Stupid Code Tricks

Microsoft Robotics Studio - What?

Malcolm Anderson — Mon, 02 Jun 2008 23:03:53 GMT

Once more, Microsoft is handing out free tools, and doing cool stuff, and putting the fun back into programming.

This is not a game, it's a full fledged robotics language, and simulator. Ever wanted to show how easy it would be to have robot navigate a maze or program a rescue robot, or a soccor droid? Well now you can. Currently the only one available is the maze navigation (complete with traps), but more will be released over the course of the year.

It looks like you can upload your program, and the winners will have their programs loaded on to real bots that will go head to head at the PDC. The be all, end all winner, gets a robot of their own.

Click here to download the MS Robotics Developer Studio 2008 (CTP April)

Click here to go to the Microsoft Robotics Developer Center

I've really got nothing more than, "This is cool"

Good Bye TDD

Malcolm Anderson — Fri, 15 Jun 2007 19:44:52 GMT

Have you seen the new buzz phrase for TDD? Pretty much, they just changed the name from Test Driven Development, to Behavior Driven Development.

It's an NLP experiment that I'm looking forward to seeing fleshed out.

Long story short, the problem with TDD is the word "Test"... *WAY* too much baggage, and charge on that one single word.

Consider:

"Testing your code is a good if you're a new programmer"

"Our department does testing at the end of a project ... if there's time"

"I'm a developer, I don't get paid to do tests"

"As a tester, I believe that most developers couldn't write a good test if their lives depended on it"

Here's the two links that I think are the most concise summations of BDD.

http://behaviour-driven.org/Introduction

http://behaviour-driven.org/GettingTheWordsRight

I don't think its ready for prime time yet, and the .net version is certainly no replacement for NUnit, *but* I think that this minor tweak can solve a lot of the communication issues that hinder adoption of TDD in the enterprise.

Alex, I'll take "Google Key Words" for 1000

Malcolm Anderson — Thu, 14 Jun 2007 18:23:32 GMT

We've got a custom control that consists of 2 radio buttons, a "normal button", and a text box, all on a single line.
Internal to the control, I have all 4 of member controls lined up so that all the text is on one line.
So far, so good. Except for one thing.
When I put my control on my form, I can not get the text on my form to line up with the text on my label, and I end up manually futzing it into something that resembles "it's place". Where are my alignment functions?

I guess it's off to google.

My first google search was
forms align text .net

This didn't get me anything useful, but I did get some stuff from asp.net.

Let's try requiring both "forms" and "align"
+forms +align text .net

That didn't help either, people with questions, but no answers

Next I tried
how do I align my textbox with my custom control

And got an interesting hit down at #6
"SnapLines not honored during resize." and the body says
"I have a custom control that contains a textbox. It has two ... the name string (based on the MeasureString length) and my Right snap ... "
let's go look at it.

http://www.devnewsgroups.net/group/microsoft.public.dotnet.framework.windowsforms/topic62981.aspx
well, would you look at that ... snaplines, is that a new keyword? I go and check out the post.
ravi.s.desai@gmail.com seems to be having the same problem I am, I have a custom control that is just a collection of windows controls, but I can't align them.
As he puts it,
"Actually, to simplify the above, it appears that when you are resizing
a control, all snaplines defined in the designer are completely
ignored, and the designer simply creates a snap-line for the edge of
the control that you are dragging, and uses that exclusively. Is this
by design? It certainly seems to limit the value of snaplines for
layout."

At this point, I stop, and write up the progress of my research up to this point.

Hopefully, the new keyword (snaplines)
a) actually is a keyword
b) produces fruit

Into google goes the phrase
snaplines custom control

The first choice is
"Using SnapLines from child controls of custom user control? - MSDN ..."

Which seems to indicate that while what I want to do, *can* be done, it's not simple, easy, or intuitive.

Now I have to ask the question, is this worth my time.

Answer, not in the short run, but definitely in the long run.

I'll be back...

Non Click-Once Version Pain

Malcolm Anderson — Fri, 18 May 2007 19:34:46 GMT

Aaaaaaaaaaaaaarrrrrg

Apparently with the release of VS2005, Microsoft has decided that if you are not using click once, you obviously don't need to use version numbers on your app.

I'm trying to set up a schema of 1.2.3.4 where

1 = major

2 = minor

3 = sprint number

4 = ccnet build number

but no matter what I do, when I right click on MY_APP.exe, and check properties, I get "1.0.0.0" as opposed to my expected "0.5.5.298"

On the other hand, another one of our apps is deployed using click once. That was easier, alter the first 3 digits in the project "publish" tab, and then add the reference to <property name="CCNetLabel"/> in or nant build for the last digit.

...

Ok, now that I'm thinking about it, the click once set up was a pain too, because MS stores the version number in 2 places, the first 3 numbers or stored in the proj file, the last number is stored in a user file buried somewhere (google applicationversion and "%2a" and you should be able to find that info)

FitNesse command line issues with .Net 2.0

Malcolm Anderson — Thu, 03 May 2007 20:27:50 GMT

There seems to be a small handful of people who are using FitNesse with .Net 2.0 and a smaller handful (me included) are having issues with the fitness command line tool (TestRunner.exe)

The error is Type <yourTestTypeName here> could not be found in assemblies.

I posted the following comment over at Cory Foy's APress blog in his highly useful entry

Getting up and running with Fitnesse.NET

If anyone else is either having these kind of problems or if you have solved this problem, please let me know. I'll make sure to post the solution when we find it.

******************************************** Begin Post

We're running the dotnet2 dlls and having an interesting issue.

We have 2 sets of 2 test runs using "Division"

Set1, Test run 1 : Run Division on localhost using the web page and dotnet2\FitServer.exe
Set1, Test run 2 : Run Division on localhost using the command line and dotnet2\TestRunner.exe

Set2, Test run 1 : Run Division on buildserver using the web page and dotnet2\FitServer.exe
Set2, Test run 2 : Run Division on buildserver using the command line and dotnet2\TestRunner.exe

Set1, Test1 : Pass
Set1, Test2 : Pass
Set2, Test2 : Pass
Set2, Test2 : Fails with:
System.ApplicationException: Type 'MyTestNameSpace.Division' could not be found in assemblies.

Is there some kind of permissions issue with the build server that could be causing this issue?

******************************************** End Post

The Perils of Pre-Optimization

Malcolm Anderson — Thu, 15 Feb 2007 15:52:00 GMT

More computing sins are committed in the name of efficiency (without necessarily achieving it) than for any other single reason - including blind stupidity.
Wulf, W. A. "A Case Against the GOTO," Proceedings of the 25th National ACM Conference, August 1972, pp. 791-97.

I came across this quote this morning and had to throw it back out into the wild.

I have run into people who, when given a choice between making their code more readable vs. shaving a millisecond off the running time of a particular process are confused by anyone who would go for readability.

Now to their credit, these same people are used to working deep in the bowels of production loops, where shaving a millisecond off of a process will compound to a point of saving hours of processing time a day. It really is a big thing where they come from, but when they enter my world, that of business software, they go insane at my complete and utter lack of care for their milliseconds. Granted, my users don't want to wait any longer than they have to for the response to their query, but when response time is measured in seconds, a millisecond here or there really isn't going to make any difference.

Using NantContrib's msbuild task

Malcolm Anderson — Sun, 04 Feb 2007 00:47:00 GMT

I'm in the process of trying to make a copy of tree surgeon that does a couple of different things. (If you don't know what tree surgeon is and you're a dotnet developer, I can not say enough good things about it, go, run, don't walk, get it)

One: I want to implement Jean-Paul Boodhoo's version of the Model-View-Presenter framework which he presented in DNRTV episode 14

Two: I want to provide a Console, a WinForm, and an ASP.Net front end.

Three: I want to upgrade treesurgeon to 2005.

In light of that, the first thing I had to do was to get the source code for treesurgeon and then upgrade it to VS2005. No big deal, just open it with 2005, up pops the wizard, it converts, again, no big deal.

My next issue was to try out the go.bat.

Uh-oh, NAnt doesn't do 2005 solutions, but nant contrib does "msbuild" tasks. Unfortunately, the nant contrib documentation assumes that you know msbuild, and doesn't explain how to chose which configuration to build with. Thankfully the blogosphere provided the answer with google pointing me over to Brig Lamoreaux's blog where he off handedly gave me what I needed to convert the nant task (Thanks Brig)

Here's the original target

Here's the final version of the target.

It was the "property" setting for "Configuration" that I would have eventually come to after studying the msbuild documentation, which I don't really want to do, so again, a big "Thanks Brig" for saving me the time.

Windows Media Player "Feature"

Malcolm Anderson — Wed, 03 Jan 2007 04:53:00 GMT

So last night my windows media player developed a fascinating bug.

Prior to last night, I could play a movie in one screen and do work in the other. But not anymore.

A Computer Express article from July of 2004 (Try the cached version here) had this to say:

"3. Video playback Windows Media: Windows Media Player automatically detects the availability of a double display configuration. When you play the media clip (videos, DVC, VCD) Media Player automatically plays it in full screen mode on the secondary display. On the primary display, the full UI version of Windows Media plays. This way, you can adjust the video properties, audio graphic equalizer settings on the base display, while full screen video is projected on the secondary screen (typically the projector). "

Peter Provost mentioned this bug back in August of 2003 and he doesn't seem to have gotten an answer for his question, "Does anyone know how to stop this behavior? Ideally it should be something I can control on the fly. It is very annoying when you don't want it, but useful when you do want it"

The worst part is that up until the last day of 2006, my media player worked with out this really nasty feature. And then *poof* January 1st 2007, a 3+ year old bug .... sorry, feature, shows up to haunt me.

Does anyone know of a solution? I would like to be able to alter the configuration of windows media player to either hijack my secondary monitor, or free it up for my use.

Lastly does anyone know someone on the windows media team who might know the answer?

[Edit]

Found it, I don't know who the realtimesoft.com people are, but they pinpointed the solution to an irritating issue.

When playing a video, it is also shown fullscreen on the other monitor
This happens if the video mirroring feature of your video card is enabled. You can disable this option on the tab for your video card under advanced display properties, Nvidia calls this feature Full Screen Video, for ATI video cards it is called theater mode, and Matrox calls it PureVideo/DVDMax.

I went in to my NVidia menu, and there on the GForce 6500 menu (it's an odd little side popout menu) was an entry for "Full Screen Video". In the second section there is a drop down for "Full screen device", some how it had gotten set to "auto select", I put it back to "Disable" and all was right with the world again.

A tool for capturing testing objects

Malcolm Anderson — Mon, 24 Jul 2006 19:34:00 GMT

I posted my code for a serialize and I just thought up an idea for making it really useful.

I'm retroactively trying to add some unit tests to some legacy code. I've got a working version of the code, and I'm thinking that I need to add a method or two to my serialization tool.

First thing, is I need a way to put a bunch of save hooks in and then have it trigger for all my objects that I care about catching their state. I'm thinking a static boolean that can be set on with a test for a field being equal to a key value.

Also, (more statics) having some kind of common name for the variables being used in that run, like if we are looking for Joe Smith we make all of our serializing tags “JoeSmith.060724.093454.classname”

I haven't implemented this yet, but it should work. This will (assuming that it works) allow me to record a specific test case's relevant values across as single run.

My Session was small, but it rocked

Malcolm Anderson — Mon, 24 Jul 2006 12:49:00 GMT

Friday night, my presentation congealed into garbage.

This was not a good thing.

I had to skip Jeff Levinson's “Test Driven Development with Team System” in order to work on my presentation.

I was competing with “Poker Bots for Fun and Profit”, with Jeff Berkowitz, Stuart Celariar's “Understanding Subversion“, Rory's “An introduction to WCF for the complete beginner“and 4 others. There was a part of me that was hoping that no one would show up, because then I could go see one of the other presentations, and another part of me wanted a ton of people.

I ended up with 4 people in my session, of which one of them was Chris Kinsman. When he said it, I was sure I knew that name, then it hit me, “were you on DotNetRocks about 10 months ago?”

And of course he said “Yes”.

crap.

I also had Chris Tavares, who just got hired on at Microsoft over at the patterns and practices group.

double crap.

These people knew what they were doing, why were they here. crap, crap, crap.

Also in the room were 2 people who were working on projects that needed exactly what I was selling. They were working on projects where the unit testing either wasn't happening, or they were completely bogged down in dependencies.

What came out of my session, besides the feeling of triumph for setting out to do, and completing, my first conference talk, were a couple of things. The two people who needed the techniques got good value, I know this because I asked. The two people who more than knew what they were doing, provided me with great feed back.

The first thing is that there is this Dependency Injection Framework, called Spring, (and in our case Spring.net), and they were expecting that I was going to be doing that talk. I've been looking into spring.net, and I'm kind of excited about it, looks cool, but I'm not yet, real clear on how it does what it does.

Second thing was that my “simple sample app” needed to still be MUCH simpler, and then I needed to walk through the process of creating my stub.

Finally, I needed to explain my constraints, which was

A) Working in a shared library, where interface changes were outside of the scope of the program.

B) That it was aimed at working with Legacy code.