Comments working again

Thanks to Al for pointing out that my comments section was failing with JavaScript errors. Apparently the blowery http compressor that this site is using was messing up the embedded ASP.NET JavaScripts so they return empty script files. Fortunately a workaround exists for this.

So if you meant to comment about one of the screw-ups in my recent blogposts, here's your chance.

Huge wind turbines in Virtual Earth

Since today is Blog Action Day and the topic is the environment, I thought I would share this amazing birds eye view of one of the 2mW windmills in the offshore windfarm "Middelgrunden" just off the coast of Copenhagen that was published last week. I've knew these windmills were BIG, but looking from the peer it's hard to tell how big they really are. The boat next to it really gives a sense of its size. I can't even begin to imagine how big the 5mW mills they make these days are... 


There are 20 of these mills, all together producing 3% of Copenhagens total electricity consumption. Around 20% of Denmarks electricity comes from windmills.

Warning... bad joke coming up: Now you know why I moved to Sunny California - The weather really "blows" in Denmark :-)

 

New Virtual Earth release

I just visited http://maps.live.com and it looks like they finally put the new release out there. A revamped design and Birds Eye View imagery inside the 3D viewer. You can now also create animations using your collections.

Nyhavn, Copenhagen - A place I missed a lot this summer

Another new feature is the new "Create model" button in the collection editor when you are in 3D mode. Using it will prompt you to download "3dvia" enabling you to model your own buildings and insert them into VE. It looks somewhat similar to SketchUp used for the same thing in Google Earth. A neat feature is that you get the map as a background to assist you in your modelling, and comes with a bunch of default textures as well:

Accurate distance calculations in UTM projections

Recently a friend of mine asked me what projection would be the best to create point-buffer circles. All map projections adds distortion so you rarely end up with a circle. If you’re lucky, you’ll have an ellipse but often they get even more complex than this. In many cases a “normal” circle is sufficient for accuracy, but still it got me thinking how I would do this 100% accurate. There are many approaches to this, fx. creating a new projection for each point, do the math in spherical coordinate systems etc, but none of them are completely trivial.

The UTM projection is one of the most commonly used projections. It’s fairly accurate to measure from point to point within small distances and close to the meridian, so this post is based on that. Most of the math here can also be applies to Mercator, where the scale reductions are applied to Y instead of X.

The Mercator projection is created by projecting the globe onto a cylinder whose center follows the Earth’s rotation axis. The Transverse Mercator is basically a "tilted" cylinder parallel to equator. Along the line that the cylinder "touches" the surface the distortion is 0, and increasing away from it in the east/west direction. There is no distortion in the North/South direction. The Universal Transverse Mercator is slightly smaller than the globe, so it will intersect the surface two places. This also ensures that the distortion between and around these two lines are minimal. Along the meridian (the center of the two lines) the distortion for the UTM projection is 0.9996, meaning that this is the amount you have to reduce an infinitely small line in the east/west direction (For Mercator this is normally 1 along Equator). As you move away from the meridian, the scale factor increases up toward infinity. Normally UTM projections are used close to the meridian, and every time you get too far east/west, you would switch to use a new UTM projection. That’s why The Earth is divided into 60 UTM ‘zones’, one for each 6 degrees. Often for practical reasons you might want to use a UTM zone even though it’s outside its defined usage area. Here you have to be specifically careful when measuring distances. For example, Denmark is covered by zone 32 and 33, but it’s common to only use zone 32 which means that distance calculations at the east end can be very inaccurate without applying a scale reduction.

Warning: The following contains a lot of math. You can skip to the bottom and download a C# class that does all the calculations for you.

The scale factor SR(x) for any given point at distance 'x' from the meridian is given by:

Since this is the scale reduction at a point, we need to sum up all of these values along a line to find the correct distance, thus the east/west distance from e1 to e0 (expressed in distance from the meridian) becomes an integral expression:

To this you need to add the north/south distance using the well-known Euclidean distance formula.

For atypical UTM using WGS84 ellipsoid and a 500000m false easting it becomes:

If you haven’t refreshed your integral math lately, or just want to write this in pure code, one can also write the expression as:

Where m is the scalefactor at the meridian, r is the earth radius and e1 and e0 have the false easting subtracted.

Often we can do with an approximate value by using the scale reduction between the start and end point. This should only be used on distances <1km in easting because it gets very inaccurate for greater distances.

Simpson’s Rule

One can also use Simpson’s rule for calculating the integral solution. My tests showed it’s a very accurate approximation for this type of integral usually giving millimeter accuracy. It’s defined as:

I performed a few performance tests for 10 million calculations using all four methods for finding a distance in a UTM projection. Below is the processing time for each method:

Euclidean 4.38 s
Average / Linear 4.49 s
Simpsons 6.94 s
Integral 9.26 s

Bottom line of this: If you want performance and distance is small use the averaging method, if you want accuracy, go with the Integral method, or if you want both, use the Simpsons approach.

Distance calculation - example

start point : { E=400,000 ; N=6,100,000 }

end point : { E=600,000 ; N=6,250,000 }

  • Distance using Simpsons rule: 249,942.5569m (error: 0.0003m)
  • Distance with approximate scale reduction: 249936.0046m (error: 6.54m)
  • Distance without scale reduction: 250,000m (error=57.46m)
Where is scale reduction=1?

Solving SR(x) =1 gives us the distance from the meridian where the scale reduction is exactly 1:

Area calculation To calculate the area of a polygon, apply the scale reduction to the line segments and then calculate the area as you normally would.Creating perfect circles

To create a circle with a fixed radius taking scale reduction into account will give you an irregular shape in the projection. You can use the approximate method (using center as scale factor) which will give you an ellipse since the scale factor for x is constant in this case. The formula for approximate circle becomes r2=(SR*x)2+y2 where SR is the scale reduction at the center of the circle.

For "accurate" circles SR becomes dependent on X. To find this solution you need to isolate e1 from the above equations to find the scale corrected distance. My math is too rusty to figure that one out yet, but if you know, feel free to post it in the comments.

You can download a C# class that performs all four types of distance calculations here.

Visual Studio 2008 JavaScript Intellisense - Take Two

...Not for namespaces either.

We all know that global variables in JavaScript API’s are bad (although they are usually OK in application-specific contexts). So therefore the solution is to create a namespace and put all your objects inside that. That way you only end up with one global, the name of the “root” namespace. You should choose a root name that probably isn’t used by anyone else, like your company name or similar.

With the Microsoft AJAX Clientside Library, Microsoft gave us registerNamespace for this thing, like for instance:

Type.registerNamespace('myCompany.Utility.Math');
myCompany.Utility.Math.Calculator = function() { }
myCompany.Utility.Math.Calculator.prototype = {
 add : function(a,b) { return a+b; }
}
myCompany.Utility.Math.Calculator.registerClass('myCompany.Utility.Math.Calculator');

Basically what Type.registerNamespace does is creating a set of objects:

myCompany = {};
myCompany.Utility = {};
myCompany.Utility.Math = {};

There's is just one problem to this. Visual Studio 2008’s intellisense is not evaluating the result of Type.registerNamespace, and thus, never realizes that those three objects exist. The result is that you don’t get any intellisense on your namespaced objects. Bummer! The workaround is either to create the namespaces yourself (like above) before calling Type.registerNamespace, or create aliases for the classes (which would re-introduce the globals we were trying to get rid of in the first place).

Of course I considered this to be a bug, and as I mentioned at the end of my previous post there was a problem with using namespaces and JavaScript Intellisense in Visual Studio 2008, that I had submitted a bug report for.

I was surprised how quickly they investigated and escalated the issue, but my biggest surprise was the result, which is pretty much this:

We are not going to support the pattern we told you to use until version 20xx.

OK, maybe a bit over-dramatized but still... It's funny because they are actually able to give me full intellisense for the namespaced objects in the MS AJAX Sys.* namespace (but they probably hardcoded that in there ;-)

With this and the fact that javascript files included by webcontrols are not intellisensed either, I’ve come to the conclusion that the hyped and long awaited JavaScript Intellisense in Orcas is pretty much useless. What do you think?

Visual Studio 2008 JavaScript Intellisense…

 …well maybe not for webcontrol users.

One of the most spoken-of features of Visual Studio 2008/Orcas is the new JavaScript Intellisense that helps the client-side developer. I’m very excited about this new feature, and put a lot of effort in documenting my JavaScript with code summaries so that I will get the full benefit when VS2008 ships.

I have been struggling to get this to work with some custom controls that emit JavaScript into the page. The JavaScript is fully documented and each file header references other assembly-embedded JavaScript files, just like the documentation tells me to do it. The problem was that I was getting no Intellisense at all for these scripts, unless I manually inserted them into the page. I also made sure that these commented JavaScript files (I also have compressed/stripped scripts embedded) of these scripts was actually registered during DesignMode.

I was reading a blog-post the other day stating that Intellisense is provided for scripts that are included in the page either as a ScriptReference with the ScriptManager, or as a normal HTML script tag in the page. My controls did just that during the PreRender events.  I emailed one of the bloggers about this, and guess what… this is not supported. If you drag a webcontrol onto the page that emits JavaScripts to the page, you are NOT getting Intellisense for those JavaScripts. So much for building a fully documented JavaScript API.

The reason is performance and security. For once they don’t want to execute user code on the client by rendering the control. But hey wait… isn’t that what is already happening in design time? The controls are rendered with specific design time layout, thus executing user code. OK VS is not doing it in the code editor but in the designer, but what's the difference from a security standpoint? Or could we at least get to allow certain assemblies we trust to do this? With regards to performance, I would think that either you want Jscript Intellisense or not, and if you want performance, you would just go and disable the Jscript Intellisense. They also seem to have nailed this one pretty well with JScript Intellisense being generated in the background while you're coding.

So the proposed workaround is to manually add a ScriptReference to each script using a ScriptManager. There are just two problems to that:

  • You rarely know the full name of the embedded script, and using Reflector to find is often a violation to the EULA’s, and furthermore the name or scripts might change with each release/service pack. -not to mention that it's not very user friendly.
  • Secondly, who says I even have a ScriptManager or want one on the page? Partial Postbacks have longer page life cycles and transmit a lot more data than callbacks, so I might just want to stick with the good old lightweight callback model, or I might not even need any AJAX functionality at all.

The funny thing is that the reply I got from Microsoft started out with saying that they get this question a lot, so there obviously is a demand for it.

I would have to wait 3+ more years for the next Visual Studio release, to harvest the benefit of all my code summaries... '-(

I actually found this other bug trying to get this thing above to work, and at least that is getting some attention. So before you start playing with JString Intellisense in beta2, don't put your classes inside namespaces.