Joey Beninghove

Git/Branch-Per-Feature - Composable Deployments/Releases?

2010-03-09T05:27:00Z

(composed in vim with love)

Now that I’m starting to use Git a lot more, I’ve been thinking about and starting to use branch-per-feature a little bit. This morning I had a term pop into my head as I was thinking about branch-per-feature: “Composable Deployments/Releases”. Now I’m still getting the hang of this whole branch-per-feature thing, so this might be old hat for a lot of you, but I noticed something in some of my recent work with Git and how my workflow was starting to shape up. Here is an example of how using Git and Branch-Per-Feature is starting to make my life easier.

Disclaimer

I’ve only been using Git for a couple weeks, so go easy on me... ;)

Life With Subversion

I manage my church’s web site and have used subversion for managing the code for a long time. As I’m working on new projects for this web site, I deploy them to a “staging” site that is used to review and test out new changes with our Pastors and ministry leaders. In the meantime, small changes and content updates get done and deployed to “production” in parallel to these new, larger projects that get pushed to “staging”. One of the issues I’ve always had is how to easily work on these new projects separately and deploy new features in isolation from one another. Using the typical trunk/branches style of subversion, I’ve always done most work directly in trunk and sometimes creating branches for larger projects. Unfortunately trying to do this with subversion is Not Fun(TM). Also this made it tough to separate “staging” deployments from “production” ones. Of course I could have managed environment-specific branches in subversion and done the merging dance with subversion, but again, Not Fun(TM).

Enter Git

I’ve been using Git for a couple weeks locally to manage local branches for my “day job”, where we still use subversion as our main VCS (for now...muwhahaha :)). But for my church site I decided to make the full move to use Git with Beanstalk [http://beanstalkapp.com], which is where I’ve been doing my subversion hosting for a while now. Beanstalk just recently added Git support and so far it’s working out great. One of the first things I noticed as I started using Git was how easy it is to branch and merge code. So I figured I’d try out this whole Branch-Per-Feature [http://www.lostechies.com/blogs/derickbailey/archive/2009/07/15/branch-per-feature-source-control-introduction.aspx] thing all the cool kids are talking about these days. Needless to say I’m liking it a lot.

Ok, so getting to the point of this post, here is a peek at the branches I’m managing right now:

master (gets deployed to production)
staging (environment branch used to push to a staging site)
faith (feature branch for a new statement of faith page)
visitors (feature branch for a new visitors page)
<insert other temporary branches here for one-off changes that get merged into master and deployed>

One of the things I *really* like about this is that I can keep work nice and isolated in feature branches and easily merge them as needed into the appropriate environment-specific branch (staging or master) when it’s ready to be deployed. My current workflow is something like this:

(Note the use of Git aliases [http://git.wiki.kernel.org/index.php/Aliases], which I’d highly recommend)

git co faith
...do some changes...
git add .
git ci -m “blah blah”
git co staging
git merge faith
test changes in staging branch and deploy to staging environment

The way I’m treating the staging branch is mainly as a “merge and deploy” branch. I’m not usually making any direct changes in the staging branch. So, in theory, once the feature in the faith branch is tested and ready for to be deployed to production, I could just merge that feature branch straight into master and then do my normal production deployment. This seems like a really nice way to work. That’s why I called it “composable deployments/releases” because it’s really nice to be able to “compose” a deployment simply by merging in the appropriate feature branches as needed.

I’m still a Git n00b for the most part, so I’d be very interested in your feedback and improvements. I’m sure my thoughts on this will change the more I learn and use Git, but so far this is really working for me quite well.

Guest Post: Suggestions for Effective Team Communication in a Virtual Environment

2009-10-09T15:27:37Z

Note: Below is a guest post from Larry Silverman, Chief Technology Officer, TrackAbout. --Joey

Preamble

Thank you for the opportunity to contribute to the Los Techies community. This site is a great resource, and I've enjoyed reading many articles published here in the past. I'm honored to have the opportunity to contribute something back.

I wrote the following article for my development team and it appears here exactly as it does on our internal dev wiki. I mention and link to several third party products by name, and I feel compelled to say we have no stake in these products other than that we've found them to be useful. We are constantly changing the mix of tools we use to better facilitate communication, and it's entirely likely that we will have moved onto other tools by the time you read this.

I welcome any feedback and lessons you may have learned on your virtual teams.

Thanks,

Larry Silverman
Chief Technology Officer
TrackAbout, Inc.
www.trackabout.com

Overview

Our development team is completely virtual, with no two developers in the same physical location. This poses some obvious communication challenges. Luckily, we have many excellent tools at our disposal, such as IRC, IM, Skype + VOIP, mobile phones, landlines, remote desktop tools, email, and so on. Here are some suggestions and rules of thumb we have learned over the years for effective communication in this kind of environment.

As you read this, no doubt you'll think to yourself, "well, duh". These suggestions may seem rather obvious. But recognizing them and practicing them are two different beasts.

Be aware of the cost of distractions

As technologists and developers, we tend to be at the forefront of the "multi-tasking" movement. Unsfortunately, and somewhat counter-intuitively, some recent studies (nytimes.com) show that multi-taskers are considerably less productive, not more. Distractions and multi-tasking can heavily impact our productivity. We should strive to be vigilant and guard against being derailed in our daily work.

A good rule of thumb is to batch your questions for others whenever possible. I learned this rule the hard way, as a green consultant on my first job out of college. After my fourth visit to my manager's office in a single morning, my exasperated manager finally exclaimed, "Holy freaking ****, would you PLEASE batch these questions?!" Whoops.

It's quick and easy to IM someone when you need an answer to a question, and we've become quite accustomed (addicted?) to the instant gratification that IM and texting provides. But, if you need to ask a co-worker a question, stop and consider whether you might have more related questions in the near future. Can you proceed in the meantime without an immediate answer? If you batch several questions together and ask them all at once, you'll have done the askee a favor, as opposed to peppering him or her with questions over time.

Consider emailing questions rather than IMing if appropriate to allow the recipient to answer the question on his/her own timetable. That having been said, it may very well be that you need a particular answer ASAP, so use your judgment.

Use the right tool for the job

We use IRC. We have a channel for #dev and a channel for #support. All the developers sit on both channels, all day. Our IRC channels are like our virtual water cooler, and we do occasionally like to amuse ourselves and our co-workers there. It can keep things fun.

An IRC channel is a broadcast medium, so it's great for when everyone needs to hear what you have to say. It's not so great when you need to address only one person, and it's horrible when you distract everyone else on the channel when intending to speak to just one person. When you need to address just one person, private message them or use a different IM system instead.

Consider configuring your IRC client such that you can quiet it down when you need to get work done. You might, for example, configure your IRC client so that it's quiet when minimized, but still chirps or does a toaster popup when your handle or specific keywords are spoken. That way you can get work done and not miss anything important.

If you're having a hard time explaining a concept, there are dozens of free online collaboration tools available at your disposal.

    * For remote desktop sharing, we used VNC in the past, but have switched over to Microsoft SharedView recently, leaving VNC only as a backup.
    * We use Google Apps at our company, which provides for real-time collaboration on documents and spreadsheets.
    * Try an online white-boarding session
    * Collaborate on a screen or page mockup using the excellent Balsamiq Mockups
    * Flowchart something using Gliffy, Visio or Powerpoint over remote desktop.
    * A couple of us recently picked up some inexpensive Wacom tablets to try out picture drawing as a quick communication tool.
    * You can draw picture on piece of paper and scan it, or take a picture with your cellphone, and upload it to a disposable file service like imgur.com or drop.io.
    * You can paste a bit of code to pastebin for discussion.

Consider security

When you do need to transmit sensitive information, do it securely using an encrypted channel. For IM, if you are a Pidgin user, consider enabling the Pidgin encryption plug-in or installing the separate Off-the-Record messaging plug-in. Hamachi chats are encrypted, as are Skype chats. Never send passwords in the clear or through email.

Know when to escalate

Don't hesitate to escalate from IRC/IM to a voice channel to save time. If the IM starts to get too long (or your hands are cramping), just switch to voice and knock it out. A simple "skype?" says it all.

Be conscious that misunderstandings can blossom seemingly from nowhere in chats and emails. If you sense you are being misunderstood or a conflict might be brewing, switch to voice immediately. Humor and wit are especially easy to misunderstand in text.

Don't be afraid to declare some "heads down" time and "go dark"

If you need to focus, it's ok to "go dark". By this I mean quieting, disabling or shutting down services that can be distracting and only checking them periodically. Some of us have production monitoring responsibilities, so we can never really go totally "dark". But for others, it is an option that can afford some much needed focused development time.

If you declare you're going dark, then everyone knows they can get you via phone if absolutely necessary.

Don't let your IM client "lie"

The idle status features of IM clients are useful to us, so leave them enabled. That is, don't configure your IM client so that it always looks like you're active. It's OK to be idle; hey, we all go to the bathroom, or take some time to read a couple chapters of a book, or whatever. What's frustrating is needing an answer from someone, seeing their IM says "active", pinging them, then sitting there waiting and wondering where they are. In a physical work environment, if someone dropped by your office, they'd be able to tell immediately if you are idle or active just by looking. Let's use the idle status to mimic that in our virtual environment. Consider reducing the amount of time your client waits before setting you "idle" to just a few minutes.

Skype Tips

    * If you are asking someone else to Skype via IM and they agree, then you initiate the call. Following this simple protocol eliminates the confusion regarding who is supposed to start a call.
    * Use a USB or Bluetooth headset, not a headset that plugs into your computer's audio in/out jacks. Those in/out jacks are usually very noisy as their analog-to-digital converters are sitting inside your machine, amongst a maelstrom of other electromagnetic activity. It's also a pain to be constantly plugging and unplugging your headset or your regular speaker system. Bluetooth headsets are fine as long as they don't introduce interference and noise.
    * If your webcam has an integrated microphone, make sure it's unidirectional and won't pick up sound coming out of your speakers. Feedback and echoes suck. A headset is likely the better choice.
    * Map those hotkeys! By default, Skype's hotkeys are not enabled. I turn them on so that Alt-Up picks up a call and Alt-Down hangs up. Additionally, I map a key combo (for me, it's Ctrl-Alt-Z) to control muting on/off.
    * Try to remember to mute when you're not speaking. Some of us type REALLY LOUDLY, or have squeaky chairs, or live in creaky houses with creaky floorboards. Or perhaps you live on a busy street and like to have the windows open. Or, you've set up shop in an aviary. Other peoples' noise is distracting, and if your noise reaches a certain level, it can cause others to be cut off.
    * Don't breathe into the microphone. Just ... don't. Move the microphone away from your line of breath. Please.
    * Keep cell phones away from headset wires. There's a tell-tale sound when someone's iPhone is resting near a headset wire, and everyone on the call can hear it.
    * For that matter, be careful of any wires that are draped across your headset wire. Some devices and wires will induce noise. Not all headset wires are adequately shielded, and shielding can spread out inside the cable with bending over time.
    * If you don't have the horsepower, try not to run anything that will exercise your CPU while you're on Skype. Sometimes it's unavoidable, like when you're hosting a remote desktop session and a Skype simultaneously. However, kicking off a build and sharing a torrent while Skyping and remotely displaying your 1920x1080 screen has the potential to make you sound like you were sucked into an inter-dimensional wormhole and are being extruded through the event horizon of a black hole. In other words, it sounds kinda bad.

Our Protocols

Over time we've adopted a small handful of common protocols that we use daily:

IRC Protocols

brb = Be right back. You are stepping away from your desk for a little while. You might also change your nick to something like "yourhandle_away", but we don't remember to do that very often.

afk = Away from keyboard, see brb

back = Hey, you just came back!

IM Protocols

otp = On the phone. Someone messages you while you are talking on the phone. You can't respond in detail immediately, but you want to acknowledge their ping and you'll get back to them later.

Try to keep your IM statuses accurate.

"lunch", "brb", "otp", "in a meeting", "do not disturb" are all useful statuses.

Conclusion

The bottom line is one of consideration -- be considerate to your fellow co-workers, minimize distractions and time wasting, and try to think, "What would I do if I were in a physical office environment?" (WWIDIIWIAPOE™)

Smart Constructor Anti-Pattern

2009-09-02T04:46:00Z

If you look at the definition of a constructor in the context of software development on Wikipedia (a completely factual and reliable source :), you'll find this simple statement of their responsibility...

"Their responsibility is to initialize the object's data members and to establish the invariant of the class, failing if the invariant isn't valid."

Simple really. Just initialize the (usually) private data members and ensure the arguments won't put the object in an invalid state. But so often I see constructors doing so much more than that. One example I've seen of this recently is in presenters.

The (perhaps contrived) scenario:
When a view loads for the first time, you need to load some lookup data that gets populated in a combo box on the view and set the number of logged in users on the view.

Here is our IView interface for reference:

        1: public interface IView

       2: {

       3:     void BindWith(IEnumerable<User> users);

       4:     bool HasUserSelected { get; }

       5:     void SetNumberOfLoggedInUsersTo(int numberOfUsers);

       6: }

“Smart” Constructor Approach

One might be tempted to start writing a spec like the one below.

       1: public class SmartConstructorPresenterSpecs

       2: {

       3:     public class when_the_presenter_is_constructed : ContextSpecification

       4:     {

       5:         private IView View;

       6:         private IUserRepository UserRepository;

       7:         private IAuthenticationService AuthenticationService;

       8:         private SmartConstructorPresenter SUT;

       9:         private IEnumerable<User> Users;

  

        protected override void When()

        {

            Users = new List<User> {new User(true), new User(true), new User(false)};

  

            View = Mock<IView>();

            UserRepository = Mock<IUserRepository>();

            AuthenticationService = Mock<IAuthenticationService>();

  

            UserRepository.Stub(x => x.All()).Return(Users);

  

            SUT = new SmartConstructorPresenter(View, UserRepository, AuthenticationService);

        }

  

        [Test]

        public void it_should_populate_the_view_with_a_list_of_users()

        {

            View.AssertWasCalled(x => x.BindWith(Users));

        }

  

        [Test]

        public void it_should_set_the_number_of_logged_in_users_on_the_view()

        {

            View.AssertWasCalled(x => x.SetNumberOfLoggedInUsersTo(2));

        }

    }

  

    public class when_attempting_to_login_and_a_user_is_not_selected : ContextSpecification

    {

        private IView View;

        private IUserRepository UserRepository;

        private IAuthenticationService AuthenticationService;

        private SmartConstructorPresenter SUT;

        private IEnumerable<User> Users;

  

        protected override void EstablishContext()

        {

            Users = new List<User> {new User(true), new User(true), new User(false)};

  

            View = Mock<IView>();

            UserRepository = Mock<IUserRepository>();

            AuthenticationService = Mock<IAuthenticationService>();

  

            UserRepository.Stub(x => x.All()).Return(Users);

            View.Stub(x => x.HasUserSelected).Return(false);

  

            SUT = new SmartConstructorPresenter(View, UserRepository, AuthenticationService);

        }

  

        protected override void When()

        {

            SUT.Login();

        }

  

        [Test]

        public void it_should_NOT_perform_a_login()

        {

            AuthenticationService.AssertWasNotCalled(x => x.Login());

        }

    }

}

And implement the presenter.

       1: public class SmartConstructorPresenter

       2: {

       3:     private readonly IView view;

       4:     private readonly IAuthenticationService authenticationService;

       5:  

       6:     public SmartConstructorPresenter(IView view, IUserRepository userRepository,

       7:                                      IAuthenticationService authenticationService)

       8:     {

       9:         this.view = view;

      10:         this.authenticationService = authenticationService;

      11:  

      12:         var users = userRepository.All();

      13:  

      14:         this.view.BindWith(users);

      15:         this.view.SetNumberOfLoggedInUsersTo(users.Where(x => x.IsLoggedIn).Count());

      16:     }

      17:  

      18:     public void Login()

      19:     {

      20:         if (view.HasUserSelected)

      21:             authenticationService.Login();

      22:     }

      23: }

And then the view.

       1: public class View : IView

       2: {

       3:     private event VoidHandler loginButtonClick = delegate { };

       4:     private readonly SmartConstructorPresenter presenter;

       5:  

       6:     public View()

       7:     {

       8:         presenter = new SmartConstructorPresenter(this, null, null);

       9:  

        // simulate a possible login button click

        loginButtonClick += () => presenter.Login();

    }

  

    public void BindWith(IEnumerable<User> users)

    {

        // bind this stuff to something

    }

  

    public bool HasUserSelected

    {

        get { return false; }

    }

  

    public void SetNumberOfLoggedInUsersTo(int numberOfUsers)

    {

        // set some label or something

    }

}

  

public delegate void VoidHandler();

If you take a look at the code above, you can see that a the view is being populated with a list of users when the presenter is constructed. While this might seem harmless at first, take a look at the second specification : when_attempting_to_login_and_a_user_is_not_selected.it_should_NOT_perform_a_login.

Notice that the only thing we care about is verifying that when a user is not selected, that we’re NOT attempting to call login on the authentication service. But since our constructor has some “smarts”, we’re required to stub out the call to the user repository just to keep the test from breaking. This may not look like a big deal in a simple case like this. But as your codebase grows, it can make your specs brittle and harder to understand.

“Dumb” Constructor Approach

Another way this could be done is by delegating the initial operations on the view to an explicit “initialize” method. Let’s see what that might look like.

First the specs of course. :)

       1: public class DumbConstructorPresenterSpecs

       2: {

       3:     public class when_initializing : ContextSpecification

       4:     {

       5:         private IView View;

       6:         private IUserRepository UserRepository;

       7:         private IAuthenticationService AuthenticationService;

       8:         private DumbConstructorPresenter SUT;

       9:         private IEnumerable<User> Users;

  

        protected override void EstablishContext()

        {

            View = Mock<IView>();

            UserRepository = Mock<IUserRepository>();

            AuthenticationService = Mock<IAuthenticationService>();

  

            SUT = new DumbConstructorPresenter(View, UserRepository, AuthenticationService);

        }

  

        protected override void When()

        {

            Users = new List<User> {new User(true), new User(true), new User(false)};

  

            UserRepository.Stub(x => x.All()).Return(Users);

  

            SUT.Initialize();

        }

  

        [Test]

        public void it_should_populate_the_view_with_a_list_of_users()

        {

            View.AssertWasCalled(x => x.BindWith(Users));

        }

  

        [Test]

        public void it_should_set_the_number_of_logged_in_users_on_the_view()

        {

            View.AssertWasCalled(x => x.SetNumberOfLoggedInUsersTo(2));

        }

    }

  

    public class when_attempting_to_login_and_a_user_is_not_selected : ContextSpecification

    {

        private IView View;

        private IUserRepository UserRepository;

        private IAuthenticationService AuthenticationService;

        private DumbConstructorPresenter SUT;

  

        protected override void EstablishContext()

        {

            View = Mock<IView>();

            UserRepository = Mock<IUserRepository>();

            AuthenticationService = Mock<IAuthenticationService>();

  

            SUT = new DumbConstructorPresenter(View, UserRepository, AuthenticationService);

        }

  

        protected override void When()

        {

            View.Stub(x => x.HasUserSelected).Return(false);

  

            SUT.Login();

        }

  

        [Test]

        public void it_should_NOT_perform_a_login()

        {

            AuthenticationService.AssertWasNotCalled(x => x.Login());

        }

    }

}

Then the implementation of the presenter.

       1: public class DumbConstructorPresenter

       2: {

       3:     private readonly IView view;

       4:     private readonly IUserRepository userRepository;

       5:     private readonly IAuthenticationService authenticationService;

       6:  

       7:     public DumbConstructorPresenter(IView view, IUserRepository userRepository,

       8:                                     IAuthenticationService authenticationService)

       9:     {

      10:         this.view = view;

      11:         this.userRepository = userRepository;

      12:         this.authenticationService = authenticationService;

      13:     }

      14:  

      15:     public void Initialize()

      16:     {

      17:         var users = userRepository.All();

      18:  

      19:         view.BindWith(users);

      20:         view.SetNumberOfLoggedInUsersTo(users.Where(x => x.IsLoggedIn).Count());

      21:     }

      22:  

      23:     public void Login()

      24:     {

      25:         if (view.HasUserSelected)

      26:             authenticationService.Login();

      27:     }

      28: }

And a slight change to the view implementation.

       1: public class View : IView

       2: {

       3:     private event VoidHandler loginButtonClick = delegate { };

       4:     private event VoidHandler load = delegate { };

       5:     private readonly DumbConstructorPresenter presenter;

       6:  

       7:     public View()

       8:     {

       9:         presenter = new DumbConstructorPresenter(this, null, null);

  

        // simulate a couple possible events

        load += () => presenter.Initialize();

        loginButtonClick += () => presenter.Login();

    }

  

    public void BindWith(IEnumerable<User> users)

    {

        // bind this stuff to something

    }

  

    public bool HasUserSelected

    {

        get { return false; }

    }

  

    public void SetNumberOfLoggedInUsersTo(int numberOfUsers)

    {

        // set some label or something

    }

}

  

public delegate void VoidHandler();

I would prefer this approach since I think it keeps the specs and the presenter implementation itself quite a bit cleaner. Notice now in the spec: when_attempting_to_login_and_a_user_is_not_selected.it_should_NOT_perform_a_login.

It cares nothing about retrieving users from the repository. All it cares about is exactly what the context/spec sentence states. And if you notice, there is even an opportunity to DRY up these specs, if you prefer, by breaking out a base context class, since they both have identical EstablishContext() methods. This is something that wouldn’t be quite as easy or clean to do in the SmartConstructorPresenterSpecs in the previous section.

Quick Tip : Load/Initialize Convention

You might be thinking that you don’t want to have to remember to create or call an Initialize() method all the time. Well recently, a co-worker of mine pointed out a simple way we could leverage some auto-wiring magic + a base presenter class to make it more of a convention that you can use when needed and to negate the need to manually call Initialize() all the time. So here is one approach to make that easier.

NOTE: The presenter specs don’t need to change at all for this refactoring.

First we add a Load event handler to our IView interface:

       1: public interface IView

       2: {

       3:     void BindWith(IEnumerable<User> users);

       4:     bool HasUserSelected { get; }

       5:     void SetNumberOfLoggedInUsersTo(int numberOfUsers);

       6:     event EventHandler Load;

       7: }

Then we introduce a base presenter class (which I’m simply calling Presenter ‘cause I think Base* class naming is the new hungarian notation…. :).

       1: public class Presenter<T> where T : IView

       2: {

       3:     public Presenter(T view)

       4:     {

       5:         view.Load += delegate { Initialize(); };

       6:     }

       7:  

       8:     public virtual void Initialize()

       9:     {

      10:     }

      11: }

Then our DumbConstructorPresenter just changes slightly to inherit and override the Initialize() method.

       1: public class DumbConstructorPresenter : Presenter<IView>

       2: {

       3:     private readonly IView view;

       4:     private readonly IUserRepository userRepository;

       5:     private readonly IAuthenticationService authenticationService;

       6:  

       7:     public DumbConstructorPresenter(IView view, IUserRepository userRepository,

       8:                                     IAuthenticationService authenticationService) : base(view)

       9:     {

      10:         this.view = view;

      11:         this.userRepository = userRepository;

      12:         this.authenticationService = authenticationService;

      13:     }

      14:  

      15:     public override void Initialize()

      16:     {

      17:         var users = userRepository.All();

      18:  

      19:         view.BindWith(users);

      20:         view.SetNumberOfLoggedInUsersTo(users.Where(x => x.IsLoggedIn).Count());

      21:     }

      22:  

      23:     public void Login()

      24:     {

      25:         if (view.HasUserSelected)

      26:             authenticationService.Login();

      27:     }

      28: }

And finally our view implementation.

       1: public class View : IView

       2: {

       3:     // in the context of a winforms app, 

       4:     // this event would be automatically declared and fired

       5:     public event EventHandler Load = delegate { };

       6:  

       7:     public event VoidHandler LoginButtonClick = delegate { };

       8:     private readonly DumbConstructorPresenter presenter;

       9:  

    public View()

    {

        presenter = new DumbConstructorPresenter(this, null, null);

  

        // simulate a possible login button click

        LoginButtonClick += () => presenter.Login();

    }

  

    public void BindWith(IEnumerable<User> users)

    {

        // bind this stuff to something

    }

  

    public bool HasUserSelected

    {

        get { return false; }

    }

  

    public void SetNumberOfLoggedInUsersTo(int numberOfUsers)

    {

        // set some label or something

    }

}

Notice in the Presenter<T> class above, we’re simply wiring up the Load event from the IView to automatically call the Initialize() method for us. Concrete presenters could then simply just override the Initialize() method if necessary.

Then notice in the updated view implementation, we don’t have explicitly call the Initialize() method anymore in our views, since that’s done for us automatically in the Presenter<T> class.

As I’m writing this I realize that the act of overriding this Initialize() method would have to be remembered as well. But I think what you gain from a convention standpoint and the fact that you don’t have to explicitly call it from the views is enough of a win to warrant this technique. But as always, I’ll probably change my mind tomorrow when I figure out a better way to do all of this. :-)

So, putting aside the possibly poor examples and the cloud of MVP in all of this, my goal was just to simply give an example of why it can be a good idea to keep your constructors simple and lean across the board, in many different types of classes.

Of course feedback is always welcome.

NHibernate + XML Columns

2009-01-14T04:42:05Z

One of the things I’ve been working on recently involves using XML columns in SQL Server. Starting out, it was simple and I was just doing vanilla ADO.NET (wrapped in a simple Query API) combined with XML serialization/deserialization, which worked pretty well for a while.

But as the complexity has grown, it seemed like too much time was being spent enhancing the persistence infrastructure in this particular area of the application. In comes NHibernate, which is already integrated and available to me on this particular project. In fact, the only reason I didn’t use NHibernate for this particular feature from day one is because I didn’t see a lot of information available regarding NHibernate and XML columns. I did find one old blog post by Ayende and a seemingly outdated article on the NHibernate site. But I admit I’ve never jumped into creating custom user types in NHibernate and wasn’t yet comfortable moving forward with that approach. Since what I needed at the time was pretty simple, I went forward without NHibernate for the time being.

Without going into too much detail, I came to a point where I wanted to spike with NHibernate to see how it handles columns with an XML data type. I’ve only tried one of a couple approaches so far, but wanted to get some feedback on it so far.

First, a couple goals:

Store a set of data as XML in a SQL Server XML column
Ability to deserialize the XML into strongly typed objects for use in the rest of the code base

Warning: contrived example ahead. The real implementation is basically for lightweight messages.

        1: public class Person

       2: {

       3:     private readonly string contactInformationXml;

       4:  

       5:     public Person(ContactInformation contactInformation)

       6:     {

       7:         // NOTE: SerializeToXmlStream extension method not shown

       8:         contactInformationXml =

       9:             new StreamReader(contactInformation.SerializeToXmlStream()).ReadToEnd();

      10:     }

      11:  

      12:     public ContactInformation GetContactInformation()

      13:     {

      14:         var xmlDocument = new XmlDocument();

      15:         xmlDocument.Load(contactInformationXml);

      16:  

      17:         // NOTE: DeserializeInto extension method not shown

      18:         return xmlDocument.DeserializeInto<ContactInformation>();

      19:     }

      20: }

And an excerpt from an example NHibernate mapping for this:

       1: <property name="ContactInformationXml" column="ContactInformation" 

       2:           type="String" not-null="true" access="field.camelcase" />

So in the Person class above, the constructor accepts a strongly typed component for the contact information, serializes it to XML and stores it in a private field as a string. Then the NHibernate mapping takes care of persisting the serialized string to the XML column named ContactInformation in the database. To see how it would get used when NHibernate loads a Person, the GetContactInformation() method is an example of deserializing the string into the strongly typed ContactInformation object which it then returns.

Now, putting aside the reasons for or against using XML in this way, using XML columns in general or the fact that serialization concerns shouldn’t be placed inside a class like this…

I’m looking for feedback on this approach and if anyone else has any better ways of doing this. I’ve yet to go down the path of creating a custom NHibernate user type, even though I think doing it that way would be a bit cleaner and flexible in the future.

Anyone have any other good examples of using NHibernate for persisting data to XML columns?

Looking for some awesome .NET developers (work from home)

2008-09-26T17:49:22Z

Are you:

an experienced and passionate .NET developer
looking to join a profitable and growing company who creates software solutions for a niche industry that is booming despite the current economy
wanting to work from home

Excerpts from our job posting:

Company:
TrackAbout, Inc.
www.trackabout.com

TrackAbout is a small, profitable and growing software company that provides Software as a Service (SaaS) asset tracking services through the combination of a hosted web platform and various mobile devices that run our custom software. We have a solid, international customer base, are experiencing high demand for our services and are looking at exciting times ahead. We are looking for an exceptional and experienced developer who is interested in playing a pivotal role in the growth of a small company. If you’re truly passionate about technology, we want to talk to you.

Location:
TrackAbout's main office is just outside of Pittsburgh, PA. However, our development staff is completely virtual (no office) and each developer works from home.

Position Details:
We're seeking a full-time developer (no contractors please) with a minimum of 5 years of experience in web development, object-oriented programming and relational database programming. Agile experience is a plus. You'll be working on all aspects of the TrackAbout offering including our ASP.NET website and web services, MS SQL Server 2005 and various wireless mobile devices.

Your responsibilities will include architecting, designing and building new features, enhancing existing features, and fixing bugs. On occasion, you may be asked to provide second-tier technical customer support for requests that require deeper technical skills than our first-tier support engineers can handle.

All members of our development team are dedicated to keeping up with the latest techniques, technologies and tools related to software development. We take time out to share our knowledge with one another and improve both our skills and our distributed development environment. We often experiment with and integrate new tools into our environment. We use ReSharper, CruiseControl.NET, TDD with NUnit, FXCop, NCover, Bugzilla, and Drupal to supplement our development efforts.

So if you're interested and want to come help me sling some code, check out the full job posting and reply there.

MVC - Free Yourself from Web Forms!

2008-09-05T13:13:41Z

Thanks to all who came out last night to hear my talk on MVC at the Richmond .NET User Group. It was a blast. I wish we had more time to dig into some more code, but maybe next time. 90 minutes flies by!

As promised, here are the slides from the presentation.

How did I get started in software development?

2008-06-27T04:06:58Z

Well apparently Derik is trying to bring me out of my blogging hiatus by tagging me with the meme of the week. :) My story is pretty boring, but I'll play along in an effort to bring me out of my recent break from the blogosphere.

How old were you when you started programming?

I suppose I was 14 (9th grade) when I first started dabbling on an Apple IIe with all the joys of line numbers and "gotos". I enjoyed it but I definitely wasn't as smart as the kid who created a program to generate one of those "magic eye" pictures that were all the rage at the time. Besides, I was a band geek for most of my high school days (drums).

What was your first language?

I guess technically Apple Basic and QBasic were my first languages where I would mess with my Mom by writing stupid little QBasic programs to simulate her computer exploding. Ahh, those were the days. But I would probably say that my Pascal class in 11th grade was where I really started to learn how to create stuff and solve problems by typing "codes" into some crazy thing called a "computer".

What was the first real program you wrote?

Probably the first "real" program I ever wrote was a Pick 3 lottery program in my Pascal class followed later by a standard 5 card draw poker game, with stunning graphics! :) Looking back, I'm not sure if those were assigned projects or if I came up with those on my own, but regardless, either the school was encouraging the students to write gambling software or I had some weird attraction to writing software that could possibly ruin people's lives. Both equally disturbing... LOL.

If you knew then what you know now, would you have started programming?

Oh yeah. I actually started my "professional" career on the hardware/networking side of things but ultimately I always knew I was a codemonkey at heart and eventually moved in that direction professionally. Besides, the money is a heck of a lot better too. :)

If there is one thing you learned along the way that you would tell new developers, what would it be?

So, what's my size limit for this answer? :D

Well of course I'd say to have fun and be passionate and all that jazz. To me, that's just a given. But also to focus on the fundamentals. Learn *why* it's important to make testability and maintainability a first-level concern when building software. Get started in the community early because you'll make some great friends and learn a ton along the way. I could say a ton more, but I don't think the internet could hold them all.

What's the most fun you've ever had ... programming?

Not sure I can pinpoint a single time, but the times I've been able to do some greenfield projects and make testability a first-class citizen in the design, I've been very happy with the results. Also the journey towards "beautiful code" is something that I enjoy very much. Getting creative and coming up with new and useful ways of wiring something together or even just naming a method well is enough to make me happy.

Recently, I'd say that my brief stint in the Rails world (not professionally) was the most fun I've had coding in quite a while. Perhaps that's why I'm getting the Ruby itch again.

The poor saps that I'm tagging?

Change I can achieve in

2008-03-05T02:21:10Z

Well, this post is more of the personal variety, but it may affect future posts in one way or another (however, not like this). In short, a season of change is upon me.

Big change #1. I'm getting out of the consulting world for a while. It's been grand and I've learned a ton about people, business and technology. But I felt it was time for something different. Because of the types of consulting firms I've worked with the past 3 years, the one thing I haven't liked is that I never got to work on a team long enough to really get to know the folks I was working with very well. Seems like just when my team and I were really getting productive together, we'd all get thrown to different projects, usually with folks we hadn't worked with before. Obviously some of this has to do with the nature of consulting in general, but I'm sure some firms see the value in keeping teams together as much as possible than others.

Big change #2. Like Chad, I also started a new job this week, for a great product company. The company is small, which I really like, especially when talking about small development teams. I'm hoping to learn a lot (drinking from the firehose this week) and be a good contributor along side the other couple developers on the team. It's an interesting product with some pretty unique challenges to tackle, from what I've seen so far. I'll be doing mostly ASP.NET work, with maybe a little mobile device stuff thrown in for good measure.

Big change #3. I'm joining the world of web workers! Yep, I'm now working from home exclusively. This might be the biggest change for me in all of this, but I'm very much looking forward to adapting to this type of lifestyle/work environment. I'm sure I'll have some things to learn, specifically about how to keep things balanced, but so far it's been enjoyable. Not to mention, it'll help me kick the "eating out" habit that I was stuck in during my traditional office jobs. :O

So far I'm finding that my reliance on tools like IM, Skype and SharedView (which is pretty decent) just grew exponentially. I'm looking forward to possibly trying out some remote pairing as soon as I get settled.

To make this post at least somewhat productive, here are a couple tools and a bonus tip that I've been introduced to over the past couple days that you may find useful.

Pidgin-Encryption (for...umm...encryption)
Hamachi (really nice and free VPN software)

Bonus Tip: I've been a huge Pidgin fan ever since it had that weird Gaim name. I use it for everything. AIM, Yahoo, GTalk, IRC, Twitter. One tip that I learned today is how to set up Pidgin to "auto join" you to a particular IRC channel as soon as you login into Pidgin. Here's how to do it...

Click Buddies > Add Chat
Choose your IRC account/server from the Account drop-down
Type the name of the channel you want to join
Give the chat an alias (I just used the channel name)
Click Add
Now find that new chat in your list of buddies, Right-Click > Auto-join
That's it!

So now when you fire up Pidgin, it'll automatically join you to that room. Most of you may already know about this, but I thought it was pretty cool.

Anyway, until next time. I have a lot of things to blog about in the hopper. Hopefully I can get around to them sometime soon. *sigh*

Scribbish Skin - For Subtext

2008-02-11T02:20:04Z

This is long overdue, but Nick Parker asked me a long time ago (sorry Nick) where I got the skin I use for my (joeyDotNet.com) blog and I told him I'd make it available. I actually ported it myself from the existing Scribbish theme that was already available for Mephisto, Wordpress and others. I really liked the simplicity of it so I just grabbed the CSS and hacked away at it myself until I got it looking close enough as a Subtext skin.

I submitted it to the Subtext skins showcase, but it may take a bit for it to get approved. So for now, you can just download it from here.

I kind of put this together in a hurry, so please let me know if you find any problems with it.

Enjoy! :)

ActiveSupport.NET - Namespace Overhaul

2008-02-09T04:01:45Z

Figured I'd try and get this nailed down as early as possible. So after some recent feedback, I decided to change the namespacing so that it focused more on "behavior/concern" rather than "types". What this means is...

Instead of:

using ActiveSupport.Core.Extensions.String; // containing string extensions for access, conversions, etc.
using ActiveSupport.Core.Extensions.Integer; // containing integer extensions for access, inflections, etc.
...

It's now just this:

using ActiveSupport; // just basic extensions
using ActiveSupport.Access // all accessor-based extensions for ALL types
using ActiveSupport.Conversions // all conversion-based extensions for ALL types
...

I'm pretty much thinking the accessor based stuff will probably get used the most. But either way, by grouping this way, I'm hoping they'll be a bit easier use. And of course when our R# 4.0 EAP shows up next (right Ilya? :) all of this namespace'ry will must magically work for us. :)

What do you think? Is this a better approach to keeping things clean and organized?

ActiveSupport.NET - Update

2008-02-08T17:59:17Z

So it seems there is some interest in getting something like this up and running. I went ahead and did some of the basic "project setup" type tasks. Here are some links of interest...

ActiveSupport.NET Google Code Project
ActiveSupport.NET.Discuss (general)
ActiveSupport.NET.Developers (for future use once we have some a group of committers)

I look forward to some more discussions about the best way to structure things to make it easy to use. On the google code project site you'll find a brief description of the project as well as a wiki page I've started to give some basic info on how the code base is structured (tools, build targets, etc...). I'm sure all of this will change/evolve as things progress.

And thanks Joe (for already submitting the first patch) and others that have said they want to get involved. Should be fun... :)

Introducing ActiveSupport.NET

2008-02-07T19:20:52Z

To my surprise, none of my googling seemed to come up with anyone doing this. So I figured I'd start it off (please let me know if you know of any other efforts already underway). As some would say, taking extension method abuse to a whole new level. :D But, I see these little utility methods to be a good fit as extensions methods to make the syntactic sugar that much sweeter.

One of the many great things about working in Rails is the core extensions in the ActiveSupport library. It just *becomes* part of the Ruby language as you work with Rails. And I wanted to get that same feeling when working in C#. So this is my attempt to start porting over the core extensions from Ruby's ActiveSupport Library.

So far I've just ported over a few string access and conversion extensions, but check 'em out and let me know if you'd find something like this to be useful. Here are some examples of what's currently supported...

 1: // some accessors 2: "blah".At(2) // 'a' 3: "blah".First() // 'b' 4: "blah".First(2) // "bl" 5: "blah".From(1) // "lah" 6: "blah".Last() // 'h' 7: "blah".Last(2) // "ah" 8: "blah".To(2) // "bla" 9:   10: // couple simple conversions (still needs lots of work) 11: "1/4/2008".ToDate() // DateTime 12: "01:15:35".ToTime() // DateTime

You can grab the source at its google code repository. You can just do a "build test" from a command line in the trunk to run the unit test suite.

Enjoy. :-)

CIFactory.Create(RichmondDNUG);

2008-01-11T02:32:09Z

Jay Flowers came down to our humble Richmond DNUG tonight to talk about Continuous Integration and his rockin' open-source product, CI Factory. It was a great demonstration of how to get a very robust automated build server up and running in literally minutes. Having set up quite a few automated build and CI environments myself over the years, I have a great appreciation for all the hard work that's gone into a product like CI Factory.

So I'm looking forward to leveraging CI Factory in my future CI server implementations (which will hopefully be soon!)

Oh and some of the cool stuff he's done in extending CCNet and NAnt was pretty impressive as well. Not to mention the extension of NCover to drive and generate code coverage for WatiN tests.

So if you need CI (and trust me, you do), then download CI Factory and check it out.

I Do What?

2008-01-06T05:25:14Z

Ok, so I know I'm crazy late on this one, but things have been a bit hectic recently. And I'm not even talking about holidays. But since our newly knighted member at LosTechies has chimed in, I guess it's about time I did too in response to Jason tagging me.

A Little History

Yeah, I know. I always hated history in school too, but bear with me. I'll just give a brief (hopefully) paragraph on the past couple years to help illustrate what I'm currently doing and why.

It's April '05 and I've got a really good corporate job as a lone software developer managing and building intranet-based ASP.NET web applications. Decent salary, great benefits, short commute, generally low-stress. Well, I had pretty much hit the "technical" glass ceiling (not wanting the "management" track) and had the desire to work with teams of other, more experienced, developers. So I took a pay cut to go work for a small, but well established software consulting firm in May '05. (BTW, make sure you always factor in health insurance costs when weighing a decision like this. That was stupid mistake #1 which ended up costing me dearly.)

So there I was in the consulting game and learning and growing by leaps and bounds. I had already long embraced OSS as a viable alternative to the Redmond standard. Which made for some interesting internal debates and experiences considering they were a "Microsoft Certified Managed Gold Partner". Anywho, I cranked away at various internal and external client projects for a couple years, getting to work with some great folks including my boys DP Bullington and Paul Laudeman. With their help, I tried to help introduce some agile engineering practices onto our projects (TDD, CI, DDD, etc.). We all spent a year on-site with a particular client trying to introduce Scrum on 4 projects simultaneously, which turned out to be quite a mess, and in my opinion, an overall failure. I mostly attribute that to poor leadership and lack of experience on ~~our~~ my part (only speaking for myself). Nevertheless, in the 2 years I worked at that consulting firm I learned a TON technically, but more importantly, personally in how to deal with people and some of the ins and outs of the software consulting business in general.

Ok, so maybe that wasn't as brief as I thought. Sorry.

Here And Now

So, ironically, in April of '07 I found myself once again at a kind of crossroads. After a year of spending ~2 hours a day on the road back and forth from the last client for which I worked, along with the fact that I was just generally not happy with how things were going, I accepted an offer from another, much larger, consulting firm. This new firm, which, among other things, had much better benefits, shorter commute and was a "hybrid" shop (unlike the previous firm, which was 1 Microsoft Way or the Highway). So overall, I felt pretty good about the move.

As I've continued down this path as a software consultant, I've grown more and more passionate about delivering quality, maintainable software to clients. I know, who would disagree with that, right? Well, in words, probably no one. But in deeds and practices, that's another story. Suffice it to say, I'm continuing to try and raise the bar, both for myself and for those I work with to ensure that we are serving our clients in the best way we know how. Which is by giving them solutions that benefit the business and that don't break and doing so without working crazy amounts of hours. Sometimes I wonder if I take my job as a consultant too seriously, which is perhaps due to my passion for the craft of building software.

What's Next?

Well, so far, I'm finding Richmond to always be playing catch up with some of the other markets like Austin, Calgary and Philly with regards to adopting a more lean development process and engineering practices. But this is where the Lord has us right now and that's all I need to know. I am looking to possibly start accepting some presentation opportunities, first internally at my current firm and also at Code Camps, DNUG meetings and screencasts. I'm still new to all of this, but hopefully the more I do it...well...you know.

Oh, almost forgot to mention my recent interest in Ruby (and of course Rails). I was really fired up a few months ago and even started "chapter blogging" my way through my first Rails book. Unfortunately, that was somewhat quenched by my work load at the "day job". I still have an iron in the Ruby fire, and am hoping to get that ramped back up again soon. Along with helping to build the LosTechies community and possibly help out on an OSS project or two. Yeah, not sure where I'll find the time.

On a side note, I'm hoping to start distancing myself away from the "dotnet" part of my persona, if you will, of "joeyDotNet". Mainly because I believe in choosing the best tool for the job at hand and sticking my name to a particular technology was probably a naive move on my part in the first place. Anyone have any thoughts on how to actually go about doing something like this?

Tools Of The Trade

Looks like other folks are listing out some of the tools they use. Well, of course, ReSharper has to be at the very top of my list. I honestly will NOT use VS without it. And you probably already know of my fondness of MonoRail which I've used for almost a couple years (I think) on various projects, back when ASP.NET MVC still meant Make Views Chief (Which sadly, most legacy projects I have to work on still embrace). Of course others like MbUnit, Rhino Mocks, Windsor, TDD.NET and NAnt are also important tools in my box of goodies.

Wrap It Up, Will Ya? yawn

Ok, so that's probably more about me than you ever wanted to know. But if you want to contact me regarding presentations, work, or just to say hey, you can find me here as well as LinkedIn, Facebook and occasionally Twitter.

Just like being there...almost.

2007-12-31T03:46:15Z

Unfortunately I haven't been able to make it to any of the jazzy conferences in 2007. There are a few that I would've really liked to have attended; altnetconf, DevTeach, RailsConf and RubyConf to name a few. Luckily, you can find quite a bit of the content from some of these kinds of conferences online. Just tonight I watched a really good presentation from RubyConf 2007, Behaviour Driven Development with RSpec. There are many more I'd like to watch from RubyConf if I can find the time. And of course the awesome MVC video from ScottGu himself at altnetconf is a must see.

So for those of you, like me, who for whatever reason can't make it to some of these conferences, here are a few good places to start:

Confreaks (these guys do an awesome job, giving you widescreen videos with the slides/code side by side with the presenter)
ScribeMedia (where I originally watched the RailsConf 2006 stuff)
Parleys (saw an awesome DDD video here once)
Google Videos (of course)

I'm sure there are many more, so leave a comment with some more good post-conference resources if you know of any.

(Hopefully my ability to join in the conference goodness will be a bit better in the coming year)