UPDATE: AH FOOEY!  I feel like a goof:).  Turns out my workaround DOESN’T WORK!!!  If you look at the logic from the original post, it makes sense that it wouldn’t work because “var map = new VEMap();” is before “var map = null;” and therefore you get Intellisense for null.  I swear it worked (giving me Intellisense for page level map variable in multiple functions) the first time I tested it.  I think I may have been a victim of the JavaScript Intellisense cache not having updated :(.  Nonetheless, I was so excited about it working, I jumped the gun with my original post.  Yes, I realize the last picture in the original post didn’t actually convey what I was trying to accomplish.  I showed the wrong scenario in my haste to get a blog post up.  Maybe if I would have shown the right scenario, I would have recognized the err of my ways:).  This entire post has been rewritten with my "plan B" workaround.

So I have been playing around with other approaches to the workaround described here.  The first thing I tried was:

On the surface, this seemed like a reasonable workaround.  However, there were still some scenarios where JavaScript errors would pop up.  For example, switching to 3D mode would cause the map control to raise an error:

A similar error would happen when you close the browser.  One could argue this annoyance is a small price to pay for JavaScript Intellisense for the map control:).  However, it is still very annoying.  Good news!  There’s a better workaround.  After fiddling with a bunch of ideas, I stumbled upon the following approach:

Everything JUST WORKS!  You get JavaScript Intellisense at design time and the map is null at run time until it is actually instantiated in pageLoad().  This makes for a nicer experience since you don’t have to put up with the annoying error messages or constantly comment/uncomment code to get Intellisense where you need it.  Of course, in production, you should replace the conditional logic with var map = null for your page level variable.

Interestingly enough, this approach works well in another scenario:

In this scenario you won’t get Intellisense for x above the if statement because Visual Studio does not perform return type analysis.  However, using the same trick, you can see that you get Intellisense after the if statement.  This makes life better in a number of scenarios.  However, it does litter your code a bit.  Use it at your discretion:).  One could write a little utility that scans .js files and removes these entries based on a regular expression.  Maybe we’ll look at writing something such a utility for our next release.

None of this is in the current release of the Virtual Earth JavaScript Intellisense Helper, nor is it in the updated code I recently blogged about at Updated Virtual Earth Map Control demo code.  However, I will get both updated in the coming days.  For the sample code, I will update the original blog post when I publish the updated code.  I will make sure we release a new version of the Intellisense helper with these changes soon.

A bug has been discovered when using the Virtual Earth JavaScript Intellisense Helper where JavaScript Intellisense fails to surface in a very common scenario.  The scenario is when you have declared a page level map variable.  You will get Intellisense in the JavaScript function where you instantiate the map variable.  However, you will not in other functions:

In the example above, you would get Intellisense for map in GetMap(), but not in Find().  The problem is that Visual Studio 2008 JavaScript Intellisense engine does not know that GetMap() gets called before Find().  Therefore, there is no way to know that the map variable used inside the Find() function is an instantiation of the VEMap class.  We are investigating the best guidance on how to handle this scenario in Visual Studio 2008. 

On the surface, the workaround seems simple.  You just need to make sure the map variable is a page level instantiation like so:

Of course, this poses a problem at runtime because VEMap doesn’t have access to the ‘myMap’ div yet.  That’s the whole reason people put their map init code in window.onload, body.onload, or pageLoad() for ASP.NET AJAX based solutions.  So what to do?  For now, here is what I am doing and what I have checked-in to source control for our upcoming release that will be in sync with the Virtual Earth Map Control 6.1:

The good news is that you really don’t have to comment/uncomment while you are developing.  The code above will "work" as is.  I say "work" because you will get an annoying script error for the "var map = new VEMap();" line of code, but the rest of your page will function as expected.  You just want to make sure you fix things for production:).  Although this is a little bit annoying, I think it is small price to pay for getting JavaScript Intellisense for the Virtual Earth Map Control.