filed in General on Feb.09, 2010
I’m a key accelerator man. I hate taking my hands out of my keyboard just to press Yes or No in a Modal Dialog. Until last week I thought there was no way to tab through the different buttons of a modal dialog.
Luckily I came across an option in System Preferences to enable the tab key navigation:
filed in General on Feb.01, 2010
Here are some little things that might speed you up when working on a Mac.
In terminal type:
1 | defaults write com.apple.finder AppleShowAllFiles TRUE |
To hide them again:
1 | defaults write com.apple.finder AppleShowAllFiles FALSE |
To copy the output of a parameter you can use pbcopy:
1 | pwd | pbcopy |
The result from pwd will be copied to the clipboard. You can seamlessly use pbpaste
Imagine you’ve got 2 folders: v1 and v2 and you want to copy the contents of v1 inside v2 recursively (merging the contents). With a normal cp command you copy v1 to v2 and there’s a bin folder in both, the bin folder in the destiny will be completely replaced with the one found in v1.
If you were in Linux you would use cp -a, in Mac you can use:
1 | cp -pPR v1 v2 |
Where:
-p = preserve
-R = recursive
-P = no symbolic links are followed — can be added but this is the default
If you want to see some colors used when executing ls you can add the folowing lines to ~/.bash_profile:
1 2 3 4 | export TERM=xterm-color export LSCOLORS=fxhxcxdxbxegedabagacad alias ls='ls -la' export CLICOLOR=1 |
The second line changes the default annoying blue color applied to folders with a more readable red color.
If you need to connect via vnc to any computer you don’t have to install a vnc client. Open Finder, press command+K and type vnc://host where host is the ip or domain you want to vnc to. A VNC client will automatically pop up.
If you want to allow incoming VNC connections to your MAC you can enable your VNC server open “System Preferences” > “Sharing” and check the “Screen Sharing” option.
There are cases where you have to execute X applications remotely but the server doesn’t have a VNC server neither the possibility.
In these cases you can use:
1 2 | xhost host ssh -X -A userName@host |
Note that to be able to execute the commands above you need to have X11. The first line basically allows incoming connections from the server. The ssh command stablishes a connection against the host indicating that any application that nees an X server will be run in the local computer.
Once logged in you can try to execute any visual application. You’ll see that the application pops up locally.
You might want to open a random file from the command line in the default desktop application
1 | open aFile.ext |
To open current directory in finder:
1 | open . |
Lots of times you navigate to a certain folder using Finder and then you want to open a Terminal in that folder. To accomplish that you can install openTerminalHere applescript http://www.entropy.ch/software/applescript/
1 2 | defaults write com.apple.dashboard mcx-disabled -boolean NO killall Dock |
1 2 | defaults write com.apple.finder _FXShowPosixPathInTitle -bool YES killall Finder |
Do you know any other must-know command or gem?
filed in General on Dec.10, 2009
We’ve been busy, very busy during the last weeks. But we’re now very happy to announce the release of:
And our first beta for the new FlexPMD Eclipse plugin that will better integrate FlexPMD into your development workflow!!
You can find all the documentation for FlexPMD on the wiki and for the plugin http://opensource.adobe.com/wiki/display/flexpmd/FlexPMD+Eclipse+plugin
If you have any question, problem or suggestion you can use the public forums.
filed in tips on Sep.08, 2009
Lately I’ve been reading several discussions around the internals of EventListener, how easy is to create memory leaks, the dependencies between IEventListener and EventListener and onther fancy things. Robert Penner has lots of details around these topics in his 2 part articles:
Even though not a solution I want to add a new parameter to this equation that might help (at least for debugging purposes).
As per the documentaqtion flash.sampler.getMemberNames returns an object containing all members of a specified object, including private members. If we pass a reference to an EventDispatcher instance we will get a private member called listeners. Yes, you’re right, it’s the list of listeners the EventDispatcher has associated.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 | package { import flash.display.Sprite; import flash.events.Event; import flash.events.EventDispatcher; import flash.sampler.getMemberNames; public class testEvents extends Sprite { public function testEvents() { var myDispatcher:EventDispatcher = new EventDispatcher(); myDispatcher.addEventListener("myEvent1", handler1); myDispatcher.addEventListener("myEvent2", handler2); var members:Object = getMemberNames(myDispatcher); for each (var name:QName in members) { if (name.localName == "listeners") { trace(myDispatcher[name].length); myDispatcher.removeEventListener("myEvent1", myDispatcher[name][0]); myDispatcher.removeEventListener("myEvent2", myDispatcher[name][0]); trace(myDispatcher[name].length); } } } private function handler1(event:Event):void { trace("handler1"); } private function handler2(event:Event):void { trace("handler2"); } } } |
The example above shows how we can get a reference to the different listeners declared in an EventDispatcher and how to remove them without having a direct reference (and knowing the event name arggg).
Even though this can help you to figure out if an EventDispatcher has listeners or not there’re still several problems you might find:
A very uggly implementation for a removeAllListeners method could be something like:
1 2 3 | for each listener in listeners for each eventName that the dispatcher can dispatch eventDispatcher.removeListener( eventName, listener ) for all phases |
If after each removeListener we check the length of the listeners list we could find out the events the listeners were listening to as well as the phase and thean provide some debugging information around it.
filed in General on Sep.02, 2009
FlexPMD is a tool that helps to improve code quality by auditing any AS3/Flex source directory and detecting common bad practices, such as:
FlexPMD can currently be invoked from:
An Eclipse plugin is under consideration.
Some good information on how to start using it can be found in this blog post.
filed in architecture, components, flex, testing, tips on Aug.25, 2009
Sometimes, due to many different reasons, there’re exceptions that happen during the invalidation/validation cycle of the flex SDK. Most times these exceptions end up being unveiled in the very core of the SDK, being not possible to catch them up because their async nature and the place where they happen.
The stacktrace of these exceptiosn, normally, looks like the lines below:
1 2 3 4 5 6 7 | ... at mx.core::UIComponent/validateProperties()[C:\autobuild\3.x\frameworks\projects\framework\src\mx\core\UIComponent.as:5807] at mx.managers::LayoutManager/validateProperties()[C:\autobuild\3.x\frameworks\projects\framework\src\mx\managers\LayoutManager.as:539] at mx.managers::LayoutManager/doPhasedInstantiation()[C:\autobuild\3.x\frameworks\projects\framework\src\mx\managers\LayoutManager.as:689] at Function/http://adobe.com/AS3/2006/builtin::apply() at mx.core::UIComponent/callLaterDispatcher2()[C:\autobuild\3.x\frameworks\projects\framework\src\mx\core\UIComponent.as:8633] at mx.core::UIComponent/callLaterDispatcher()[C:\autobuild\3.x\frameworks\projects\framework\src\mx\core\UIComponent.as:8582] |
Today, snooping around the UIComponent class, I’ve found a nice unexpected thing. Basically the invalidation pattern in the SDK is built around the LayoutManager and the UIComponent classes. The callLater method is the responsible to delay the execution of certain functionalities to the next frame. callLater implementation relies on the RENDER and ENTER_FRAME frames. The handler for these events is the method callLaterDispatcher.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 | // At run-time, callLaterDispatcher2() is called // without a surrounding try-catch. if (!UIComponentGlobals.catchCallLaterExceptions) { callLaterDispatcher2(event); } // At design-time, callLaterDispatcher2() is called // with a surrounding try-catch. else { try { callLaterDispatcher2(event); } catch(e:Error) { var callLaterErrorEvent:DynamicEvent = new DynamicEvent("callLaterError"); callLaterErrorEvent.error = e; systemManager.dispatchEvent(callLaterErrorEvent); } } |
Here’s the surprise. If UIComponentGlobals.catchCallLaterExceptions is true, the execution of the validation cycle will be wrapped with a try..catch and a DynamicEvent will be dispatched through the systemManager with information about the error.
This opens the doors for at least catch the error and log it. In some cases, depending on the severity of the error and if the application is not too corrupted, we’ll even be able to display a message to the user notifying him the error. This is not a fix to FP-1499 nor FP-444 but under some circumstances it can be really helpful if combined with great logging practices as explained by Tom.
I’ve created a simple class that hooks into the systemManager, listens for the callLaterError event and logs it using the standard SDK logging mechanism. The AsyncErrorLogger is MXML friendly so that it can be instantiated in the root application. It provides an enable property to switch the behavior on and off. Additionally the property redispatchErrorEvent, which by default is false, redispatches the error causing the traditional flash player error dialog to appear (in debugger player versions) and not fail silently.
Here’s an example on how to use it:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 | <?xml version = "1.0" encoding = "utf-8"?> <mx:WindowedApplication xmlns:mx = "http://www.adobe.com/2006/mxml" xmlns:utils = "com.adobe.ac.utils.*"> <mx:Script> <![CDATA[ import mx.logging.LogEventLevel; //--------------------------------------------- // THAT'S THE WAY TO SIMULATE THE ASYNC ERROR //--------------------------------------------- private var doItChange : Boolean = false; private function doIt() : void { doItChange = true; invalidateProperties(); } override protected function commitProperties() : void{ super.commitProperties(); if( doItChange ) { doItChange = false; throw new Error("CallLater exception"); } } ]]> </mx:Script> <mx:TraceTarget level = "{ LogEventLevel.ALL }"/> <utils:AsyncErrorLogger/> <mx:Button click = "doIt()" label = "Throw exception"/> </mx:WindowedApplication> |
Here’s the code. You can download the class from here as well.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 | package com.adobe.ac.utils { import flash.events.ErrorEvent; import flash.events.EventDispatcher; import mx.core.Application; import mx.core.UIComponentGlobals; import mx.events.DynamicEvent; import mx.events.FlexEvent; import mx.logging.ILogger; import mx.logging.Log; import mx.managers.ISystemManager; [Event(name="error",type="flash.events.ErrorEvent")] public class AsyncErrorLogger extends EventDispatcher { //------------------------------------------------------------------------ // // Constants // //------------------------------------------------------------------------ private static const CALL_LATER_ERROR:String = "callLaterError"; private static const LOG:ILogger = Log.getLogger("com.adobe.cairngorm.AsyncErrorLogger"); //------------------------------------------------------------------------ // // Properties // //------------------------------------------------------------------------ //------------------------------- // systemManager //------------------------------- private var systemManager:ISystemManager; //------------------------------- // logStackTrace //------------------------------- public var logStackTrace:Boolean = true; //------------------------------- // redispatchError //------------------------------- public var redispatchError:Boolean = false; //------------------------------- // enabled //------------------------------- [Bindable] public function get enabled():Boolean { return UIComponentGlobals.catchCallLaterExceptions; } public function set enabled(value:Boolean):void { UIComponentGlobals.catchCallLaterExceptions = value; } //------------------------------------------------------------------------ // // Constructor // //------------------------------------------------------------------------ public function AsyncErrorLogger(systemManager:ISystemManager = null) { enabled = true; init(systemManager); } //------------------------------------------------------------------------ // // Private Methods // //------------------------------------------------------------------------ private function init(systemManager:ISystemManager):void { if (systemManager) { initializeListener(systemManager); } else { var application:Object = Application.application; if (!application.initialized) { application.addEventListener(FlexEvent.CREATION_COMPLETE, creationCompleteHandler); } else { initializeListener(application.systemManager); } } } private function creationCompleteHandler(event:Event):void { var application:Object = Application.application; application.removeEventListener(FlexEvent.INIT_COMPLETE, creationCompleteHandler); initializeListener(application.systemManager); } private function initializeListener(systemManager:ISystemManager):void { this.systemManager = systemManager; if (this.systemManager) { LOG.debug("AsynErrorLogger initialized."); this.systemManager.addEventListener(CALL_LATER_ERROR, errorHandler); } } private function errorHandler(event:DynamicEvent):void { var error:Error = event.error as Error; LOG.error("{0}", error.message); if (logStackTrace) { LOG.error(error.getStackTrace()); } if (redispatchError) { var errorEvent:ErrorEvent = new ErrorEvent(ErrorEvent.ERROR, false, false, error.getStackTrace(), error.errorID); dispatchEvent(errorEvent); } } } } |
filed in air, testing on Aug.16, 2009
One of the biggest problems around testing installed AIR applications is that testers can’t see when a runtime exception is triggered. Lots of times the symptom of a problem will be something completely different to the real cause and developers will have to spend much more time trying to figure out what was the root cause of the problem. Even though you have a good logging policy if there’s an uncaught exception there’s not too much you can do.
Here are some techniques I’ve came across to make testing less painful for everybody.
One thing I’ve been thinking for a while is that there are not two different versions of the AIR runtime; one with debugging enabled and the other without it (the way Flash Player has). This means that there might be a way to enable and disable debugging in the same runtime. This would allow, at least in some controlled environments, to see Runtime Exceptions and make everybody’s life easier. At least until we have global error handling implemented directly on AIR or Flash Player. Yes, I agree, It’s unbelievable that on its 10th version we still are not able to capture global exceptions: FP-444 and FP-1499
Using Procmon I realized that when an AIR application is launched it tries to read a file in {INSTALLATION_FOLDER}/META-INF/AIR/debug. See the screenshot. What surprised me was the result “NAME NOT FOUND”. Basically the file didn’t exist in the app I was playing with. After some more trial / error, yes, you’re right, this is the flag that switches on and off debugging capabilities.
If an AIR application is launched and an empty file with name debug is found in {INSTALLATION_FOLDER}/META-INF/AIR/debug then, any runtime exception will be displayed in a system modal window! In OS the path for this file will be {APPLICATION_FOLDER}.app/Contents/Resources/META-INF/AIR/debug
So if testers (or users during the earlier versions of your app) are using your application, make sure they create the debug file and attach any stack trace they see to their bug reports.
This is not as easy as the previous solution but it gives the testers more options.
Once installed, an AIR application, is not more than the swf file produced from FB, its application descriptor and a native launcher. Given that we have access to the main application swf file and its corresponding descriptor we can use ADL to launch the installed application instead of the native launcher.
Here’s how:
1. Installed the SDK you want to use from http://opensource.adobe.com/wiki/display/flexsdk/Downloads
2. Uncompress the zip in the location you prefer. I’ll refer to this folder as {FLEX_HOME}
In Windows
3. Right click on “My Computer” > Properties
4. Go to “Advanced” tab
5. Click in the option “Environment Variables”
6. In the “System variables” find the “Path” definition and add this to it (double clicking on it): “;{FLEX_HOME}/bin”
7. Make sure your replace {FLEX_HOME} with the path to the folder where you’ve uncompressed the SDK
In OS X
3. Open a terminal
4. Execute this command: open ~/.bash_profile
5. Add the following line to the opened file:
PATH=${PATH}:{FLEX_HOME}/bin
6. Make sure your replace {FLEX_HOME} with the path to the folder where you’ve uncompressed the SDK
In Linux
3. Open a terminal
4. Edit your profile file ~/.profile
5. Add the line: export PATH=$PATH:$FLEX_HOME
Then to launch the application you will have to:
In Windows
1. Open a command line window. Start > “Run …” > cmd
2. Navigate to the folder where the application is installed: cd “Program Files”/MyApp
3. Execute this command: adl META-INF/AIR/application.xml .
4. Don’t forget the dot at the end of the command
In OS X and Linux
1. Open a terminal
2. Navigate to the folder where the application is installed: cd /soft/testAirApp.app/Contents/Resources where testAirApp is the name of your application
3. Execute the following command: adl META-INF/AIR/application.xml .
4. Don’t forget the dot at the end of the command
If during the execution process there’s an exception, ADL will show it.
filed in flash builder, tips on Jun.09, 2009
Flash Builder now includes the ability to filter code hints depending on whether the the hint is a property, an style, an event or an effect. Simply press ctrl+space to cycle through them all.
filed in flex, flexunit, functional testing, presentations on Jun.09, 2009
Here are the slides I used during my presentation around testing, FlexUnit4 and RIATest.
filed in flex, presentations on Jun.09, 2009
Here are the slides and sample files I used for the Hello Gumbo presentation I did for the Spanish Flex User group in the context of the Global AUG tour.