Beige Sunshine

You know, sometimes, you really shouldn’t test your code.

What?

Yes. You heard me. You shouldn’t test your code.

Well, let me explain that better. I recently took a course where I was paired with someone where neither one of us knew what it was that we _really_ wanted to do. We started to flush out an API for wiring up a hand rolled IOC automatically, and every time we did so, we hit a brick wall. The reason is that we kept trying to ‘test first’.

Don’t get me wrong, I am a complete advocate of test first development. HOWEVER, if you don’t know what the hell you’re doing, you can’t really test first. Being able to test first implies that you can think ahead of yourself and write something that does exactly what you know you need.

In our case, we kept wasting time trying to write tests that just didn’t work with how we NEEDED to do things. How did we solve this dilemma? We sat down and spiked out some code. Just hammered it out and got something working. Now, let me tell you, we wrote some pretty damn horrible and untestable (except via black box testing) code. However, what we did gain was an insight into the process and steps we needed to take to accomplish our task.

However! We didn’t keep that working code we spiked out. Once we had a spiked set of code and a solid idea of what it was we needed, we were able to test and rewrite the code we’d spiked in a much better, tested and well designed manner that would have taken forever if we’d just kept banging our heads against the wall.

To reiterate:

1) TDD with no idea what we were trying to accomplish = 3 hours wasted.

2) Spiking out some code, getting our heads around the steps needed, wiping it clean and restarting the TDD approach with an idea of what we needed to do = less than 1 hour to completion.

Remember, testing first doesn’t mean you should beat your head against a wall just to write a test before writing some code. Sometimes, in order to write a test, you need to write some code first. I guess some people will write tests afterwards, I personally prefer to rewrite, as I find I spend just as much time fixing the code when retrofitting tests as I do just rewriting it out.

In the end, you just need to remember that even if you’re doing TDD, you shouldn’t be afraid to spike out code. It’s the best way to learn and give yourself some direction.

I just added an MsBuild task for rubicant tonight. I’m not very familiar with the command line options for MsBuild so some of these could probably be improved a bit, but until I know more I will leave them as they are.

The MsBuild Task is used to create tasks which will use MsBuild. This should help with people who need to build WPF applications (since MsBuild is the only way right now), or (fingers crossed) migrate from MsBuild.

The MsBuild Task is used as follows:

   1: MsBuildTask.new(:task_name => :dependencies) do |msb|

   2:   msb.no_autoresponse

   3:   msb.target

   4:   msb.logger

   5:   msb.distributed_logger

   6:   msb.console_logger_parameters

   7:   msb.validate

   8:   msb.verbosity

   9:   msb.no_console_logger

  10:   msb.max_cpu_count

  11:   msb.ignore_project_extensions

  12:   msb.file_logger

  13:   msb.distributed_file_logger

  14:   msb.node_reuse

  15:   msb.properties

  16:   msb.file_logger_parameters

  17:   msb.project_file

  18: end

Each of setters matches a command line option for MsBuild. For more information on what each one does, please view the documentation at http://msdn.microsoft.com/en-us/library/ms164311.aspx.

Values that can only be set (for example, /noautoresponse) can be set to true (set) or nil (unset).

Values, such as /property that take a list of option settings, must be passed a hash where the keys are the options and the values are the option values. For example, to set the output directory and Warning Level for a csproj compile, you could use:

   1: MsBuildTask.new(:compile) do |msb|

   2:   msb.properties = { :OutputDir => ".\\Output", :WarningLevel => 2 }

   3:   msb.project_file = ".\\Project1\\project.csproj"

   4: end

As of yet, I haven’t added an MsBuild helper to help create this function. I will probably get this done fairly shortly.

The mbunit function in rubicant has been changed from just a function to a complete task.

MbUnitTask

Behind the scenes, what MBUnitTask does is copies over the minimal necessary support files from an MBUnit directory and then runs the tests. It uses the MBUnit Console Runner and is only tested with MBUnit 2.4. It is used as follows:

   1: MbUnitTask.new(:task_name => :dependencies) do |mbunit|

   2:   mbunit.mbunit_dir = "/path/to/mbunit" 

   3:   mbunit.test_dir = "/path/to/testing/dir" 

   4:   mbunit.test_files = ["tests.dll"]

   5:   mbunit.assembly_path => FileList["/assemblies/to/include/*.dll"],

   6:   mbunit.report_folder => "sets /rf:",

   7:   mbunit.report_name_format => "sets /rnf:",

   8:   mbunit.report_type => "sets /rt:",

   9:   mbunit.show_reports => "sets /sr:",

  10:   mbunit.transform => "sets /tr:",

  11:   mbunit.filter_category => "sets /fc:",

  12:   mbunit.exclude_category => "sets /ec:",

  13:   mbunit.filter_author => "sets /fa:",

  14:   mbunit.filter_type => "sets /ft:",

  15:   mbunit.filter_namespace => "sets /fn:",

  16:   mbunit.verbose => "sets /v{optionvalue}",

  17:   mbunit.shadow_copy => "sets /sc:" }

  18: end

Mbunit helper

   1: mbunit(name, dependencies, mbunit_dir, test_dir, test_files, opts={}) 

Usage is:

   1: mbunit(:run_tests, [:compile_tests], "/path/to/mbunit/dlls",

   2:         FileList["/test/dlls/test.dll"],

   3:         {:working_dir => "/dir/to/run/tests/in",

   4:          :assembly_path => FileList["/assemblies/to/include/*.dll"],

   5:          :report_folder => "sets /rf:",

   6:          :report_name_format => "sets /rnf:",

   7:          :report_type => "sets /rt:",

   8:          :show_reports => "sets /sr:",

   9:          :transform => "sets /tr:",

  10:          :filter_category => "sets /fc:",

  11:          :exclude_category => "sets /ec:",

  12:          :filter_author => "sets /fa:",

  13:          :filter_type => "sets /ft:",

  14:          :filter_namespace => "sets /fn:",

  15:          :verbose => "sets /v{optionvalue}",

  16:          :shadow_copy => "sets /sc:" } 

Runs the tests in test_files in the test_dir with the dlls from mbunit_dir using the options in opts.

For more information on the command line arguments for the mbunit console runner, please visit: http://docs.mbunit.com/help/html/gettingstarted/MbUnitConsoleRunner.htm

Since I already had rubicant building as a gem, it made sense to make it available as a gem. I have moved the project to github and it is now located at: http://github.com/mendicantx/rubicant/

In addition, you can also install rubicant via gem:

   1: gem sources -a http://gems.github.com

   2: sudo gem install mendicantx-rubicant

You should only ever have to run the gem sources command once. This adds github to your sources of remote gems. From what I understand, there’s a ton of great projects there.

Now you don’t have an excuse to try it out!

Don’t forget to check out Part 1 and Part 2.

Rubicant basically has 4 main parts, as mentioned in the first part of this series. They are as follows:

The CscTask

The CscTask is used for compiling C# files. It’s usage is like:

   1: CscTask.new(:task_name => :dependencies) do |csc|

   2:   csc.target = 'library'

   3:   csc.references = FileList["libs\*.dll"]

   4:   csc.out = "hello_world.exe"

   5:   csc.debug = "full"

   6:   csc.files = FileList["src/**/*.cs"]

   7: end

target: Any target that the csc compiler takes (exe, winexe, library, etc)

references: The list of assemblies that are referenced during compilation

out: The output filename

debug: if this is set(to anything), the compiler will be passed /debug:full to create debugging information

files: The list of files to be compiled

CscTask also has a shortcut that can be accessed as follows:

   1: csc(name, dependencies, opts)

   2:  

   3: ex:

   4: csc :name, [:dependencies], { :target => 'library',

   5:                               :references = FileList["libs\*.dll"],

   6:                               :out = "hello_world.exe",

   7:                               :debug = "full",

   8:                               :files = FileList["src/**/*.cs"] }

The MBUnit Helper

The MBUnit helper is used to run tests using mbunit v2. It is used as follows:

   1: mbunit(mbunit_dir, test_dir, test_files, opts={})

   2:  

   3: ex:

   4: mbunit("/path/to/mbunit/dlls",

   5:         FileList["/test/dlls/test.dll"],

   6:         {:working_dir => "/dir/to/run/tests/in",

   7:          :assembly_path => FileList["/assemblies/to/include/*.dll"],

   8:          :report_folder => "sets /rf:",

   9:          :report_name_format => "sets /rnf:",

  10:          :report_type => "sets /rt:",

  11:          :show_reports => "sets /sr:",

  12:          :transform => "sets /tr:",

  13:          :filter_category => "sets /fc:",

  14:          :exclude_category => "sets /ec:",

  15:          :filter_author => "sets /fa:",

  16:          :filter_type => "sets /ft:",

  17:          :filter_namespace => "sets /fn:",

  18:          :verbose => "sets /v{optionvalue}",

  19:          :shadow_copy => "sets /sc:" }

For a complete listing of the mbunit command line options, please visit http://docs.mbunit.com/help/html/gettingstarted/MbUnitConsoleRunner.htm.

This runs the tests in test_files in the test_dir with the dlls from mbunit_dir using the options in opts. Currently, this will physically copy the minimum required mbunit dlls and mbunit.cons.exe to the test dir, run the tests and then remove the files.

The CopyFiles Task

The copy files task will copy files from one directory to another, while maintaining the folder structure.

   1: CopyFiles.new(:name) do |cf|

   2:   cf.base_dir = "/some/dir"

   3:   cf.file_globs = ["**/somefiles/*.cs", **/otherfiles/*.cs"]

   4:   cf.target_dir = "/another/dir"

   5:   cf.excludes = ["AssemblyInfo.cs"]

   6: end

base_dir: The base directory to start copying files from

file_globs: The globs of files to be copied from base_dir

target_dir: The target directory to place the files in

excludes: (Optional) list of files to be excluded from the copy.

The Expand Template File Task

Usage is:

   1: expand_template(template, output_file)

This will read in the file specified by template and expand it out and write it to the location of output_file.

WARNING: I am fairly certain that the only reason this is working right now for me is because I use configatron, which is accessed as a singleton. It is entirely possible that variables declared in your build script will not be accessible during template expansion. I will be testing this in the near future, however.

Framework Selection

Currently, only the .Net 2.0 and .Net 3.5 frameworks are configured. .Net 3.5 is the selection by default. However, you can choose which framework you want to use by adding framework={framework} in the build line. For example:

   1: rake deploy_local framework=net20

This will run the deploy_local task in the default rakefile using the .Net 2.0 framework.

Valid framework values are:

net20 - .Net 2.0

net35 - .Net 3.5

These frameworks are discovered using the registry, so you shouldn’t have to put in the location of your csc.exe file.

What else?

Well, that’s about all it really does right now. The rest is just built in rake and ruby goodness. I would like to add msbuild support for those that are building WPF apps. I would also like to add some nunit support, as well as some support for other things like easy creation of assemblyinfo.cs files and possibly some deployment tasks for iis7. Who knows. We’ll see how I like things and if I really take to using this (I think I will).

Update: There is also a Part 1 and a Part 3.

Alright, so here is a sample build script for rubicant. First you will have to get rubicant installed. To do that, you need to get the source from https://rubicant.googlecode.com/svn/trunk/. I believe the username is rubicant-read-only.

In the root directory type:

   1: E:\projects\rubicant>rake gem

   2: (in E:/projects/rubicant)

   3: c:/ruby/lib/ruby/gems/1.8/gems/rake-0.8.3/lib/rake/gempackagetask.rb:13:Warning:

   4:  Gem::manage_gems is deprecated and will be removed on or after March 2009.

   5: mkdir -p pkg

   6: WARNING:  no rubyforge_project specified

   7: WARNING:  RDoc will not be generated (has_rdoc == false)

   8: WARNING:  deprecated autorequire specified

   9:   Successfully built RubyGem

  10:   Name: Rubicant

  11:   Version: 0.0.1

  12:   File: Rubicant-0.0.1.gem

  13: mv Rubicant-0.0.1.gem pkg/Rubicant-0.0.1.gem

  14:  

  15: E:\projects\rubicant>

Now you’ve got the gem built, you just need to install it:

   1: E:\projects\rubicant>gem install pkg\Rubicant-0.0.1.gem

   2: Successfully installed Rubicant-0.0.1

   3: 1 gem installed

   4:  

   5: E:\projects\rubicant>

There. Now for a test application. Suppose we’ve got 3 projects that make up a console app. The UI, the Engine (I KNOW) and the Tests. They look like:

For now, let’s ignore the internal workings and focus on building it.

The build script looks like this:

   1: require 'rubicant'

   2: require 'configatron'

   3: include Rubicant

   4:  

   5: BASE_DIR = File.expand_path(".")

   6: SRC_DIR = "#{BASE_DIR}/src"

   7: SRC_APP_DIR = "#{SRC_DIR}/app"

   8: SRC_TEST_DIR = "#{SRC_DIR}/test"

   9: CONFIG_DIR = "#{BASE_DIR}/config"

  10: BUILD_DIR = "#{BASE_DIR}/build"

  11: DEPLOY_DIR = "#{BASE_DIR}/deploy"

  12: TOOLS_DIR = "#{BASE_DIR}/tools"

  13: MBUNIT_DIR = "#{TOOLS_DIR}/mbunit"

  14: RHINO_DIR = "#{TOOLS_DIR}/rhino"

  15:  

  16: UI_SOURCES = FileList["#{SRC_APP_DIR}/Rubicant.Test.App.UI/**/*.cs"]

  17: ENGINE_SOURCES = FileList["#{SRC_APP_DIR}/Rubicant.Test.App.Engine/**/*.cs"]

  18: TEST_SOURCES = FileList["#{SRC_TEST_DIR}/**/*.cs"]

  19:  

  20: #### Build Initialization

  21: task :initialize_build do

  22:     rm_rf BUILD_DIR if File.exists?(BUILD_DIR)

  23:     mkdir_p BUILD_DIR

  24:     rm_rf DEPLOY_DIR if File.exists?(DEPLOY_DIR)

  25:     mkdir_p DEPLOY_DIR

  26: end

  27:  

  28: #### Loading Properties

  29: task :load_local_properties do

  30:     configatron.configure_from_yaml("local.properties.yml")

  31: end

  32:  

  33: task :load_test_properties do

  34:     configatron.configure_from_yaml("test.properties.yml")

  35: end

  36:  

  37: #### Deployment Targets

  38: task :deploy_local => [:load_local_properties, :deploy]

  39:  

  40: task :deploy_test => [:load_test_properties, :deploy]

  41:  

  42: task :deploy => [:test, :expand_app_config] do

  43:     cp FileList["#{BUILD_DIR}/{hello.exe,engine.dll}"], DEPLOY_DIR

  44: end

  45:  

  46:  

  47: #### Compile Targets

  48: csc :compile, [:initialize_build, :compile_engine], {:files => UI_SOURCES,

  49:                                                      :target => "exe",

  50:                                                      :out => "#{BUILD_DIR}/hello.exe",

  51:                                                      :references => FileList["#{BUILD_DIR}/*.dll"] }

  52:                 

  53: csc :compile_engine, [], {:files => ENGINE_SOURCES,

  54:                           :target => "library",

  55:                           :out => "#{BUILD_DIR}/engine.dll"}

  56:                 

  57: #### Testing Targets                

  58: task :test => [:compile_tests, :run_tests]

  59:  

  60: csc :compile_tests, [:compile], {:files => TEST_SOURCES,

  61:                                           :target => "library",

  62:                                           :out => "#{BUILD_DIR}/tests.dll",

  63:                                           :references => FileList["#{BUILD_DIR}/**/*.{dll,exe}", "#{RHINO_DIR}/*.dll", "#{MBUNIT_DIR}/*.dll"]}

  64:  

  65: task :run_tests do

  66:     mbunit(MBUNIT_DIR, BUILD_DIR, FileList["#{BUILD_DIR}/tests.dll"], {:assembly_path => FileList["#{BUILD_DIR}/engine.dll", "#{RHINO_DIR}/*.dll"].to_a})

  67: end

  68:  

  69:  

  70: #### App.config expansion

  71: task :expand_app_config do

  72:     expand_template "#{CONFIG_DIR}/App.Config.template", "#{DEPLOY_DIR}/hello.exe.config"

  73: end

The point of this build file is to be able to churn out two different builds based on two different sets of properties. This is based off of my builds that I do at work where we need to deploy and test in a test environment before going into production. It would be as simple as adding a production.properties.yml and a deploy_production target to simulate this fully in this script.

So basically, we call deploy_test (or deploy_local), which in turn loads the test properties, and then the deploy task.

Deploy calls the test task, which builds the app, builds the tests, runs the tests. If any tests fail, the build will fail here, making sure we don’t deploy any code that doesn’t pass tests.

Finally, we expand our app.config in a way suitable for the environment we’re going into.

Finally, run the executable to see that it worked!

   1: E:\projects\Rubicant.Test.App>deploy\hello.exe

   2: Hello, World! (Test Version)

   3:  

   4: E:\projects\Rubicant.Test.App>

You can download the source from http://blog.beigesunshine.com/wp-content/uploads/2009/01/rubicanttestapp.zip and try it out for yourself. The default rakefile goes to an exe and an assembly. You can also run both deploy_local and deploy_test and create a single exe file by using rake -f rakefile_single_exe.rb deploy_local.

Coming up, a post on some of the specifics of rubicant.

Update: Don’t forget to check out Part 2 and Part 3.

In previous posts I have mentioned that I think that there might be a better way to automate builds than NAnt. I also mentioned that I had looked at a few of them and been mostly underwhelmed. I’ve spent a large amount of time building a process for builds that works well for me and none of them seemed to erase my pains.

I also mentioned that I would be working on some items that may even drive some content to this blog, among other things. Today, I introduce the fruits of my labour.

Please note that this is NOT recommended for any sort of production environment. It is currently at best a way to try out Rake on .Net projects.

What is rubicant?

rubicant is my own personal set of tasks that sit on top of rake, a rubyish interpretation of make. I’ve used ruby for some scripting in the past, and also used rake in a rails app that I developed. I felt that the combination of both would alleviate some of the pains that I (and I’m pretty sure I’m not alone here) feel when I use NAnt.

I was admittedly scared to even try building my own set of tasks, and actually didn’t even know where to start. I kept rolling the idea around. One day I was going to start doing it now. Then the next I wasn’t. As a father of young children my spare time is very limited and very few and far between, especially for something that would demand so much of it.

And then the I started coming across others experimenting with exactly what I was talking about. First it was a post by Dave Laribee, as well as another post in his comments. And now as I read over them again, I’m seeing that there’s more people feeling like I do. I started to google and found some great posts by Jay Fields and to a lesser extent one by Martin Fowler.

Anyway, in my looking it seemed like no one had actually put out a reusable library that anyone could just pick up and automatically compile a project within 5 minutes.

And thus, with some much needed inspiration from the above sources, rubicant was born.

What isn’t rubicant?

Well written for one thing. Take a look at my tests and they’re enough to make someone blow a gasket (I kid, I LOVE his twitter rants) I started writing them thinking I knew what I was doing and let me clear something up for you. At the time, I had a greatly misguided view on BDD/Specification testing. Either way, it’s still got my code covered and has helped to drive out some of the design, especially when loading framework information.

Another thing it is not is a complete solution. I had 4 goals in functionality that I wanted. They were:

1. Build C# code
2. Test C# code (with MBUnit)
3. Copy selected files in a meaningful hierarchy
4. Template expansion using configatron to allow for easy configuration swapping.

Why rubicant?

First off, automated builds are the only thing that I feel I truly have a grasp on. I feel comfortable enough in myself and my process that I could come up with something useful. I don’t think of myself as a guru, but I do know that I have a set of build scripts at work that are robust and involved while trying to stay (somewhat simple). In other words, I don’t consider myself a guru, but I do feel I have a grasp on the subject.

I was also feeling too much pain with NAnt. As well, I didn’t feel that any of the other solutions out there really felt like what I was looking for. I was also looking for a learning project in ruby. I wanted to try driving out a design via testing using ruby. A lot of things. Either way, I felt like I wanted to do something for myself, but that also might provide some sort of benefit to the community, even if it just ended up as a failed attempt. In a sentence: I wanted to try and provide SOMETHING back to a community that I feel has given me so much.

Even if this ends in failure, it’s been exactly the experience that I’ve wanted thus far, and though I’m sure I’m in for some rough waters ahead, I think that it will continue to drive me to learn, improve and better myself. Okay, so there is definitely some selfishness there, but hopefully you can forgive me in the name of Continuous Improvement, even if it’s my own.

How do I get it? How do I use it?

Well, if you want to get it, you can go to the project site on google code and download the source. There are instructions there on how to install it. As for how to use it… stay tuned for a sample project and build file. Hopefully very soon.

Update: Don’t forget to check out Part 2 and Part 3.

I told myself I would do a whole lot more writing than I’ve been doing. I’ve even got ideas of things I would like to write on! Honest. Anyway, between Christmas, kids and a project (that’s related to this blog!) that I’ve been working on, I haven’t had a lot of time to sit down and focus.

So to my whopping 0 subscribers, just you wait! There will be content. Oh yes. There will be content.

I’m not typically one to outright throw my beliefs in the faces of others just to be noticed, and that isn’t the intent here either. For those of you who haven’t heard or don’t know, Canada has a parliamentary form of government which you can kind of boil down to the fact that it’s possible for someone to run the government, but without a clear majority. This week in Canada, the left wing parties have decided to try and form a Coalition Government to overthrow the Conservatives.

As well, some of you may notice the reference in my blog to The Dead Milkmen a humorous punk band from the 80’s and 90’s.

Well a friend of mine and I have taken this opportunity to show not only a tribute to the Milkmen, but also to make light of the situation in the country. So here, for all posterity I give unto you my version of Beach Party Vietnam:

Bloc Party Quebecois

Duceppe and Bouchard with Michaëlle Jean
When Duceppe got a letter from Mr. Dion
Said get Layton and everyone
We’re gonna start a co-ali-tion.

With the Bloc Party Quebecois
Parle anglais? Je ne c’est pas!
Seperatists eating foie gras
It’s a Bloc Party Quebecois!

Hey Duceppe! Aren’t you gonna try and be PM?
I’m afraid I can’t do that Bouchard.
Why not?
Because I don’t have any support in english Canada!!!
(AAAAAAAAAAAAAAAAHHHHHHHHHHH!)

Bloc Party Quebecois
Parle anglais? Je ne c’est pas!
Seperatists eating foie gras
It’s a Bloc Party Quebecois!

Duceppe he wants Harpers head
Harper wants the Bloc dead.
Hangin on the east side
Gonna be some party genocide

It’s a Bloc Party Quebecois
Parle anglais? Je ne c’est pas!
Seperatists eating foie gras
It’s a Bloc Party Quebecois!

YEAH!!!

Today I was suddenly reminded of a question I saw somewhere a couple weeks back. I wasn’t going to admit it at first, but it was on Stack Overflow. I actually went looking for the question, but can’t seem to find it, so here’s my version of the story from memory.

The question basically said that after every commit to a code base, it would trigger a build on their build server. The build generated a file (I think it was called revision.txt) that was tagged with the current revision number, and then checked back into source control, which would then trigger a build, which would generate a file……. and so on.

I was so focused on the fact that they would be generating something that could be retrieved at any point in time from the repository that I failed to see the value of what was really being asked. I the fact that you would create a file with a revision number, and check it into the repository, effectively nullifying that revision number was so consuming for me that I didn’t stop to think that it might actually be helpful to not build on every commit.

Then this morning I was working on some test documentation for a deployment that we’re doing soon. As I made changes and checked them into the repository, I kept triggering builds. Then it hit me. I realized that there’s really no reason why a test doc that has nothing to do with a the build. I mean, in this case, the build doesn’t take long, but we have some pretty big projects here that can really hit our build server hard (which causes emails to be sent out to our server admins).

It didn’t take me long to find the Filtered Source Control Block and update our configuration to use it.

But it got me thinking. Though I don’t fully agree with the process being described in the original question, I was so focused on what was wrong that I completely missed out on learning about what was actually a valid, useful and soon-to-be relevant feature in my build server.