It seems that mobile and embedded computing is finally being recognized as an actual career path.  Purdue University has a class and a Mobile Computing Lab.  Check out the projects students have done so far.  Very promising indeed.

Here's a quick look at the object model so far (the Aircard555 is where the bulk of it resides).  The documentation is a bit sparse yet for descriptions, and there are methods that I've not yet added - largely for voice calls - but it's coming along well.

Anyone have interest in the voice end of things anyway?

I've got a basic code line counter that I like to run on my projects occasionally.  Just hit the SDF source tre and this is what I got:

code lines=51090 comment lines=32619 files=473 projects=12

A typical rate of development for an experienced developer is probably 100 lines of debugged code per man-week.  That means that the SDF has roughly 9.8 man-years of development invested.  Now consider you can get it for free, including source.  Even if you were superhuman and could write 1000 lines of code per week, it would still take you a year to replicate it.

OpenNETCF has a budget of about zero.  We've paid for absolutely no marketing whatsoever, so any “market awareness” is purely by word of mouth and search engine results. 

So tonight I got to wondering, many compaies pay thousands for search engine ranking and hire teams to better themselves.  I wonder how we're faring. Here are the results from some quick Google tests.  I list the search term and the placement we got:

Compact Framework : #2
Compact Framework Code : #1 (beating Microsoft even)
Compact Framework Samples : #15 (surprisingly low based on the last 2)
.NET Mobile : Not in first 50 (though DePaul University has a CF course, neat!)
Managed Code Mobile : #49
Pocket PC VB : #12
Pocket PC C# : #28
Pocket PC Managed : not in top 50
CE Managed Code : #7
Compact Framework Consulting : #4

So it seems that we've got pretty good placement for what I'd think are common search terms. I'd love to see higher ranking on some, but it seems we're getting more than we pay for.

As is typical, I started what I thought would be a simple project - providing managed access to a Sierra Wireless AirCard 555 CDMA modem from managed code. I did a basic eVB wrapper in COM for it about a year ago and figured it was going to be cake. Little did I know...

The first problem is that it needed a native callback.  Second, that callback takes a union parameter, which has some 50+ members.  Third, the few pieces of info retrieved without the callback would be used to update a UI and had long retrieval times, so retrieval must be on a Thread, yet most users won't want to deal with using Invoke for everything.

So the short story is that it took a lot longer to develop than originally planned.  I guess that's how development typically works.

It's only been through light testing, and I still need to add a lot of XML comments and run it through NDOC, but it's pretty much complete.  This is going to be a for-sale product through OpenNETCF Consulting, the “commercial arm“ of the OpenNETCF project (we've got to pay the bills somehow).  If you're interested, email us.

Here's a snapshot of my test harness, which gives you an idea of the properties exposed.

Anyone taken the time to test out this sample code in the CF and see how it compares to the full framework?  I was surprised at the desktop results, and if it's true for the CF then I'd like to be aware of it.  Does the JITter use registers for locals when available?  If so, how many registers (the FFW has 64 of them)?  Is it processor dependent?