Rajeeshcv.com

Sharing my knowledge

EmberJS Error Logging in Production

In order to log all errors in an EmberJS application, EmberJS provides a global hook called onerror

Here is a link to the official documentation - http://emberjs.com/guides/understanding-ember/debugging/#toc_implement-an-ember-onerror-hook-to-log-all-errors-in-production

Code sample from the documentation (v 1.5.1) look like this

  Ember.onerror = function(error) {
    Em.$.ajax('/error-notification', {
      type: 'POST',
      data: {
        stack: error.stack,
        otherInformation: 'exception message'
      }
    });
  }

What this piece of code does is, whenever an error occurs that stack trace is send to the server using an AJAX request. This will works great for most of the scenarios but....

Consider a situation where an error occurs inside a mouse move event handler. Then you will be trying to sending hundreds of requests to server in short interval. This can bring down your server if not handled properly, it's like DDOS-ing your own server.

Good approach will be to control the rate at which errors are send to the server for logging. This can be done using a concept called Throttling - which ensures that the target method is never called more frequently than the specified period of time.

Google for JavaScript Throttling you can find lot different implementations. EmberJS also provide it's own throttling function Ember.run.throttle

Here is the example from the official EmberJS website

  var myFunc = function() { console.log(this.name + ' ran.'); };
  var myContext = {name: 'throttle'};

  Ember.run.throttle(myContext, myFunc, 150);
  // myFunc is invoked with context myContext

This throttling function can be used for sending errors to the server. Here is a working demo which shows the throttled error logging -
http://cvrajeesh.github.io/emberjs/errorhandling.html

EmberJS - How to Detect User Inactivity

Scenario

I am working on a web application that has a Single Page Application module which is built using EmberJS framework.

This EmberJS application is a page designer where user can drag and drop items to the design surface and change properties of these items. Finally when the page is ready, he can save the changes back to server.

Problem

Whenever a user works on the page designer for long period of time without saving the changes in between - finally when he try to save all of his changes in one shot, server side session was already timed out.

Solution

Implemented an API on the server side which can refresh the session if this API is invoked.

So on the EmberJS application side we had to calculate how long a user not active. Then we can check for this inactivity time periodically and decide whether to refresh session or not.

An easy way to do it in EmberJS is by hooking to the 'mousemove', 'keydown' events of the ApplicationView. Details can be found in the EmberJS documentation http://emberjs.com/api/classes/Ember.View.html#toc_responding-to-browser-events

Here is code sample which does this

First I am creating the application object with an additional property for tracking the last active time.

var App = Ember.Application.create({
    lastActiveTime: null,
    ready: function(){
        App.lastActiveTime = new Date();
    }
});

After that, hook the events in which you are interested in ApplicationView, which will update the last active time defined in the application object.

  App.ApplicationView = Ember.View.extend({
    templateName: 'application',

    mouseMove: function() {
        App.lastActiveTime = new Date();
    },

    touchStart: function() {
        App.lastActiveTime = new Date();
    },

    keyDown: function() {
        App.lastActiveTime = new Date();
    }
});

In our case, I am interested on 'mouseMove', 'touchStart' (for touch devices) and 'keyDown' events. When they are getting fired I will update the last active time with current time.

Now you can use the last active time property to find out how long the user is active.

Demo

http://cvrajeesh.github.io/emberjs/inactivity.html

This demo uses moment.js for calculating time difference as well as for formatting the time.

Visual Studio : Debugging Silverlight Application with Chrome and Firefox

Here is a tip if you want to debug Silverlight applications from Visual Studio 2012 if it is running in Google Chrome or Firefox

Google Chrome

  1. Run the application and select Debug –> Attach to Process from the menu
  2. From the Available Processes list, look for process named “chrome.exe” with type “Silverlight”
  3. Select that process and click the “Attach” button

Here is screenshot from my machine

Mozilla Firefox

Firefox has a slightly different approach

  1. Run the application and select Debug –> Attach to Process from the menu
  2. From the Available Processes list, look for process named “plugin-container.exe”
  3. Select that process and click the “Attach” button

Here is the screenshot

Voila!!! now you are able to get the breakpoints in Visual Studio

One of the Top E-Commerce Website in India Stores Passwords Incorrectly

Disclaimer: I am not a security expert nor a hacker, but just a developer.

This incident made me feel very bad – couple of weeks before I ordered a product from naaptol a well-known E-Commerce website in India.  It was the first time I was ordering something from them. So I had to register first, after the registration was completed got a confirm email from them

Oops!!! My password is in clear text.

I thought they might be generating the email during the registration process before it is stored in their database. Without bothering much, I ordered the item. Thought about investigating about the “clear text password” little bit, but somehow forgot it.

After few days when I got the product which I have ordered, it reminded me about this password incident. So in order see whether they are hashing my password or not, I used their “Forgot my password” option. It asked for my email address. After the submission got a message saying

We' have sent you an e-mail at the submitted ID including instructions. You'll be back to your shopping place in no matter of time.

As expected, got an email from them

It reads “Here is your new Login and Password” and surprisingly password they gave is same as my old password even though in the email it is mentioned that they are sending a new password. It confirmed that they are not hashing the passwords.

Will they be storing it in plain text? Who knows…

Did someone tell you internet is not a good place to store your secrets?

I tried to play a nice role, so sent them an email telling about about the password hashing problem. You know what happened… no reply from them till now.

If you are in the naaptol technical team, convince your boss about the importance of securing the password and push that functionality in the next release.

If you are a non-technical manager in naaptol, tell your developers to read this article from Jeff - You're Probably Storing Passwords Incorrectly

Automating WYSIWYG editors using Selenium web driver

Automating web application using selenium for acceptance/integration tests is very easy. Recently I have faced few issues when automating a page with a JavaScript based WYSIWYG editors.

The reason why it is hard because most of these editors create a new editable HTML document inside an iframe. There are two ways I aware that you can set text on these editors

  1. Executing a JavaScript code in the current page
  2. Sending keys on the editable iframe

Here is how I automated two WYSIWYG editors using selenium web driver.

bootstrap-wysihtml5


private void SetBootstrapWysihtml5Content(IWebElement textArea, string content)
{
    string script = string.Format(@"$('#{0}').data('wysihtml5').editor.setValue('{1}');", textArea.GetAttribute("id"), content);
    ((IJavaScriptExecutor)this.WebDriver).ExecuteScript(script);
}

What this method does is, it accepts the TextArea web element which you are converting to an editor and the actual content to be set. Then it executes a piece of JavaScript on the current WebDriver context.

For e.g. if you have a TextArea element with id “summary” then executing below JavaScript statement will insert a “Test” as h1 tag to the editor

$('#summary').data('wysihtml5').editor.setValue('<h1>Test</h1>');

TinyMCE

private void SetTinyMCEContent(IWebElement textArea, string content)
{
    string textAreaId = textArea.GetAttribute("id");
    string frameId = string.Format("{0}_ifr", textAreaId);
    var frame = this.WebDriver.FindElement(By.Id(frameId));
    this.WebDriver.SwitchTo().Frame(frame);
    ((IJavaScriptExecutor)this.WebDriver).ExecuteScript("document.body.innerHTML = '" + content + "'");
    this.WebDriver.SwitchTo().DefaultContent();
}

This method also accepts the TextArea web element which you are converting to an editor and the actual content to be set. Then it finds the corresponding iframe and set the innerHTML by executing a JavaScript statement.

Workaround for Web Essentials to support TypeScript 0.8.1.0

If you have installed the latest version of TypeScript(0.8.1.0) ,  Web Essentials extension you have installed on the VS 2012 may not be working correctly. Here is a work around to get it working

  1. Open command prompt as an administrator and go the TypeScript installation directory. In my case it is C:\Program Files (x86)\Microsoft SDKs\TypeScript.
  2. If the older version of TypeScript (0.8.0.0 or below) is there in that directory, delete it
    Create a symbolic of the older version, which points to the new version using the following command
    mklink /D 0.8.0.0  0.8.1.0
  3. Now open VS 2012 and go to Tools -> Options -> Web Essentials -> TypeScript. Set “Compile all TypeScript files on build” to “False”

Introducing SourceCodeReader

Have you ever tried to understand a project just by going through the source code?

That’s how I do most of the time, till now I haven’t seen any up to date documentation which gives the clear understanding of how the code is implemented. In reality after a certain period during the development, documentation goes out of sync with the source code. So some one joins the project lately has only one way to understand the project “Go through the source code

Sometimes I go through the open source projects hosted on either Github, Codeplex or Google code. All the hosting site provide you with a user interface for browsing through the source code with syntax highlighting support, which is nice. The problem I had with this is – if I see a object creation statement, method call there is no way for me to find out how it is implemented other than manually going through the files and finding it out.

Most of the full blown editors provide a feature where you can navigate to the implementation directly from where it is used for e.g. Microsoft Visual Studio provides “Go To Definition” feature. I find this feature very much useful for understanding the project. Only draw back with this approach is you have to completely get the latest version of source code to your local machine.

SourceCodeReader is trying solve this navigation problem on the web. With this application you can open a project and browse through the files with code navigation support.

Source code for this project can be found at https://github.com/cvrajeesh/SourceCodeReader

Limitations

  1. Now supports only Github code repository
  2. Only supports .Net C#  projects, other type of projects also works without code navigation support

* Caution: This is an early release you will see bumpy roads.

 

How to use this application

 

Go to SourceCodeReader and enter the URL of a C# project on Github

image

 

Once you have entered a Github project link and open the project, it get the source code from the Github and present you with file browser.

image

 

When you navigate to a C# file you will be able to see clickable links for identifiers which takes you the file location where that is declared.

image

 

Sample projects

  1. http://sourcecodereader.apphb.com/#/open/cvrajeesh/SourceCodeReader
  2. http://sourcecodereader.apphb.com/#/open/ayende/ravendb

Hope you liked this idea, please provide me with your valuable feedback.

Attach to Any ASP.NET Web Server from Visual Studio in One Click

This is an update to my previous blog post Attach to Visual Studio Development Server with One Click.

The Visual Studio Macro from previous article doesn’t support IISExpress or IIS; it only supported the Visual Studio Development Server, more over it doesn’t detect latest Development Web Server “WebDev.WebServer40.exe”.

Now I have updated the Macro so that it will automatically detect the Web Server setting from the project properties and attach it accordingly.

Read more...

Attach to Visual Studio Development Server with One Click

Update: Enhanced version of the Macro created here is available in the new article -  Attach to Any ASP.NET Web Server from Visual Studio in One Click

In my day to day work, during the development I had to attach an ASP.NET application to the development server (Cassini) several times in order to debug and fix a problem.

This task is little bit time consuming because this is how we normally do it

  1. Click on the “Attach to Process” menu under the Debug menu
  2. Select the correct process from the list of available processes
  3. Either double click on the select process or click the “Attach” button

You can reduce these into two steps, if you assign a short cut key to the “Attach to Process” command.

What I found is most of the time is lost in finding and selecting the correct process from the available list of processes in the “Attach to Process” dialog.

Read more...