Creating a thread save queue

We’ve been working on a WPF project for a costumer some time now. One of the requirements of the application was the storage and live streaming of multiple network cameras.

Due to the project setup we decided to split the storage and the viewing part. A client application will stream the camera’s at a rate of 25 frames per second and a server application will record the frames at a rate of 2 frames a second. The camera’s (Mobotix and Axis) can handle the 2 streams and also the network capacity is high enough to create 2 different streams to the cameras.

For the server application I started with implementing the AForge framework (http://www.aforgenet.com/) to handle the camera streams. The Aforge Library has excellent support for audio and video streaming/editing. In our server application we only make use of the AForge.Video namespace. Setting up a stream to a network camera is quite simple.

public CameraStream(string ipAddress, int frameRate)
{
	string url = string.Format(@"http://{0}/control/faststream.jpg?stream=full&fps={1}", ipAddress, frameRate);
	MJPEGStream mjpegStream = new MJPEGStream(url);
	mjpegStream.NewFrame += VideoNewFrame;
	mjpegStream.VideoSourceError += VideoSourceError;
}

private void VideoNewFrame(object sender, NewFrameEventArgs eventArgs)
{
	//Add implementation
	throw new NotImplementedException();
}

private void VideoSourceError(object sender, VideoSourceErrorEventArgs eventArgs)
{
	//Add implementation
	throw new NotImplementedException();
}

Create a MJPEGStream with the URL of the camera as parameter. Attach 2 event handlers, one when a new frame is received, one if an error occurs. The error handling is not part of this post where we focus on asynchronously and thread save usage of a queue. Next part is to implement the event handler when a new frame is received. In the event argument a bitmap is exposed as current frame. We can save this bitmap to disk.

private void VideoNewFrame(object sender, NewFrameEventArgs eventArgs)
{
	//Get the bitmap from the stream
	Bitmap bitmap = eventArgs.Frame;

	//Set the path where we want to save the image
	string path = string.Format(@"d:\Temp\{0}.jpg",DateTime.Now.ToString());
	
	//Save the image
	bitmap.Save(path);
}

This seems allright and if we test with low number of cameras it seems to be running just fine. If we deployed the application to the test environment with 24 cameras we noticed the camera stream stopped every minute for a few seconds. After some debugging we found that the disk where the file was saved had some lack and blocked the saving part of our code for 20 – 30 seconds. Due to our implementation to save directly in the event handler all streams were halted and no images got recorded.

Up to asynchronously saving of the images. If access to the disk was blocked for a reason the image streams have to keep on streaming and the images need to be put in a queue. When the disk is writable again we can empty the queue and start saving to the disk. We first created a helper class to hold the images and to manage the queue. First implementation was a simple list that could hold objects with as parameters a Bitmap and a string (the image and the path where it has to be saved).

public class SaveImageQueue
{
	public List<ImageToSave> Queue { get; set; }
	private Object thisLock = new Object();

	/// <summary>
	/// Constructor
	/// </summary>
	public SaveImageQueue()
	{
		Queue = new List<ImageToSave>();
	}
	
	public void AddImage(Bitmap image, string path)
	{
		lock(thisLock)
		{
			Queue.Add(new ImageToSave(){FilePath = path,Bitmap=image});
		}
	}
	
	public ImageToSave GetImage()
	{
		lock(thisLock)
		{
			return Queue.FirstOrDefault();
		}
	}
}

/// <summary>
/// Helper class to add images and their filepath in the queue
/// </summary>
public class ImageToSave:IDisposable
{
	public string FilePath { get; set; }
	public Bitmap Bitmap { get; set; }

	/// <summary>
	/// Disposes the Bitmap in the helper class
	/// </summary>
	public void Dispose()
	{
		Bitmap.Dispose();
	}
}

In the code above you can see we use a lock object to be sure only one thread at the time can access the List<T>. In the new frame event handler we can just call the AddImage method to add an image to the queue. In a separate thread we will consume the GetImage method to fetch the first image from the queue and when it’s not null we can save it to disk.

If we put this to test we could notice almost directly that the List<T> got locked to much by the 24 camera streams and the consumer to save the images. Our simple solution wasn’t up to this load. When searching for a solution we stumbled upon the white paper “Thread-safe Collections in .NET Framework 4 and Their Performance Characteristics” you can find in this blog post. From the 4.0 version .Net has his built in thread-safe collections that perfect fit our needs. We dropped our own written lock system and added the BlockingCollection from the System.Collections.Concurrent namespace.


/// <summary>
/// Queue to async save images to disk
/// </summary>
public class SaveImageQueue
{
	public BlockingCollection<ImageToSave> Queue { get; set; }

	/// <summary>
	/// Constructor
	/// </summary>
	public SaveImageQueue()
	{
		Queue = new BlockingCollection<ImageToSave>();
	}

	/// <summary>
	/// Start the queue handling. Will check the queue and then saves the images one by one.
	/// </summary>
	public void StartQueue()
	{
		Task.Factory.StartNew(() =>
								  {
									  while (true)
									  {
										  ImageToSave imageToSave = null;
										  if (Queue.TryTake(out imageToSave))
										  {
											  Log.DebugFormat("1: Saving image from queue to {0}", imageToSave.FilePath);
											  try
											  {
												  imageToSave.Bitmap.Save(imageToSave.FilePath);
												  imageToSave.Dispose();
												  Log.DebugFormat("1: Queue is holding {0} images", Queue.Count);
											  }
											  catch (Exception ex)
											  {
												  Log.Error("1: Error reading and executing queue", ex);
											  }
										  }
									  }
								  });
}

/// <summary>
/// Helper class to add images and their filepath in the queue
/// </summary>
public class ImageToSave:IDisposable
{
	public string FilePath { get; set; }
	public Bitmap Bitmap { get; set; }

	#region Implementation of IDisposable

	/// <summary>
	/// Disposes the Bitmap in the helper class
	/// </summary>
	public void Dispose()
	{
		Bitmap.Dispose();
	}

	#endregion
}

As you can see we can now use the TryTake method to fetch an object from the collection and to save the image. The BlockingCollection will take care of the locking of the object and if we want we can add multiple consumers so the queue will be emptied faster. All this is put is a asynchronous Task so the camera streams will not notice anything from the saving of the Bitmap.

In the camera new frame event we can add the committing of the image to the queue. One note here is that you best add a clone of the bitmap you extracted from the event arguments. If the queue is emptied rapidly and you don’t use a clone of the image you’ll get GDI exceptions all around the place due to concurring accessing the same memory address space.

private void SaveImage(string filePath, Bitmap bitmap)
{
	queue.Queue.Add(new ImageToSave() { Bitmap = (Bitmap)bitmap.Clone(), FilePath = filePath });
	bitmap.Dispose();
}

Using jQuery in ASP.NET applications – part 2 MVC

This is the second post in the “Using jQuery in ASP.NET applications” series where we tackle the how you can access your code base with a jQuery ajax call.

Setting up the project

Open up Visual Studio and create a new ASP.NET MVC 3 Web Application

Enter a name and a location and click the OK button. Visual Studio will now ask you if you want to create an empty, an intranet or an Internet application. Let’s choose the last one. For the view engine choose Razor. We don’t need any unit tests in this demo so leave the “create a unit test project” unchecked and click the OK button.

Visual Studio will now create a standard MVC web project where jQuery and Modernizer are included. When our project is created it will look like this:

We will reuse the simple demo application we created in the previous blog post that will let a user add his favorite book title to his wish list.

Alter the default view

Like in the Web Forms application we’ll alter the default view Visual Studio created for us. Open up the solution explorer and go to the Home folder in the Views folder. Double click Index.cshtml file. Remove the default message and add a textbox, a button and a listbox so our view looks like:

@{
    ViewBag.Title = "Home Page";
}

<h2>@ViewBag.Message</h2>
<input name="BookTitle" id="txtBookTitle" />
<input type="button" id="btnAddBookTitle" value="Add book title"><br/>
<select id="lstBookTitles" name="lsstBookTitles" size="4"></select>

Hit run (F5) in Visual Studio and our fabulous designed application will open up in a browser. As mentioned before, the focus is this post is the implementation of the ajax call, not the design of the application.

Add our business object

In our application we’re going to use a class to exchange data from the server to the client in the ajax call. Open up the solution explorer and right click on the Models folder and choose Add => Class. Name the class Book and add the following properties so our class looks like this:

using System;
using System.Collections.Generic;
using System.Linq;
using System.Web;

namespace Demo.Models
{
    public class Book
    {
        public long BookId { get; set; }
        public string BookTitle { get; set; }
    }
}

Reference and create javascript files

Unlike in Web Forms, a MVC application has default a reference to the jQuery javascript file. In solution explorer open up the Shared folder in the Views folder. There you’ll find a _Layout.cshtml class. This class is more or less the MasterPage for a MVC application. In the head tag you’ll already find a reference to the jQuery-$version$.min.js file. We don’t want to include our custom javascript in the view as it then will be loaded in every screen of our application. This is first unnecessary and secondly will slow down the page as he has to load multiple sources from the server. In MVC you can create a section in the master view. In your “child” view you can refer to that section to add custom scripts our styles into the page. Add the @RenderSection helper in the head block of the _Layout.cshtml view as shown below.

<!DOCTYPE html>
<html>
<head>
    <meta charset="utf-8" />
    <title>@ViewBag.Title</title>
    <link href="@Url.Content("~/Content/Site.css")" rel="stylesheet" type="text/css" />
    <script src="@Url.Content("~/Scripts/jquery-1.5.1.min.js")" type="text/javascript"></script>
    <script src="@Url.Content("~/Scripts/modernizr-1.7.min.js")" type="text/javascript"></script>
    @RenderSection("JavaScript", required: false)
</head>
<body>
.....

Create our custom javascript file

In solution explorer right click on the Scripts folder and choose Add => New Item. In the window that will appear select Jscript file, name it Home.js and click the Add button.

Reference our custom javascript file in our view

We still have to create a reference to our custom created Home.js file in our view. Open up the index.cshtml again and add the section tag at the top of the view. In that section add a reference to the Home.js file.

@section JavaScript
{
    <script src="../../Scripts/Home.js" type="text/javascript"></script>
}
@{
    ViewBag.Title = "Home Page";
}

<h2>@ViewBag.Message</h2>
<input name="BookTitle" id="txtBookTitle" />
<input type="button" id="btnAddBookTitle" value="Add book title"><br/>
<select id="lstBookTitles" name="lsstBookTitles" size="4"></select>

Hit Run (F5) again in Visual Studio and our page should still look the same.

Create the Ajax Method in the controller

In MVC adding a ajax method is as simple as adding an action in our controller. Instead of returning a View we’ll return a JsonResult that will convert our .NET book class to a Json representation.

private static string sessionName = "BooksSessionName";

public ActionResult AddBook(string bookTitle)
{
	List<Book> books;
	if (Session[sessionName] != null)
	{
		books = Session[sessionName] as List<Book>;
		if (books == null)
			books = new List<Book>();
	}
	else
	{
		books = new List<Book>();
	}
	Book book = new Book() { BookId = GetNextBookTitleId(), BookTitle = bookTitle };
	books.Add(book);
	Session[sessionName] = books;
	return Json(book,JsonRequestBehavior.AllowGet);
}

private long GetNextBookTitleId()
{
	if (Session[sessionName] != null)
	{
		List<Book> books = Session[sessionName] as List<Book>;
		if (books != null)
		{
			long ticket = books.Max(t => t.BookId);
			ticket++;
			return ticket;
		}
	}
	return 1;
}

I’ve added a static string to hold the session name and a private helper method to generate an id for the added book based on the list in the session. In the return statement we create the Json object we want to send to the client. Don’t forget the JsonRequestBehavior.AllowGet flag to allow GET requests to this method!

Create the Ajax call in jQuery

Now our code behind method is available we only deed to call this method from the client. When the button is clicked we’ll call the method and add the returned object to our list. Let’s start with creating an event handler for the click event on the button. Open up the Home.js file we created and add the following lines:

$(document).ready(function () {
    $("#btnAddBookTitle").click(function () {
        alert('Button is clicked !!');
        return false;
    });
});

If we now run our application (hit Run in VS or F5) and we press the ‘Add book tile ‘ button well see a javascript popup telling we’ve clicked the button. Now we can add the ajax implementation.

$(document).ready(function () {
    $("#btnAddBookTitle").click(function () {
        var $bookTitle = $("#txtBookTitle").val();
        $.ajax({
            type: "GET",
            url: "/Home/AddBook/",
            data: {bookTitle:$bookTitle},
            contentType: "application/json; charset=utf-8",
            dataType: "json",
            success: function (msg) {
            },
            error: function () {
                alert('Error calling AddBook');
            }
        });
        return false;
    });
});

I’ve added a variable $bookTitle that holds the value the user has entered in the text box. After that line we’ll start the ajax call. I’ve added some settings in that call:

  • Type: can be POST or GET, in this case: GET
  • url: The url where the values should be posted to: the controller name + action name
  • The data: we have to send the booktitle to our code behind method. This we will rap in a JSON format (more info on http://www.json.org)
  • ContentType: We declare we want to use JSON to communicate with the server
  • Datatype: We declare we want to use JSON to receive data from the server
  • Success: what we have to do when the call is succeeded (not filled out yet)
  • Error: what we have to do when the call throws an error.(alerting a message box in this case)

If we run the application and open up a Firebug screen (in Internet explorer you can use the Developer Tools as you can use these in the Chrome browser). After clicking the button we see a async request to the server that returns an JSON object representing our book.

Implement success function

All we now have to do is to implement the success function so our returned book is added to the list of book titles.

$(document).ready(function () {
    $("#btnAddBookTitle").click(function () {
        var $bookTitle = $("#txtBookTitle").val();
        $.ajax({
            type: "GET",
            url: "/Home/AddBook/",
            data: {bookTitle:$bookTitle},
            contentType: "application/json; charset=utf-8",
            dataType: "json",
            success: function (msg) {
                var $book = msg;
                var $option = new Option($book.BookTitle, $book.BookId);
                $("#lstBookTitles").append($option);
            },
            error: function () {
                alert('Error calling AddBook');
            }
        });
        return false;
    });
});

As we’re receiving an javascript object (JSON) we can create the option using the parameters of the object.

If your un the application now you’ll see the book titles are added to the list, all with their id. From here you can implement delete, edit and other functions.

Next post will be how we can refactor the above so we create an API that’s not tied to a specific controller.

 

Using jQuery in ASP.NET applications – part 1 Web Forms

After the previous post (Why using jQuery for AJAX calls before the Ajax Control Toolkit)we going to see in this post how you can set up a Web Forms project and how you can easily can create Ajax calls to your code behind.

Setting up the project

Open Visual Studio and create a new ASP.NET Web Application.

Enter a name and a location and click the OK button. Visual Studio will now create a standard Web Forms application. In this example I’m using the .Net 4.0 framework although older versions will work as well (up to 2.0).

When our project is created it will look like this:

As demo we’ll create a simple application that allows the user to add a book to his wish list. The user has to fill his favorite book title. After pressing add we’ll show the added book in a list underneath. The user can add as many book titles as he wishes. Let’s start with updating the Default.aspx page with some text and a textbox where the user can add the book title. Under this textbox we place a button to add the title to the list and the list itself. As you can see this is a VERY simple application where the emphasis lies on the creation of the Ajax call, not the application itself.

<%@ Page Title="Home Page" Language="C#" MasterPageFile="~/Site.master" AutoEventWireup="true"
    CodeBehind="Default.aspx.cs" Inherits="Demo._Default" ClientIDMode="Static"  %>

<asp:Content ID="HeaderContent" runat="server" ContentPlaceHolderID="HeadContent">

</asp:Content>
<asp:Content ID="BodyContent" runat="server" ContentPlaceHolderID="MainContent">
  <h2>
        Welcome to ASP.NET!
    </h2>
    <p>
        Enter your book title for your wish list
    </p>
    <asp:TextBox runat="server" ID="txtBookTitle" Width="200"></asp:TextBox>
    <asp:Button runat="server" ID="btnAddBookTitle" Text="Add book title" /><br/>
    <asp:ListBox runat="server" ID="lstBookTitles"></asp:ListBox>

</asp:Content>

One side note: I declared the ClientIDMode Static in the page directive. This will be necessary to implement the jQuery with the same id’s we have given the controls in the aspx page. More info on this blog post of Scot Guthrie.

Our fabulous application now should look like the image below.

Add our business object

In our application we are going to use a class to store our book title information in. In the root of the application add a class. Name this class Book and add the following properties so our class looks like this:

using System;
using System.Collections.Generic;
using System.Linq;
using System.Web;

namespace Demo
{
    public class Book
    {
        public long BooktitleId { get; set; }
        public string BookTitle { get; set; }
    }
}

Reference and create javascript files

In the solution explorer right click on the Scripts folder and choose Add => New item. In the window that appears select a JScript file, name it Default.aspx.js and click ok. Open up our Default.aspx page and add following lines in the HeaderContent block:

<asp:Content ID="HeaderContent" runat="server" ContentPlaceHolderID="HeadContent">
    <script src="Scripts/jquery-1.4.1.min.js" type="text/javascript"></script>
    <script src="Scripts/Default.aspx.js" type="text/javascript"></script>
</asp:Content>

This will add the jQuery library and our own created javascript file to the project.

Now we’re ready to start coding.

Create Web Method

To access our code behind file from javascript we have to add a WebMethod. Open up the Default.aspx.cs code file and create the following method:

        public static long AddBook(string bookTitle)
        {

        }

Above these method we can add an annotation to tell the framework we want to access this method directly. Therefor we need the System.Web.Services namespace. Add this namespace in our using declarations:

using System.Web.Services;

Above the method Add Book add [WebMethod(EnableSession = true)]

        [WebMethod(EnableSession = true)]
        public static long AddBook(string bookTitle)
        {

        }

Now we can create the method body. It’s not that much work, we’re going to store the added book titles in a session object.

        [WebMethod(EnableSession = true)]
        public static long AddBook(string bookTitle)
        {
            List<Book> books;
            if (HttpContext.Current.Session[sessionName] != null)
            {
                books = HttpContext.Current.Session[sessionName] as List<Book>;
                if (books == null)
                    books = new List<Book>();
            }
            else
            {
                books = new List<Book>();
            }
            Book book = new Book() { BooktitleId = GetNextBookTitleId(), BookTitle = bookTitle };
            books.Add(book);
            HttpContext.Current.Session[sessionName] = books;
            return book.BooktitleId;
        }

        private static long GetNextBookTitleId()
        {
            if (HttpContext.Current.Session[sessionName] != null)
            {
                List<Book> books = HttpContext.Current.Session[sessionName] as List<Book>;
                if (books != null)
                {
                    long ticket = books.Max(t => t.BooktitleId);
                    ticket++;
                    return ticket;
                }
            }
            return 1;
        }

I’ve created a helper method to get the next available Id so our list can’t contain the same id twice. Build your application and hit Run. Our Default page should open in a browser window without any changes.

Creating our Ajax call in jQuery

Open up the Default.aspx.js file we created. In this file we will add an event handler on the submit button. When the button is clicked we’ll call the code behind method and add the title to the list stored in the session variable. Add the following lines to the file and run the application.

$(document).ready(function () {
    $("#btnAddBookTitle").click(function () {
        alert('Button is clicked !!');
        return false;
    });
});

When you click the “Add book title” button a javascript popup will appear. The return false line will stop the page from executing a post back as we are going to call a ajax method instead.

Adding the ajax call is not more than a few lines:

$(document).ready(function () {
    $("#btnAddBookTitle").click(function () {
        var $bookTitle = $("#txtBookTitle").val();
        $.ajax({
            type: "POST",
            url: "Default.aspx/AddBook",
            data: "{ 'bookTitle':'" + $bookTitle + "' }",
            contentType: "application/json; charset=utf-8",
            dataType: "json",
            success: function (msg) {
            },
            error: function () {
                alert('Error calling AddBook');
            }
        });
        return false;
    });
});

We’ve added a variable $booktitle where we fetch the title that our user has entered in the textbox. After that line we start the ajax call. There are a few settings we’ve got to set before the call will work:

  • Type: can be POST or GET, in this case: POST
  • url: The url where the values should be posted to: the page name + the method name (“Default.aspx/AddBook”)
  • The data: we have to send the booktitle to our code behind method. This we will rap in a JSON format (more info on http://www.json.org)
  • ContentType: We declare we want to use JSON to communicate with the server
  • Datatype: We declare we want to use JSON to receive data from the server
  • Success: what we have to do when the call is succeeded (not filled out yet)
  • Error: what we have to do when the call throws an error.(alerting a message box in this case)

This is all we have to add to contact our code behind method. If you run the application and open up Firebug in firefox or the developer tools in Internet Explorer 9 or the developer tools in Google Chrome well see in the NET tab we make a async call to the server and receive the id of the book title we’ve added.

Response: {"d":4}

Implement success function

All we have to do now is add this data to our list view. Therefor we’ll have to implement the success function in our ajax call:

            success: function (msg) {
                var $bookId = msg.d;
                var $option = new Option($bookTitle, $bookId);
                $("#lstBookTitles").append($option);
            }

We create a variable $bookId to store the id in we receive from the server. As you can see the return message is wrapped in a “d” variable. This is changed in the .NET 3.5 version. More info in Dave Ward’s post on Encosia.
After that we create an option to be added in the listbox and append it to this box.

If we run the application now well see that every time a book title is added, it’s send asynchronously to the server. We receive an id from the server and the book title is added to the listbox.

Next post will be about implementing AJAX calls in ASP.NET MVC 3 project.

Why using jQuery for AJAX calls before the Ajax Control Toolkit

Every starting developer in ASP.NET who wanted to integrate AJAX functionality in his web application first started with the Ajax Control Toolkit. It’s easy to understand, in Web Forms it’s almost not more than a drag and drop implementation.

After a while you start to notice the limitations. You want to create a functionality that’s not integrated in the toolkit. You start with extending methods, extending properties and get stuck every time again. The amount of update panels start to grow above your head and on every postback you’re have to test if the postback is a Ajax post back or not. In your UI you start to see some disadvantages, the focus is lost when a update panel is updated, you’re page jumps back to the top and so one.

One of the most disturbing issues I found was the load from and to the server. For a simple update of a textbox, the whole content of the update panel is send in the background to update your page. In the first screenshot you see a (very) simple example. The value typed into the textbox has to be added to the list underneath without submitting the page. After creating the update panel and added a script manager we implemented the async postback.

The only thing we needed back from the server is the ID of the value entered and the value itself (although that isn’t needed to make a round trip) I we check the Console or Net tab in Firbug we see the response:

1|#||4|471|updatePanel|UpdatePanel1| <span id="lblFirstName">Bart</span><br /> <span id="lblLastName">De Meyer</span><br /> <span id="Label1">Ticket</span><br /> <input name="txtTicket" type="text" id="txtTicket" /><br /> <input type="submit" name="btnSave" value="Save" id="btnSave" /><br /> <select size="4" name="lstTickets" id="lstTickets">     <option value="1">aaaa</option> </select> |148|hiddenField|__VIEWSTATE|/wEPDwUJNTU0OTM3MDUxD2QWAgIDD2QWAgI DD2QWAmYPZBYCAgsPEA8WAh4LXyFEYXRhQm91bm RnZBAVAQRhYWFhFQEBMRQrAwFn ZGRkfXCqCy7IiEyl+lHROA5vfn1oYNq3t8MhnzZlEt DTPXA=|80|hiddenField |__EVENTVALIDATION|/wEWBALF19m3BALEsPqoBgKct7iSDAKjn6ufDly+66xedWvvj /fQnBQjlOsBTWNL9UuTy1VDVvKmkOGe |15|asyncPostBackControlIDs|| btnSave,btnSave|0|postBackControlIDs|||26|updatePanelIDs||  tUpdatePanel1,UpdatePanel1|0| childUpdatePanelIDs|||25| panelsToRefreshIDs|| UpdatePanel1,UpdatePanel1|2| asyncPostBackTimeout||90|16|formAction||UpdatePanel.aspx| 

A 1024 character response, including all controls that are in the update panel.

For test, I’ve implemented the same functionality with an Ajax call with jQuery. Check the response:

{"d":{"__type":"Webform.Object.Ticket","TickedId":3,"TicketName":"aaaa"}} 

A 73 character response, only containing the values we needed: the id of the value entered and the value entered

As you can see in this (very) simple example, the difference between the implementing a async callback with the toolkit and jQuery gives us an advantage of 951 characters. You can imagine what the difference would be in a page with let’s say 50 controls on their.

For simple Ajax functionality the toolkit is an excellent choice for a beginning developer. A developer who doesn’t want to know what’s going on in the background. He updates a drop down list and the correct value appears in the textbox underneath. What triggered it and how it’s used, he (or she) doesn’t care. They still maintain and renew the toolkit as you can see on Stephen Walter’s blog.

I do care. I’m one of those guys who want to know what’s going on in the background, so I can extend functionality, create my own controls and so one. And I do care about the traffic I generate.