Using HttpWatch and WatiN 2.1

calendarJune 25, 2012 in Automation , C# , HttpWatch , Internet Explorer

In a previous post we described how to interact with a web page using WatiN 1.3 while recording HTTP/HTTPS traffic in HttpWatch. It was a popular post, used by many customers to build automated web page tests that used HttpWatch to provide performance metrics, check for HTTP level errors and to look for opportunities to improve performance.

Since then, WatiN 2.1 has been released providing significant improvements and the ability to interact with Firefox 3.6 as well as Internet Explorer. Unfortunately, a change in the WatiN assembly caused a type conflict with HttpWatch over the definition of IE’s IWebBrowser2 type. This could give rise to the compilation errors in a Visual Studio C# project:

error CS1758: Cannot embed interop type ‘SHDocVw.CommandStateChangeConstants’ found in both assembly ..\WatiN\bin\net20\Interop.SHDocVw.dll’ and ‘…\obj\Debug\Interop.SHDocVw.dll’. Consider setting the ‘Embed Interop Types’ property to false.

If you had marked the project to build against .Net 2.0 there isn’t even an option to set ‘Embed Interop Types’ to false.

The way to avoid this error is to delete the ‘Interop.SHDocVW’ reference that you may have added from the WatiN bin directory:

In HttpWatch 8.3.19 we’ve added a WatIN sample program and documentation to provide a starting point for using WatiN 2.1.

The sample program shows you how to fill out a simple web form:

and retrieve values from the resulting page:

There are two ways to get HttpWatch and WatiN to work on the same instance of IE. The first is to create the new instance with WatiN and then attach HttpWatch:

// Attach HttpWatch to an instance of IE created through WatiN
WatiN.Core.IE watinBrowser = new WatiN.Core.IE();
HttpWatch.Plugin plugin = control.IE.Attach((SHDocVw.IWebBrowser2)watinBrowser.InternetExplorer);

Or you can create the instance in HttpWatch and attach WatiN:

// Attach WatiN to an instance of IE created through HttpWatch
HttpWatch.Plugin plugin = control.IE.New();
WatiN.Core.IE watinBrowser = new WatiN.Core.IE(plugin.Container);

For more information about using WatiN with HttpWatch please take a look at the sample program that is installed with HttpWatch Basic and Professional Editions:

http://apihelp.httpwatch.com/#WatiN Form Fill Sample.html

 

 

 

 

 

A Guide to Automating HttpWatch with PHP

calendarMarch 11, 2011 in Automation , HttpWatch

Stoyan Stefanov, the creator of smush.it, architect of YSlow 2.0 and engineer at Facebook, has written an excellent three part guide to controlling HttpWatch from PHP:

He’s even published a class that wraps the HttpWatch API and makes it easier to use from PHP.

In the rest of this blog post we wanted to follow up on a few points mentioned in Stoyan’s blog posts. These items don’t just apply to PHP. You may find them useful when automating HttpWatch using other languages such as C# or Ruby.

Hiding the IE window

The ‘A better experience in IE’ section of Automating HttpWatch with PHP shows how you can hide the IE browser window during tests by separately creating IE and attaching HttpWatch to it.

It’s also possible to do this using the Container property without having to separately create and attach to IE:

$controller = new COM("HttpWatch.Controller");
$plugin = $controller->IE->New();
$browser = $plugin->Container; // Only works with IE
$browser->Visible = false;

If you do decide to do this in order to stop windows popping up during a test please bear these points in mind:

  1. Orphaned instances of iexplore.exe will be left running in the background if your script ever terminates before calling CloseBrowser.
  2. In IE 8 and earlier, HttpWatch will not record a Render Start event because the hidden IE window does not get updated by Windows. However, the event will be recorded in Firefox 3.5+ and IE 9.
  3. Your performance measurements may not directly match user experience. It’s possible that current or future browsers may avoid certain rendering and processing actions if they detect that the output will not be in a visible window. For that reason, we recommend running tests in visible browser windows on a normal interactive desktop either on a physical machine or VM. Also, viewing a browser test through Remote Desktop is likely to have a significant negative impact on performance as the graphics and text making up the page have to be transferred over the network.

Opening the HttpWatch Plugin Window

It’s often handy to open the HttpWatch window in the browser when you are developing an automation script so that you can check that it is working as expected.

The OpenWindow method allows you to do this and specify whether you want the HttpWatch window docked or undocked. For example, here is the PHP code to open HttpWatch as an embedded window in the browser:

...
// Open docked HttpWatch window in browser
$plugin->OpenWindow(false);
...

Handling Differences Between HttpWatch Basic and Professional Editions

In part 3, Stoyan mentions using try-catch to handle the errors that occur when attempting to access data that is restricted in HttpWatch Basic Edition. While this is a valid approach, there is a risk of hiding other errors that might be occurring that are not due to the restrictions in HttpWatch Basic edition.

There are a couple of properties in the HttpWatch automation interface that help you handle the differences. The first is the IsBasicEdition property on the Controller class.

For example, here’s a high level test in PHP:

$controller = new COM("HttpWatch.Controller");
if ( $controller->IsBasicEdition )
{
    echo "\nThis test requires HttpWatch Professional Edition";
}

At a lower level, you can also check each request to see if it has been restricted using the IsRestrictedURL property:

...
if ( $entry->IsRestrictedURL)
{
    // Goes here in HttpWatch Basic Edition for URLs outside Alexa Top 20
    echo "\nSome of the properties for this request are restricted";
}
else
{
    // Goes here in HttpWatch Basic Edition for URLs in Alexa Top 20
    // or in HttpWatch Professional Edition for any URL
    echo "\nAll the properties of this request are available";
}
...

How to check HTTP Compression with HttpWatch

calendarJuly 10, 2009 in Automation , C# , HttpWatch , Optimization

HTTP compression is one of the easiest and most effective ways to improve the performance of a web site. A browser indicates that it supports compression with the Accept-Encoding request header and the server
indicates the compression type in the Content-Encoding response header.

This screenshot from the Stream tab of HttpWatch shows these headers and the compressed content being returned from the server:

Compressed Page

Here’s another screenshot of a page that is not compressed:

Page with no compression

The browser still indicated that it accepted gzip and deflate compression, but the server ignored this and returned uncompressed HTML with no Content-Encoding header.

The easiest way to check the amount of compression achieved is to use the Content tab in HttpWatch to view a ‘200 OK’ response from the server:

Compressed Content

Don’t try checking for compression on other HTTP status codes. For example, a ‘304 Not Modified’ response will never have any compression saving because no content is returned across the network from the web server. The browser just loads the content fom the cache as shown below:

Content Tab with 304

So, if you want to see if compression is enabled on a page, you’ll either need to force a refresh or clear the browser cache to make sure that the content is returned from the server. The HttpWatch Automation API lets you automate these steps. Here’s an example using C# that reports how many bytes were saved by compressing a page’s HTML:

// Set a reference to the HttpWatch COM library
// to start using the HttpWatch namespace
//
// This code requires HttpWatch version 6.x
//
 
using HttpWatch;
 
namespace CompressionCheck
{
    class Program
    {
        static void Main(string[] args)
        {
            string url = "http://www.httpwatch.com";
            Controller controller = new Controller();
 
            // Create an instance of IE (For Firefox use
            // controller.Firefox.New("") )
            Plugin plugin = controller.IE.New();
 
            // Clear out all existing cache entries
            plugin.ClearCache();
 
            plugin.Record();
            plugin.GotoURL(url);
 
            // Wait for the page to download
            controller.Wait(plugin, -1);
 
            plugin.Stop();
 
            // Find the first HTTP/HTTPS request for the page's HTML
            Entry firstRequest = plugin.Log.Pages[0].Entries[0];
 
            int bytesSaved = 0;
            if (firstRequest.Content.IsCompressed)
            {
                bytesSaved = firstRequest.Content.Size
                                   - firstRequest.Content.CompressedSize;
            }
 
            System.Console.WriteLine("Compression of '" +
                firstRequest.URL + "' saved " + bytesSaved + " bytes");
 
            plugin.CloseBrowser();
        }
    }
}

Tip: If you access a web site through a proxy you may not see the effect of compression. This is because some proxies strip out the Accept-Encoding header so that they don’t have to process compressed content. Tony Gentilcore’s excellent ‘Beyond Gzipping’ talk at Velocity 2009 described how 15% of visitors to your site will not receive compression due to problems like this. A simple way to effectively bypass proxy filtering for testing purposes is to use HTTPS if it is available. For example, try https://www.httpwatch.com if you don’t see compression on http://www.httpwatch.com.

Ready to get started? TRY FOR FREE Buy Now