There is following code on page 79Is it a correct sequence to detect available version of MSXML parser?
I believe that in case Microsoft.XMLHTTP ProgID is not registered there can hardly be Msxml2.XMLHTTP on the system. I think the sequence should be in reverse order. Am I wrong?
And probably the alternative approach presented on the same page that checks support for ActiveXObject first is not practical because it forces IE7 to use ActiveXObject instead of its XMLHttpRequest object (because it supports both variants).

__________________
Alexey Gorkov

In the book "Professional Ajax" (also Wrox) they create the XMLHttpRequest object like this:

function createXMLHttp() {
    var aVersions = [ "MSXML2.XMLHttp.5.0",
        "MSXML2.XMLHttp.4.0","MSXML2.XMLHttp.3.0",
        "MSXML2.XMLHttp","Microsoft.XMLHttp"
    ];

    for (var i = 0; i < aVersions.length; i++) {
        try {
            var oXmlHttp = new ActiveXObject(aVersions[i]);
            return oXmlHttp;
        } catch (oError) {
            //Do nothing
        }
    }
    throw new Error("MSXML is not installed.");
}

I’m not sure what the best approach is, though. I use the following:

var globalXMLHttpRequestObj = null;
function createXMLHttpRequestObject()
{
    if (globalXMLHttpRequestObj != null)
        return globalXMLHttpRequestObj;

    if (window.XMLHttpRequest)
        return globalXMLHttpRequestObj = new XMLHttpRequest();
    else
    {
        var aVersions = [ "MSXML2.XMLHttp.6.0", "MSXML2.XMLHttp.5.0",
            "MSXML2.XMLHttp.4.0", "MSXML2.XMLHttp.3.0", "MSXML2.XMLHttp",
            "Microsoft.XMLHttp" ];

        for (var i = 0; i < aVersions.length; i++)
        {
            try
            {
                globalXMLHttpRequestObj = new ActiveXObject(aVersions[i]);
                return globalXMLHttpRequestObj;
            }
            catch (oError)
            {
                //Do nothing
            }
        }
        throw new Error("Can't create XMLHttpRequest object.");
    }
}

DanJ,
Thanks for your reply. Well, I've read Professional Ajax (several times actually :) and I know the method presented there.
What I'd asked here about wasn't actually a question, rather a statement and I just wanted to hear from authors regarding this subject. To put it more precisely I think that both code snippets in “Try it out� section on page 79 are erroneous because of reasons presented earlier. I could be wrong of course. My apologies if it is the case.
And two additional remarks: probably it is not a good idea to name the method CreateXmlHttpRequestObject(). I don’t think anyone would use such a verbose name in real project. And it also probably not quite accurate to place such method in a global namespace (i.e. declare it globally) so despite of the phrase “The preceding example, which created the XmlHttpRequest object in a cross-browser fashion will be
encapsulated in a function called CreateXmlHttpRequestObject() and included in a script file named
CommonAJAXLibrary.js.� this method is not actually encapsulated.

Why do you consider this not to be encapsulated?

coderevo,

Function is certainly a type of encapsulation because it combines a set of instructions but it is probably the most basic level of encapsulation.
What I meant is (and I think it was clearly defined in the same sentence) that it would be better to use encapsulation on an object level, combining related functionality and data pertaining to XMLHttp request object and not to define such a function globally polluting the global namespace. Sorry if I am wrong.
But isn't it better to present real world examples? Have you seen such kind of standalone function used in any of existing JavaScript/Ajax frameworks/libraries?