Creating a TFS Work Item from ClojureCLR

At my day job we use Team Foundation Server 2008 (TFS) for our automated builds, iteration management, and source control. TFS may not be the most ideal way to manage these processes but in our MS environment it has helped us communicate with our non-technical team members and customers.  So we’ve been looking into ways to add bugs automatically when automated tests fail (see Creating A TFS WorkItem From IronRuby) or errors are reported from our applications via a message queue.

This morning I had a little time to do some ‘research’ on how to programmatically create a new bug work item.  My goal was to write code that would create a new work item bug with an image attached to it.  Before I get into the code let me take a second to introduce you to a TFS Bug Work Item.

A TFS Bug Work Item

In TFS Work Items are a way to track work that needs to be done. There are a total of five different type of Work Items available but we typically only use three: Task, Scenario, and Bug. Each Work Item type UI has different fields. Here is what a Bug Work Item looks like in our environment.

In this example we will create a new bug and enter text into the highlighted fields plus attach an image file. In order to create the bug I need to do a little setup.

The Setup

There are three assemblies needed to create a TFS Bug Work Item. The are: Microsoft.TeamFoundation.dll, Microsoft.TeamFoundation.Client.dll, and Microsoft.TeamFoundation.WorkItemTracking.Client. All three of these DLLs can be found in the C:\Program Files (x86)\Microsoft Visual Studio 10.0\Common7\IDE\PrivateAssemblies directory. I copied the three Dlls into the project’s libs directory.

The Code

The first step is to load the TFS assemblies from libs directory using the assembly-load-from function. I follow that call up with the import function pulling in the specific classes I need.  Once the assemblies and classes have been loaded I can start working on creating a bug work item. First, I created the -main function. The first few lines of –main are responsible for creating the necessary objects for WorkItem creation.

The first line gets the server object by making a call to the static method TeamFoundationServerFactory.GetServer passing in the name of the server I am working with.  The returned server object is used to create a WorkItemStore object.  A WorkItemStore encapsulates the data store that contains all work items on a particular server.  The next line gets the TFS project that I want to work with by using the filter function.  The last setup line uses the project object to retrieve the WorkItemType object that represents a bug.  Now I 'm are ready to create a new bug WorkItem.

Creating the Bug WorkItem

Since I will be creating bugs that have attachments and bugs that do not I created a function that creates and returns a WorkItem object. As I stated previously, the work items created this way will have the bare minimum values: title, area, iteration and description. The code is straight forward, just create the WorkItem object by passing in the WorkItemType object. Next I set the properties of for the four parameters. When everything has been set I return the WorkItem object.

Adding the Attachment

Now that I have my WorkItem object its time to add my image file as an attachment. I wrote another small and simple function to create the Attachment object. It takes three parameters, the WorkItem, the path to the file to attach and a description of the file.  First I use the path and desc parameters to create the Attachment object.  After the Attachment is created I can added it to the WorkItem object’s Attachments collection by calling the Add method passing the Attachment object as its only parameter. 

At this point I have a WorkItem object with an Attachment in memory.  At this moment the WorkItem is not visible through the TFS toolset.  I still have one last line of code to write to ensure that the WorkItem is saved. Surprisingly enough all I need to do is call the WorkItem’s save method.

(.Save work-item)

Now, If I look at the Pending Bugs report in TFS I will see my newly created bug in the list.

Summary

This post is another illustration of how easy it is to use .NET Assemblies from ClojureCLR.  In this case I walked you through the simple process of creating a Bug WorkItem in TFS.  Having the ability to programmatically create and report bugs will help us close the feedback loop in our applications. 

In The Future

In my ‘free time’ I will be checking into TFS’s build and source control APIs to see what we may be able to do improve our processes.  If and when I find anything of interest I will be sure to do a post or two about my findings.

Resources

TFS API: http://msdn.microsoft.com/en-us/library/bb130146(v=vs.90).aspx

My Source (this blog’s code is the 5-create_tfs_work_item):https://github.com/rippinrobr/clojure-clr-intro/zipball/master

My First Attempt at Porting Libraries to ClojureCLR

A while back David Miller wrote a couple of blog posts (Porting libs to ClojureCLR: an example and Porting effort for Clojure contrib libs) about porting Clojure contrib libs to ClojureCLR.  In the two posts he made it look like porting most of the libs wouldn’t be too terribly hard.  After a long period of telling myself, “tomorrow I will pick a lib to port and get started”, I actually did do just that a few days ago. I am about to start a project that needs to process command line arguments and it just so happens that David flagged the tools.cli lib as a trivial port.  So I decided to start my porting career with tools.cli.

The Setup

I created a directory off of my usual code directory called clj-ports just to help me differentiate that code from everything else I’m working on. Next I cloned the existing github project (https://github.com/clojure/tools.cli).  After cloning the project I attempted to compile the project with clojure.compile. For tools.cli the compilation went off without a hitch. Now that I have a clojure.tools.cli.clj.DLL  I’m ready to start testing.

The Testing Process

Now that I have my DLL I’m ready to start up the REPL and run the tests. The first step to running the is to load the clojure.test and the clojure.tools.cli-test libraries. After loading the libraries I can run the tests.  When I ran the tests and very quickly I hit my first error.

The error was a System.MissingMethodException on a call to a split method. I scoured the cli.clj file for a call to split but couldn’t find one. After taking a closer look I realized it was in the TESTING code. I opened up cli-test.clj, changed split to Split. When I re-ran the tests I hit another error. This time the test failed because it expected a Char[] and was passed a String.

The problem is in the testing code and is one of the parameters to the Split method call.  Since it was looking for a Char[] and I’m passing in a one character string I thought I’d try to use (first “-“) to get around the issue.  So I updated the test code and the change comment and re-ran the tests.  The error message I received this time was in the cli.clj file and not the test file.  Now that I have resolved the issues in the test file I can add my end of line comments following David’s instructions in his post Porting effort for Clojure contrib libs.

The new error is:

actual: System.MissingMethodException: Cannot find member startsWith matching args at .CallSite.Target (:0)

I opened up the cli.clj file and searched for startsWith calls. I found four places where the java startsWith method was used.  I changed all four to use the .NET equivalent StartsWith (and of course I commented the lines accordingly).  Remember, after you make a change to the source file you will need to re-compile (at least that is the way I did it) which means you’ll need to exit out of your REPL session first.  After recompiling and getting my REPL session back in order I re-ran the tests.  This time they all PASSED!

Now that I have all greens on my test it is time to create a new project on github called cljclr.tools.cli. After creating the project on github and setting up my machine I copied all files from the clojure.tools.cli project into the new cljclr.tools.cli project.  Next, I changed the namespaces in all code files to reflect the new project name and re-ran the tests.  All green! I checked the ported code into github and called the port complete. The ported project lives at cljclr.tools.cli.

Sample App

Since this was my first time porting a library I wanted to make sure it worked I used the ‘application’ above to check.  The first part of the main function is where the call to the cli function is which is where the available command line options are configured. The parameters are the args passed to the main function followed by a vectors that illustrate the allowed command-line options.  The returned vector has three members, a map with the parameters and their values, a vector with any additional parameters passed on the command line, and the string that contains the ‘banner’ which is used to show the available parameters.  After the call the cond function  is used to figure out what to do.  Each option has a simple println if one of the available parameters is passed in.  If no parameters are passed the usage banner is printed.  After compiling and running the app I was convinced that the port worked! 

Since That was Easy…

I thought I’d try my hand at another lib.  This time I chose to port tools.macro.  I went through the same configuration process, cloning, compiling and starting up the REPL so I can run the tests.  This time I hit NO errors when I ran the tests. That is the easiest type of port! 

Like a gambling addict I pressed my luck and chose one more library to port.  Since the algo.monads library was flagged as trivial I grabbed it.  When I ran the tests I was hoping for the same results as the tools.macro port but this time I hit an error.  The first error was another simple swap a java class for a .NET class: java.lang.IllegalArgumentException for System.ArgumentException and re-ran the tests.  The second error was in an extend-protocol call  using java.lang.String.  I simply swapped the java.lang.String for a System.String call.  When I ran the tests again all tests passed.  Another port complete!  The project can be found at cljclr.algo.monads on github.

Summary

Porting libraries to ClojureCLR wasn’t nearly as time consuming or complicated as I thought it would be.  In fact I have probably spent more time on writing this blog post than I did doing the ports.  Of course I picked libraries flagged as trivial ports which is probably why it didn’t take long to finish.  What porting these libs over did teach me was the process: Clone, Compile, Test, fix, retest.  For my next porting attempt, the data.csv library I will add a step to search the cloned files for easy fixes like startsWith to StartsWith and other mappings that David laid out in his posts.  Please join the porting party and port one today!  I have not tested the ports against 1.2, all of my ports were done with 1.3.  I will continue to test only with 1.3 unless others feel that we need to test against 1.2.  If you use any of the libraries I’ve ported and run into a bug please do not hesitate to let me know.

Resources

Blog Posts: Porting libs to ClojureCLR: an example and Porting effort for Clojure contrib libs

Ported Projects: cljclr.tools.cli and cljclr.algo.monads

Projects that work ‘As Is’: clojure.tools.macro

In order to help others see which libraries have been ported and/or tested for ClojureCLR I have created a page that lays out which libraries have been ported and the location of the github project. If you have ports you’d like listed on the page please let me know. The ClojureCLR Ports Status Page is here.

Intro to Clojure-clr: Creating a UI with Windows.Forms

In this post I will walk you through the creation of a Windows.Forms based user interface using ClojureCLR.

The Setup

To follow along with the blog you will need to have ClojureCLR installed and have access to a MySQL and SQL Server instances. If you haven’t installed it yet follow the directions in my Getting Started with Clojure-clr post. If you need to install one or both of the bases you can download and install them from here: SQL Server Express and MySQL.  Once you have downloaded and installing the databases the last step you’ll need to take is to grab the mysql_schema.sql file and run it in your MySQL database. If you are not familiar with MySQL you can run the file from the command line by running this:

mysql –user=username –password=password < mysql_schema.sql

This will create a database and three tables: Master, Schools and PlayerSchools. The three tables will be migrated into a new SQL Server database. These tables are from a database created by the group baseball-databank.org who have collected baseball statistics from the beginning of professional baseball to the present. I find using this data more exciting then the usual demo data.  Anyway, once you have the loaded the schema into your MySQL instance you are ready to start building the user interface.

UI.clj

Surprisingly enough the ui.clj file is where the user interface code lives. Here’s what it looks like when it is first started.

I won’t bore you with a line by line analysis of the source but I will show you highlights of the UI creation process, walking you through the steps below. Creating Windows.Forms apps follow the same process without regard to the language that they are being creating in. So there aren't any magic steps here. However, creating the UI in Clojure was a lot of fun.

Step 1. Including the necessary assembly and classes

The first step in creating the UI is to bring in the System.Windows.Forms assembly which I did by calling LoadWithParitalName. Next, I used :require to bring in the classes I needed create the UI with. Since I want to be able to run this app outside of the REPL, I call :gen-class to generate ui.exe.

Step 2. Instantiating the UI Objects

After I have loaded and required everything I need I can start creating objects. As you can see the creating objects is pretty straight forward, all I do is new up all the objects. In the next step is where the configuration of the objects and the event handlers are added.

Step 3. Setting Properties

The snippet above contains a sampling of the property setting code for the title label, the group box, the CheckedListBox and the 'Load Table' button. If you’d like to see the entire file you can do so here. You may notice all the calls to methods with names like set_Text and wonder where the methods come from since you do not see them when you look at the Label class in the Object Browser. In ClojureCLR we access the .NET objects at the CLR 'layer' which is different than the we access them through C#. The set_Text method is the CLR representation of the C# code Label.Text = "Some Text". If you’d like to see what other methods are available you can use the ildasm tool, which is a part of the .NET SDK. Here is an example of some of the set_ methods for the Label class:

Step 4. Adding the UI Objects to the Form and displaying the it

Now that I have all the objects customized I would like to show them to the user. In order to do that I need to add them to the Form object. The Form class has a Controls collection where all of its child objects are stored. I added the objects to the collection by calling the Controls.Add method for each object. When all of the objects have been added I call the ShowDialog method to display the UI.

Step 5. Adding Click Event Handlers to the Buttons.

A snazzy UI is great but if it doesn’t do anything what good is it? Handling the Button.Click events for both of the buttons is where all the action happens. In order to wire up the Button.Click event handlers I needed to use the gen-delegate macro. The macro’s signature looks like this:

gen-delegate
([type argVec & body])

The Load Table Click Event Handler

Here's what the 'Load Table' button's click handler looks like:

In my call to gen-delegate the parameters begin with passing the type System.EventHandler since we want a EventHandler delegate created. The next argument is the a vector which in this case contains the usual Click event parameters: sender and args. The last parameter is where the code of the event handler reside. In this case when the 'Load Table' button is clicked all of the table names from my bdb_post_2010 database are returned and are added to the CheckedListBox. After clicking the 'Load Table' button the app looks like this for me. Your table listing will look different.  The SQL File I provide only has three tables.

The tables are retrieved by creating a MySQL connection which is then passed to a function that retrieves the list of tables in the specified database.

The returned lazy sequence I passed through doseq so the tables can be added to the table list.  The table names are added by making a call to CheckedListBox.Items.Add method.  Once all of the table names have been added I close the database connection.

The Migrate Tables Event Handler

The ‘Migrate Tables…’ button’s event handler does a similar task as the ‘Load Table’ handler does, except it interacts with both MySQL and SQL Server databases. 

First it creates a connection to the MySQL database and uses the value from the 'New DB Name' TextBox to create the new database on the SQL Server instance. Next, doseq is called to process the values in the CheckedListBox.CheckedItems collection. Each of the selected table names is passed to mysql/get-columns to retrieve a sequence of maps that contains information about each column in the table. The returned sequence is then passed with the table name and SQL Server database object to sql/create-table which to creates the table in the SQL Server database. After all of the selected tables have been migrated to SQL Server the MySQL connection is closed and a MessageBox is displayed to inform the use that the migration is complete.

db/core.clj and db/sqlserver.clj

Since this is a blog post about creating a UI I will not be delving into the minute details of how the database related code works. However, I do think it is worth giving you a quick overview of the code. The DB code in db/mysql.clj, db/core.clj and db/sqlserver.clj are based on the code I wrote in Intro to Clojure-clr: Connecting to SQL Server and MySQL.

The db/core.clj file contains one function named run-sql which does what you might expect, it executes a query. In my previous database connectivity post I used (.Read reader) to loop through the results of a query but here I've switched to using the System.Data.DataTable class. Using the DataTable class allows me to close the reader right away and retrieve the column names for the query very easily. I used the column names with the zipmap function to map the column name to its values for each of the returned rows. This allows me to return a sequence of maps that represented the results of the query.

The db/mysql.clj file has three methods: get-connection, get-tables and get-columns. Each of these functions does what their names imply. get-tables uses 'show tables;' and get-columns uses 'describe [table name];'.

On the migration side of the app, the db/sqlserver.clj file makes use of the Microsoft.SqlServer.Smo assembly to create the new database, tables and columns. If you want more information on the visit the Microsoft.SqlServer.Management.Smo MSDN Page.

Running the App

Starting the application from the REPL is straight forward.  Start the REPL in the same directory that as the ui.clj file. When the REPL is ready enter:

You should then see the UI appear with an empty CheckListBox.  Once the app is up and running click on the ‘Load Table’ button.  If are used the SQL file I mentioned above you should see three tables:  master, schools, and schoolsplayers.

Go ahead and click all three tables.  You will need to click on the table once to highlight it and then either click on the check box or press the space bar to put a check in the box.

Next, enter the new database name.  My database will be named clr-intro-4. Finally, I kick off the migration click on the ‘Migrate Tables…’ button and within a few seconds you will have a new database with the three tables added. When the migration has completed you will see something similar to the image below.

Summary

Building a UI with System.Windows.Forms in ClojureCLR is straight forward. You only need to bring in the System.Windows.Forms assembly and follow the typical process of creating a Forms based application.

Adding event handlers is as easy as making use of the gen-delegate macro passing it the ‘guts’ of what you want to do when a particular event is fired.

About the Data

The data I’m using for this blog post comes from the Baseball Databank project.  The project has gathered all the baseball status from previous season and offers the data in many different formats.

Resources

Previous Posts: Getting Started with Clojure-clr, Intro to Clojure-clr: Connecting to SQL Server and MySQL

Database Downloads: SQL Server Express, MySQL

Data: mysql_schema.sql

My Source: https://github.com/rippinrobr/clojure-clr-intro/zipball/master

More Examples and Information: Another Windows Forms Example, Microsoft.SqlServer.Management.Smo MSDN Page, and Ildasm.exe page on MSDN