You have to see this amazing performance….

Took me a while to understand but she is recording herself and the replaying it while continue to add voices, rhythm etc… and finally produces a wonderful sound and a quite catchy song.

One thing that really surprised me was that when tests fail in my build script the build script happily ran along after that!? Isn’t one of the ideas with Continuous Integration that the build script only will emit working versions?

Well in this article Aaron Hallberg (Swedish relatives?) shows us how to handle this. What he basically does is to check the status of the test-run and then update the status of the complete build.

I extended his code to use the test-status-flag on the tasks that follow the test, i.e. the publish task. This task deploys the executables to my test server, update the documentation etc.

So now I’ve got my desired result which is that failing test leads to failing the whole build.

When you have a testing framework in place you of course want to test it each build. But of course, you also want to know much code your tests are covering, a.k.a. CodeCoverage.

It is in a way quite easy to accomplish in your build script. The only catch is the complete lack of documentation. Benjamin Day led the way (he he - it rhymes).

What you need to do is two things:

1. First configure the tests to run. This is easily done by adhering to good naming standards (all test assemblies ending with tests.dll). Then you can add create an item group called TestContainer, like below. This will run through all your tests.

   <ItemGroup> <TestContainer Include="$(OutDir)\%2aTests.dll" /> </ItemGroup> 

2. Secondly (and here is the secret) you need to specify that the tests are to be run under code coverage....

This is a great song from a great old show - Macken.

I just love his way of taking out the worst in life.

One thing that have irritated me with the code coverage tool in Visual Studio 2005-8 is that you can’t exclude things from the code coverage result.

For example - in VB.NET the MySettings-class is generated for each assembly that is created. The code in it is automatically generated and you shouldn’t change it (it even says so in a comment from the tool…). However it is counted when the code coverage result is calculated.

The same thing applies for proxy-classes to web services for example.

This gives an incorrect picture on how much of your code that is tested. But now I’ve found an attribute that might help:

<debuggernonusercode()>

This attribute tells the compiler that the code that is decorated with the attribute is not written by the user and should not be step into during debugging sessions. So when the testing tool encounters the attribute it will simply not...

After reading the great patterns book i was talking about earlier i soon ran into a problem that at first seemed pattern-like but was not after some examination.

The case that we are integrating against a back end AS400 system, via Microsoft Transaction Integrator. The “problem” is that the department that is responsible for the integration is creating a web service for each program (“method”) we are accessing. I brushed on this in an earlier post about naming those web services.

A lot of the stuff we are sending back and forth are equivalent but not the same, for example an header that all methods expect. They are equivalent but not the same since they are located in different namespaces, since the header are generated once for each web service.

So what we wanted was to create an interface that all the headers could implement. To create...

Everybody who knows me knows my feelings on doing complex WebGUI. In short i think it is stupid - and also i have never met a developer yet who can show me a robust and simple way of managing the events of a web page. It is always, always fixes like; hidden fields, setting stuff in prerender etc etc.

So my personal opinion of this is - don’t do it! Web is for simple stuff. The moment a request about a tab control or thing posting back is being said I would raise a big warning flag. You’ll get into trouble. (Of course I can only speak for ASP.NET, and love to be convinced otherwise).

Luckily there is help on the way.

• You could either stick with simple HTML and use [ASP.NET MVC](http://weblogs.asp.net/scottgu/archive/2007/10/14/asp-net-mvc-framework.aspx) which removes all the crazy event-handling from the ASP.NET WebForm model. Basically...

During the last three weeks or so I’ve been reading a great book, Head First Design Patterns. Now i have finished it!

As i said earlier; This is the best, funniest and most informative IT-related book I’ve ever read - and that is quite a few if you count my years at the university.

I have tried to wrap my brain around Design Patterns many times but failed, often due to the very academical language that is used when you speak about such abstract thing. I cannot count the number of times i got an “aaah”, “so it was this simple?” or “finally!”-feeling while reading this book.

Thank you!

Now if the Head First team only could put out some .NET stuff and I’ll be buying lots and lots of those.

I can’t withhold this sweet picture. It is from the Youth Brass Band Championships in Great Britain held last weekend. Note that she’s not even holding a normal cornet, it is smaller than a regular cornet. But she is still playing a singing E. What a concentrated face!

I’m back - yesterday was a day with speed since we had to prepare the Scrum demo we’re going to hold today. And as frosting on the cake i caught a cold (kind of).

Well, nothing to do about that - the sprint demo today will be running smoothly even if i am a bit slower than usual.

The night at the shelter was quite quiet (:)) for me. I got to go to bed at around 2345 and woke up 7 hours later without a single disturbance, part from a soar throat. A very small contribution from me.

[UPDATED] The demo was a success! The product owner said that she wanted to do “the wave”, which was quite a good reception, don’t you think.