Martin Hinshelwood (MrHinsh) on Visual Studio ALM

A Scottish software developer | SSW Solution Architect | Microsoft Visual Studio ALM MVP | Microsoft Visual Studio ALM Ranger | Scrum Developer Trainer
posts - 443, comments - 722, trackbacks - 57

My Links

News

Profile photo of Martin Hinshelwood Martin Hinshelwood on LinkedIn
Martin Hinshelwood on Twitter
Martin Hinshelwood on Facebook
Martin Hinshelwood on Messenger
Martin Hinshelwood on Feedburner

Martin is a Solution Architect at SSW, and also one of three Microsoft Visual Studio ALM MVP in the UK and has over 9 years experience in the software industry. Martin has recently joined the ranks of the Microsoft Visual Studio ALM Rangers deliver out of band solutions for missing features or guidance.

Microsoft Visual Studio ALM MVP

Martin speaks at DDD Scotland, one of the top events in the UK along with a number of User Groups across UK and Europe. And is excited to be presenting the Professional Scrum Developer course which he is one of only three qualified trainers in the UK.

Professional Scrum Developer Trainer Certified ScrumMaster

He aims to continue improve the engineering practices of development teams in the UK and Europe. He does this by migrating them to, and coaching them in the use of, Microsoft’s ALM offering in combination with Scrum. These offerings include Team Foundation Server (TFS) and Visual Studio.

Microsoft Certified Technical Specialist: Team Foundation Server  

Disclaimer

The opinions expressed herein are my own personal opinions and do not represent my employer's view in any way.

Locations of visitors to this page

Twitter












Tag Cloud

Article Categories

Archives

Post Categories

Image Galleries

Blogs I read

Blogs of Friends

Personal

Projects

VSTS

Finding the problem on a partially succeeded build on Team Foundation Build Services 2010

Now that I have the Build failing because of a genuine bug and not just because of a test framework failure, lets see if we can trace through to finding why the first test in our new application failed. Lets look at the build and see if we can see why there is a red cross on it.

First, lets open that build list. On Team Explorer Expand your Team Project Collection | Team Project and then Builds. Double click the offending build.

image
Figure: Opening the Build list is a key way to see what the current state of your software is.

 image
Figure: A test is failing, but we can now view the Test Results to find the problem

 

image  
Figure: You can quite clearly see that the test has failed with “The device is not ready”.

To me the “The Device is not ready” smacks of a System.IO exception, but it passed on my local computer, so why not on the build server?

Its a FaultException so it is most likely coming from the Service and not the client, so lets take a look at the client method that the test is calling:

bool IProfileService.SaveDefaultProjectFile(string strComputerName)
{            
    ProjectFile file = new ProjectFile()
    {
        ProjectFileName = strComputerName + "_" + System.DateTime.Now.ToString("yyyyMMddhhmmsss") + ".xml",
        ConnectionString = "persist security info=False; pooling=False; data source=(local); application name=SSW.SQLDeploy.vshost.exe; integrated security=SSPI; initial catalog=SSWSQLDeployNorthwindSample",
        DateCreated = System.DateTime.Now,
        DateUpdated = System.DateTime.Now,
        FolderPath = @"C:\Program Files\SSW SQL Deploy\SampleData\",
        IsComplete=false,
        Version = "1.3",
        NewDatabase = true,
        TimeOut = 5,
        TurnOnMSDE = false,
        Mode="AutomaticMode"
    };

    string strFolderPath = "D:\\"; //LocalSettings.ProjectFileBasePath;
    string strFileName = strFolderPath + file.ProjectFileName;

    try
    {
        using (FileStream fs = new FileStream(strFileName, FileMode.Create))
        {
            DataContractSerializer serializer = new DataContractSerializer(typeof(ProjectFile));
            using (XmlDictionaryWriter writer = XmlDictionaryWriter.CreateTextWriter(fs))
            {
                serializer.WriteObject(writer, file);
            }
        }
    }
    catch (Exception ex)
    { 
        //TODO: Log the exception
        throw ex;
        return false;
    }

    return true;
}

 

 

 

Figure: You can see on lines 9 and 18 that there are calls being made to specific folders and disks.

What is wrong with this code? What assumptions mistakes could the developer have made to make this look OK:

  1. That every install would be to “C:\Program Files\SSW SQL Deploy”
  2. That every computer would have a “D:\\”
  3. That checking in code at 6pm because the had to go home was a good idea.

lets solve each of these problems:

  1. We are in a web service… lets store data within the web root. So we can call “Server.MapPath(“~/App_Data/SSW SQL Deploy\SampleData”) instead.
  2. Never reference an explicit path. If you need some storage for your application use IsolatedStorage.
  3. Shelve your code instead.

What else could have been done?

  1. Code review before check-in – The developer should have shelved their code and asked another dev to look at it.
  2. Use Defensive programming – Make sure that any code that has the possibility of failing has checks.

Any more options?

Let me know and I will add them.

What do we do?

The correct things to do is to add a Bug to the backlog, but as this is probably going to be fixed in sprint, I will add it directly to the sprint backlog.

  1. Right click on the failing test Select “Create Work Item | Bug”
    image
    Figure: Create an associated bug to add to the backlog.
     
  2. Set the values for the Bug making sure that it goes into the right sprint and Area. Make your steps to reproduce as explicit as possible, but “See test” is valid under these circumstances.
     image
    Figure: Add it to the correct Area and set the Iteration to the Area name or the Sprint if you think it will be fixed in Sprint and make sure you bring it up at the next Scrum Meeting.
    Note: make sure you leave the “Assigned To” field blank as in Scrum team members sign up for work, you do not give it to them. The developer who broke the test will most likely either sign up for the bug, or say that they are stuck and need help.
    Note: Visual Studio has taken care of associating the failing test with the Bug.
     
  3. Save…
     

Need Help?

www.SSW.com.au

SSW was the first company in the world outside of Microsoft to deploy Visual Studio 2010 Team Foundation Server to production, not once, but twice.

Team Foundation Server

Visual Studi2010 ALM SSW provides expert Visual Studio ALM guidance including installation, configuration and customisation through our four Microsoft Visual Studio ALM MVP’s in three countries; Australia, Beijing and the UK. They have experience deploying to small development shops all the way through to large blue chips.

Professional Scrum Developer Training

Professional Scrum Developer Training SSW has six Professional Scrum Developer Trainers who specialise in training your developers in implementing Scrum with Microsoft's Visual Studio ALM tools.

Technorati Tags:           

Print | posted on Thursday, March 04, 2010 5:48 PM | Filed Under [ .NET WCF TFBS Testing Design Developing SSW Scrum VS 2010 TFS ]

Feedback

Gravatar

# re: Finding the problem on a partially succeeded build on Team Foundation Build Services 2010

there is no way to accomplish that without writing some code. Here is a sample console application that I created for this purpose. It allows you to see the list of registered templates, add your own, and even change which ones are the default and upgrade templates there can be only one default and one upgrade template per team project.
8/30/2010 8:37 AM | best casino games online
Post A Comment
Title:
Name:
Email:
Website:
Comment:
Verification:
 
 

Powered by: