off it’s full original size.  I wasn’t sure how I was going to solve the problem until I found this post on the Google forums:

Yes, the images from the camera are 2048px by 1536px.  If you convert

to a bitmap, thats 2048 x 1536 x 2 = 6291456, or 6MB.  This is right

at the allotted space an Activity gets.  If you can live with smaller

images, I would suggest using a sampleSize of > 1 when decoding the

image.  This will result in a smaller image size, thus less memory

used.  I have used the following code in the past to keep my images

under a set byte count.  Alternatively, you could forget about

calculating the sample size and just set it to 2.  This would use

1.5MB instead of 6MB on a standard image.

6MB, that’s not that much, but for a mobile device, yeah I suppose I can see how that is overkill.  I was unaware that the ‘allotted space an Activity gets’ was around 6MB so that is why my app is running out of memory (basically I was trying to allocated 2 uncompressed images in memory.

The posting on Google gives a solution which I’ve tried.  I haven’t spent any time playing with the calculation but I plan on it as the default sizing loses a lot of the image detail.  I have some wiggle room in there because at any one time I’ll only be using 2 images so I can probably bump up the quality/sizing a bit.

I use Jetbrains IntelliJ Idea at work.  I have for quite a while now.  In the early days they didn’t really exist but since starting college and TurboC I’ve used an IDE.  Since I’m spending my free time learning Android development and I know that Jetbrains have an Android plugin for Idea I figured it would be natural to use IntelliJ.  Unfortunately the plugin for Android is pretty simplistic.  It’s not bad (the recent community beta for Idea is pretty flakey) but it leaves a lot to be desired as filling out the XML files for configuration can be a pain without examples.

I know Google recommends Eclipse so I downloaded it instead and while it’s integration with Android is awesome I find the IDE too cluttered, it does TOO many things.  I never though that I’d say it, but sometimes you can have too much functionality and the user can get confused.  Regularly at work I’ve got into arguments with product managers with my statement  “If the user doesn’t want to use the functionality they’ll just ignore it” but in the case of Eclipse you’re so surrounded by all these features that you can’t really escape it.

I don’t really like how projects are laid out either in Eclipse.  It seems like they try to hide the directory structure/layout of your build process with their “projects” mentality but likely it’s because I’m so used to having my code laid out a different way that it’s just foreign to me.

Either way, learning new stuff is fun but can be frustrating when you just learning to walk and you want to run.

So I finally got back into developing some Android code, I have some projects in mind but since I”m a lazy mofo I won’t talk about them until I feel like they’re turning out how I want.  Typically when I start a new project I think of splitting the key functionality of the application into pieces then start doing some reading and googling to see how to do those pieces.  Then it becomes a matter of taking those pieces and putting them together.

I did some reading on how to do multi-touch.  Pinch to zoom, drag, that sort of thing.  I downloaded the latest beta for JetBrain’s IntelliJ because I knew they had the Android framework included, and I had used their framework in an earlier version (8.x) and it worked fine with the emulator, albeit slow.

I found an interesting article online regarding how to do multi-touch and proceeded to copy/paste some of the code into my project.  After getting everything built properly I proceeded to run my app.  For some reason my app wasn’t showing up.  I know the emulator is slow but damn..  I started looking through IntelliJ’s logs to see if I was missing something and the log says .. my app is running.   WTF?!  Well it turns out that I had my phone plugged in and the Android SDK connector (ADB) automatically uploaded my app to the phone.  I turn the screen on and there it is.  Funny that I totally forgot that the phone was plugged in and that I had USB debugging turned on by default.  Even more exciting was that the process was instant, no more waiting 5 minutes for the emulator to start.  Fantastic stuff.

I found this interesting post on another Android developer’s website (Steven Osborn):

Android 2.2 isUserAMonkey() check

Per the documentation: Returns “true” if the user interface is currently being messed with by a monkey.

There is another gem in the Log class, they’ve added a new log level ‘wtf‘ which in this case means “What a Terrible Failure”

Great stuff, pretty easy to tell the Google developers are geeks like myself, I love shit like that in source code, makes writing stuff a lot more fun.

When working on my own applications at home I realize how much infrastructure code I take for granted from my full time job.  For example, my application at home needed to display a table of data.  Java tables by themselves are pretty ugly, they work but they look very boring, offer very little in ways of formatting and require you to reinvent the wheel in order to get some nice functionality behind them.

I figured “surely there must be some open source tools/code that I can use”.  Well I stumbled upon only ONE that actually seems pretty decent.  Objectplanet has a decent table but if you dig around you can see that you need to buy a commerical license to get full access to it.  And the price for that?  A cool $1000.  This is a reasonable price really considering the amount of work to write a full-fledged table kit on your own but MAN.  We have our own set of table tools in my full-time job that walks circles around Objectplanet’s offering that I never think about when working but in reality it alone would be worth thousands of dollars and it’s such a tiny piece of all the software we use/have developed.

The same goes for a wizard framework.  Again it was one of the first things I helped write when I moved over from MFC->Java back in 2002 in my full time job.  Looking around now there are very few examples/working prototype code that I imagine something I wrote as a requirement at work would likely be worth a lot of money out in the wild.

I’m not really sure why I thought about these things when I worked on native code type applications.  Perhaps because the controls in Java are so bare-bones that to get something really usable/pretty you have to put a lot of effort into it, whereas for win32/MFC type applications the base controls will do 80% of what you really get done and that last 20% is a luxury that most people won’t miss.

Doesn’t really make me want to go back to MFC though.  My skills on Windows development now are probably so far behind now I doubt I’d even remember where to begin.