Tim's Blog

Category: Software Development (page 1 of 2)

ifweallvote.com

I spent the weekend creating a site that holds daily mock presidential elections to inspire us to vote. Record voter turnout will change things for the better. Please use it, share it, get your friends to share it, get Cher to share it.

 

www.ifweallvote.com

HTML-NORMALIZER

I just wrote a library to help developers write DOM tests; specifically DOM assertions.  It allows you to express expected HTML (DOM structure) using React’s JSX.  All the details are in the readme.  The library is called html-normalizer and is available on npm and github. Here’s the link.  https://github.com/TimothyRHuertas/normalizer

 

Happy testing

 

Rails, Resque, Elastic Beanstalk and Ruby 2.1 (Puma) platform

There are several articles on using elastic beanstalk hooks to create and manage (start|stop|restart) background services (this example uses up-start) that govern rake tasks. This is not one of them; rather it is an amendment to those posts intended to provide you with an updated example. I recently upgraded my Elastic Beanstalk application to use Ruby 2.1 (Puma) and my services stopped working. I made some minor changes to get stuff working again.

Disclaimer: There are better ways to accomplish this (see rails sidekiq), but you may still find this useful. Consider it the “copy/paste path of least resistance” for upgrading older projects.

commands:
create_post_dir:
command: “mkdir /opt/elasticbeanstalk/hooks/appdeploy/post”
ignoreErrors: true
create_pid_dir:
command: “mkdir /var/app/current/tmp/pids”
ignoreErrors: true
files:
“/etc/init/resque.conf”:
mode: “000755”
content: |
description “Starts a Resque worker. For example: start resque.”

start on runlevel [2345]
stop on runlevel [!2345]

respawn
respawn limit 5 20

script
exec /bin/bash <<“EOF”
EB_SCRIPT_DIR=$(/opt/elasticbeanstalk/bin/get-config container -k script_dir)
EB_SUPPORT_DIR=$(/opt/elasticbeanstalk/bin/get-config container -k support_dir)

. $EB_SUPPORT_DIR/envvars
. $EB_SCRIPT_DIR/use-app-ruby.sh
EB_APP_USER=$(/opt/elasticbeanstalk/bin/get-config container -k app_user)
EB_APP_DEPLOY_DIR=$(/opt/elasticbeanstalk/bin/get-config container -k app_deploy_dir)

PIDFILE=$EB_APP_DEPLOY_DIR/tmp/pids/resque.pid
LOGFILE=$EB_APP_DEPLOY_DIR/log/resque_workers.log
cd $EB_APP_DEPLOY_DIR
exec su -s /bin/bash -c “bundle exec rake workers:start >> $LOGFILE 2>&1” webapp
EOF
end script

post-stop script
exec /bin/bash <<“EOF”
EB_SCRIPT_DIR=$(/opt/elasticbeanstalk/bin/get-config container -k script_dir)
EB_SUPPORT_DIR=$(/opt/elasticbeanstalk/bin/get-config container -k support_dir)

. $EB_SUPPORT_DIR/envvars
. $EB_SCRIPT_DIR/use-app-ruby.sh
EB_APP_USER=$(/opt/elasticbeanstalk/bin/get-config container -k app_user)
EB_APP_DEPLOY_DIR=$(/opt/elasticbeanstalk/bin/get-config container -k app_deploy_dir)

PIDFILE=$EB_APP_DEPLOY_DIR/tmp/pids/resque.pid
LOGFILE=$EB_APP_DEPLOY_DIR/log/resque_workers.log
cd $EB_APP_DEPLOY_DIR
exec su -s /bin/bash -c “bundle exec rake workers:killall >> $LOGFILE 2>&1” webapp
EOF
end script
“/opt/elasticbeanstalk/hooks/appdeploy/post/999_start_resque_job.sh”:
mode: “000755”
content: |
#!/usr/bin/env bash
. /opt/elasticbeanstalk/support/envvars
set -ex

initctl restart resque || initctl start resque
echo ‘resque restarted’

Beanstalk + Rails Passenger + Rails 4 + AWS == evil

I was super excited to learn about AWS and beanstalk. I created my rails 4 project and used beanstalk to push it. When I went to the url I was greeted by my old friend the nginx error screen. “502 bad gateway”. WTF! I ssh’d in to the box and in the logs I find “You must set config.secret_key_base in your app’s config”. I added config.secret_key_base = ‘dfsdsfsd’ to my application.rb and Bam! Error solved.

InetAddress.getLocalHost().getHostAddress() 127.0.0.1

Today I wasted about 4 hours on this.  I was running in to an issue because InetAddress.getLocalHost().getHostAddress() was returning “127.0.0.1” aka the loop back address and not the “real” IP address I expected.  It sounds silly, but after hours of Google-ing and head banging it turns out the cause of this was a corrupt hosts file (which I had been looking at all day trying to figure out what I needed to do to it to make stuff work).  The moral of the story is if you run in to this problem back up your current host file then replace it with a default one.

FABridge.js play nice with Chrome

Today at work I got an email that said, “The site text editor is broken in Google Chrome. Remember about 3 years ago when you thought FABridge.js was a good idea?” In hindsight I would not have used FABridge. I think the levels of indirection out weigh the ease of use (especially if you need tons of control). Nonetheless I had to fix it. Once I found the bug it was pretty easy to fix. I narrowed the error down to FABridge__bridgeInitialized in FABridge.js. FABridge__bridgeInitialized looks for either an embed or object tag based on the browser type. A check for Chrome did the trick.

In FABridge.js if you change:
if (/Explorer/.test(navigator.appName) || /Konqueror|Safari|KHTML/.test(navigator.appVersion))
to:
if((!(/Chrome/.test(navigator.appVersion))) && ((/Explorer/.test(navigator.appName) || /Konqueror|Safari|KHTML/.test(navigator.appVersion))))

you should be in good shape.

So if you hear FABridge.js does not work in Chrome. Or your JavaScript wrapper to the Flash widget is broken this may solve your challenge.

THERE IS A BETTER WAY TO DO THIS. HERE IT GOES
/**This function searches for the flash dom object that has the flex movie in it.
*In some browsers this is an object tag. In other browsers this is an embed tag.
*We know we have the right thing when we find a dom object with the function getRoot.
**/
function FABridge__bridgeInitialized(bridgeName)
{
var searchStr = “bridgeName=”+ bridgeName;
var flashInstanceWithBridge = null;

var flashInstances = document.getElementsByTagName(“object”);

//search object tag
if (flashInstances.length == 1)
{
flashInstanceWithBridge = flashInstances[0];
}
else
{
for(var i = 0; i < flashInstances.length; i++) { var inst = flashInstances[i]; var params = inst.childNodes; var flash_found = false; for (var j = 0; j < params.length; j++) { var param = params[j]; if (param.nodeType == 1 && param.tagName.toLowerCase() == "param") { if (param["name"].toLowerCase() == "flashvars" && param["value"].indexOf(searchStr) >= 0)
{
flashInstanceWithBridge = inst;
flash_found = true;
break;
}
}
}

if (flash_found) {
break;
}
}
}

if(!flashInstanceWithBridge|| !flashInstanceWithBridge[“getRoot”] || !(typeof(flashInstanceWithBridge.getRoot) == “function”) ) //search embed tag if the object tag does not have the needed methods
{
flashInstanceWithBridge = null;

flashInstances = document.getElementsByTagName(“embed”);
if (flashInstances.length == 1)
{
flashInstanceWithBridge = flashInstances[0];
}
else
{
for(var i = 0; i < flashInstances.length; i++) { var inst = flashInstances[i]; var flashVars = inst.attributes.getNamedItem("flashVars").nodeValue; if (flashVars.indexOf(searchStr) >= 0)
{
flashInstanceWithBridge = inst;
}

}
}
}

if(flashInstanceWithBridge)
{
FABridge.attachBridge(flashInstanceWithBridge, bridgeName);
}

return true;
}

JNI + OSGI + Refactoring + Idiot = 2 wasted days.

I just wasted two days.  I am working on a desktop application in JAVA.  It needs to sniff web proxy settings from the Window’s registry (and some day other platform specific locations).  I decided to use JNI  to do this.  Everything worked like a charm in my sandbox JAVA application.  I then embarked on moving it to and OSGI bundle.  When I ran the thing I received a java.lang.UnsatisfiedLinkError.  WTF!!  WHY!! 

When I debugged my call to System.load(“somedll”) worked, but an exception was triggered when I invoked the native method.  I then thought well maybe its and OSGI thing.  After some Google-ing I found a few write ups on how to tweak the manifest to load the system specific library.  I tried effing with the manifes eight ways from Sunday, but still no workie.  I then came across a forum where someone had the same issue I did and they eventually discovered that that they had renamed their JAVA class, but did not regenerate the JNI lib.  It then dawned on me that I had refactored my JAVA class in to a new package, but did not update the generated .h file.  

The morale of the story is if your JNI call works outside of OSGI, but then stops working when you move it in to OSGI; ask yourself “did I refactor this thing when I moved it in to a bundle?” 

Motion Detection Revisited

About 2 years ago I did some experiments with ActionScript 3 and motion detection.  Looking back I had no idea what what the ActionScript I pieced together was doing.  I have since gained a better understanding of some of the low level bitmap data functions available in AS3.  I threw this together last night and am quite pleased with the result.  It is a virtual drum game.  The game works best if you get out of you chair and position yourself such that your hand is the same size as a drum.  I have separated the motion detection logic so that it can be reused on any DisplayObject.  Also it is not nearly as memory intensive as the first implementation.  I took the sounds from http://www.themaninblue.com/experiment/JS-909/ (thanks Cameron Adams).  Here is a demo.  Here is the source.  This is a proof of concept and certainly not the first AS3 web cam motion detection experiment.  I am curious to hear what you think of my implementation.

Flex Module with in a Module Bug (Fixed in Gumbo)

I know I am not the first to encounter this, but I thought it worth a mention anyway.  I ran in to an issue loading a module from with in a module using flex.  The module would load, but the ModuleEvent.READY would not fire.  The instance of the module was ready and setup was true.  WTF!!! why!  After searching the net I came across this bug https://bugs.adobe.com/jira/browse/SDK-14669.  It said a workaround is to load all the modules sequentially.  Now I know this sucks because 1 by 1 is not exactly a model of efficiency.  Anyhoo…I threw the following class together and thought it would help:


/**
* this class is to get around this bug
* https://bugs.adobe.com/jira/browse/SDK-14669
* when we move to gumbo (flex 4) the bug will be fixed
* */

package com.timothyhuertas.utils
{
import flash.events.EventDispatcher;
import flash.events.IEventDispatcher;
import flash.system.ApplicationDomain;
import flash.system.SecurityDomain;
import flash.utils.ByteArray;

import mx.events.ModuleEvent;
import mx.modules.IModuleInfo;

public class SequentialModuleLoader extends EventDispatcher
{
private static var _instance:SequentialModuleLoader;
private var _pool:Array;
private var _current:IModuleInfo;

public function SequentialModuleLoader(value:SingletonEnforcer)
{
_pool = new Array();
}

public static function get instance() : SequentialModuleLoader
{
if(!_instance)
{
_instance = new SequentialModuleLoader(new SingletonEnforcer());
}
return _instance;
}

public function load(info:IModuleInfo, applicationDomain:ApplicationDomain=null, securityDomain:SecurityDomain=null, bytes:ByteArray=null) : void
{
var queued:Object = new Object;
queued.info = info;
queued.applicationDomain = applicationDomain;
queued.securityDomain = securityDomain;
queued.bytes = bytes;
_pool.push(queued);
processPool();
}

protected function processPool() : void
{
if(_pool.length && _current==null)
{
var queued:Object = _pool[0];
_pool.splice(0,1);
_current = queued.info as IModuleInfo;
_current.addEventListener(ModuleEvent.READY, handleModuleResponse, false,0,true);
_current.addEventListener(ModuleEvent.ERROR, handleModuleResponse, false,0,true);
_current.load(queued.applicationDomain, queued.securityDomain, queued.bytes);
}
}

protected function handleModuleResponse(e:ModuleEvent) : void
{
IEventDispatcher(e.target).removeEventListener(e.type, handleModuleResponse);
_current = null;
processPool();
}

}
}
class SingletonEnforcer{}

An example of its usage is as follows:


var m:IModuleInfo = ModuleManager.getModule("Mod1.swf");
m.addEventListener(ModuleEvent.READY, modLoaded);
SequentialModuleLoader.instance.load(m,ApplicationDomain.currentDomain);

protected function modLoaded(e:ModuleEvent):void
{
IEventDispatcher(e.target).removeEventListener(e.type, modLoaded);
IModuleInfo(e.target).factory.create();
}

I hope this helps

Using C++ to read the Window’s registry.

I am working on a project that requires me to deploy files to Firefox’s plugins directory.  The biggest challenge was finding the location of this directory.  People can have multiple versions of Firefox and may change the default install location.  Since the project targets Windows machines a coworker recommended reading the registry.  The idea sounded good, but it had been a while since I had cracked C++ (let alone Visual C++).  So I took to the web for some examples.  I came across this (http://msdn.microsoft.com/en-us/library/ms235431.aspx) on Microsoft’s web site.   It is a list of sample code for various Visual C++ tasks.  It helped me throw this together.  Below is source for a program that will write out every installed version of Firefox along with the location of the plugins directory.  Enjoy….


// registry_read.cpp
// compile with: /clr
using namespace System;
using namespace Microsoft::Win32;

int main( )
{

RegistryKey^ rk = nullptr;
rk = Registry::LocalMachine->OpenSubKey(“SOFTWARE”)->OpenSubKey(“Mozilla”);
array^subKeyNames = rk->GetSubKeyNames();

for ( int i = 0; i < subKeyNames->Length; i++ )
{
RegistryKey ^ tempKey = rk->OpenSubKey( subKeyNames[ i ] );
RegistryKey ^ extensionKey = tempKey->OpenSubKey( “extensions” );
if (extensionKey!=nullptr)
{
Console::WriteLine( “\nFound {0} here is the plugins dir {1}.”, tempKey->Name, extensionKey->GetValue( “Plugins” )->ToString() );
}

}

return 0;
}

Olderposts