Friday, October 16, 2009

First Snow

This picture was just too cute not to post. Here’s Brinna, my 1-year old daughter about a week ago when we had our first snow of the year. This is her first experience with snow, at least as a walking human being. I’m sure she doesn’t remember it from last winter. You just have to wonder what’s going through her mind right at this moment.

Probably something like: “Why the heck are we getting snow in early October?!?”

Sunday, August 23, 2009

Using a Mobile Source Code Repository

corona_laptop

If you’re a consultant or just do a lot of coding on the side either for profit or to keep your skills sharp, it’s often nice to be able to store your code in a source code repository. The benefits of a personal repository are almost the same as what you get with a repository shared by a team of developers (history-tracking of your code, the ability to branch and merge, easy roll-back of uncommitted changes), the only different is that you’re the only user.

But setting up a reliable source code repository is not trivial, especially if you want to be able to access it while on the go. The easy route is to pay for a hosted solution, but that requires an internet connection (at least for doing source code operations) and they usually cost money if you want something decent.

I’ve come up with a couple simple approaches that allow you to have a personal repository that you control, that can be accessed from anywhere whether you have an internet connection or not, and that can get backed up in the event of data corruption.

Virtualization

Before I jump into the details, I’d like to say a couple things about virtualization since it is a big component of the solutions I came up with (especially the second). While you can easily code using tools installed directly on your host OS, I highly recommend isolating your software development environments on one or more virtual machines (VM’s). There are several reasons for this:

  • It allows you to develop on operating systems that are more comparable to what your application may be running on in production. Most workstation OS’s (like Windows Vista or Windows 7) do not come with all of the server-side components or are just not quite configured the same as the server OS your app may be running on in production (Windows 2000 Server, Windows Server 2003, Windows Server 2008, etc). Some server-side application services (like SharePoint) won’t even run on a non-server OS.
  • It allows you to have multiple different development environments. If you have one project or client that uses Visual Studio 2005 and another that uses 2008, it’s safer to run them on separate operating systems to prevent interference.
  • It allows you to easily play with new beta development tools without hosing up your working development environments. Nothing like installing the latest beta of VS2010 and finding out VS2008 no longer loads your paying customer’s solution.
  • If your host OS is not Windows (ex: Mac OSX) and you want to develop Windows applications, you don’t have much of a choice!

Which virtualization option you go with is pretty irrelevant. For the longest time, I was a big user of Microsoft’s Virtual PC. It was free and it worked great. Since then I’ve moved to VMWare’s Workstation and Player on Windows and Fusion on the Mac. VMWare seems to have a slightly more superior product at the moment, but Microsoft’s virtualization offering is not too far behind, especially when you consider the new functionality available in Windows 7.

Source Control System

There are a few good choices out there for source control.  Personally, I like Subversion (SVN) the best.  It’s free (open source) and a lot of very useful third-party tools are out there.  TortoiseSVN is an excellent Windows GUI client and is free.  AnkhSVN (now being developed by Collabnet, the company that hosts the SVN source control itself) provides source control integration into Visual Studio and is also free.  Finally, there’s VisualSVN.  They make a free server application called VisualSVN Server that allows you to get a Subversion repository up and running in minutes on a Windows server, complete with HTTPS access over Apache.  VisualSVN also makes a Visual Studio source control add-in that’s comparable to AnkhSVN, although arguably more robust.

The two solutions I came up with (Method 1 and Method 2 below) both use VMWare Workstation for the virtualization platform and Subversion for source control.  I’m sure one could adapt these steps fairly easily to work with other systems since the concepts behind them are what are important.

Method 1: Single VM with a Local Repository

When I first had the need to manage my own source control, I had a single VM I used to work on my code. 

single-vm

The source control solution in this case was quite simple since I could pretty much put everything on that VM.

Install a local repository

First I used TortoiseSVN to create a local file-based repository.  You can use the command-line Subversion tool to do this as well, but TortoiseSVN makes it easy:

 create-local-repository

I usually put the repository itself in a standard location (like C:\SVN) so the repository URL in my Working Copy becomes something like file:///C:/SVN/my-repository:

checkout

Configure repository backup

You can do this a number of ways, but you want to make sure that whatever is backing up the repository is doing it automatically on a scheduled basis, so you don’t have to think about it.  An online backup service (like Mozy.com) works well, but there’s a monthly service charge.  I do all my backups on my wife’s computer, which is a MacBook.  Therefore I can use TimeMachine, which is integrated into Mac OS, to backup everything important, including my source code repository.

However, my repository is still sitting in the file system of my VM which isn’t hosted on the MacBook – it’s on my laptop, which is running Windows.  To sync the repository files to the MacBook so they can get backed up I use Windows Live Sync (used to be called FolderShare).  It’s a free file syncing tool from Microsoft that can sync between multiple PC’s and Mac’s via your local network or over the internet and even through firewalls.  All you need is a Windows Live ID to get started.

So as long as all my source code is developed on a single VM, I can keep the source control repository locally on that VM as a simple file-based SVN repository and use Live Sync and TimeMachine to keep my repository files backed up.  Also, if I go mobile and/or the MacBook isn’t connected to the internet (so Live Sync can’t sync) I still have access to the repository so I can do all the source control operations I may need to keep developing code.  Then the next time my VM is able to connect the Mac, it will sync the repository files so they can be included in the next backup.

Method 2: Multiple VM’s with a Shared Repository

The above solution works great until you decide to write your code on more than one VM.  Perhaps you have a couple projects going that require different development stacks.  Maybe one is an ASP.NET application running on IIS6 (Windows Server 2003) and another is on IIS7 (Windows Server 2008).  In that case you really should develop your applications on separate VM’s, each running the correct version of Windows.

second-dev-vm

Of course, now the problem is accessing the file-based source control repository across multiple VM’s.  One option is to move the SVN repository files to the host OS.  Most virtualization solutions like Microsoft Virtual PC and VMWare Workstation have a file sharing feature where a VM can access files on the host OS.  While this works, there’s a very noticeable performance hit.  The other drawback to putting the SVN repository on the host OS itself is that it’s mixing a part of your development environment infrastructure into the host, which conflicts with why you moved everything to virtual machines in the first place.

A better approach is to move the repository to a dedicated VM that can act as a light-weight server and handle source control operation requests from all the client VM’s.  Subversion can be run as a server in this way.  It uses Apache to handle HTTP requests from clients and it seems perform nearly as well as with the local file-based repository.

Let’s take a look at how to set all this up.

Create a source control server VM

The first step is to create this server VM which will host our shared repository and handle client SVN requests from the other VM’s.  This VM won’t require a lot of RAM since it’s primarily just going to serve up request for source control operations.  I created mine using Windows Server 2008, which has a minimum memory requirement of 512MB, so that’s what I went with.

server-vm

Before we move on to installing additional software on the source server VM, we need to take care of some necessary networking infrastructure.

Create a private virtual network to connect all the VM’s

One of the other advantages of most virtualization systems out there is the ability to create virtual networks.  If you have multiple VM’s that need to communicate with one another and especially if that communication needs to be private (or doesn’t need to occur over a public network), you can create a virtual network.

In my situation, I ended up using one of the available custom networks installed by VMWare Workstation (vmnet2 in my case).  You don’t want to use the Host-Only, Bridge, or NAT networks as we need something that’s private that we can dedicate to these VM’s.  In my case, to turn this on, I just added a network adapter to all the VM’s, including the server VM, and connected it to the vmnet2 virtual network:

virtual-network

This new network adapter showed up in the OS (Windows) of all the VM’s.  To make them easier to manager, I renamed them from “Local Area Connection” and “Local Area Connection 2” to “Public LAN” and “Private LAN” in each VM:

Windows-network-adapters

Set up DNS and DHCP services on the server VM

Simply connecting the VM’s together with our new private network connection doesn’t make it so they can communicate (at least not very well) over the private network.  It’s the equivalent to connecting a bunch of computers to a Ethernet hub or switch with network cables.  Each host will default to a 169.254.x.x address and name resolution won’t work at all if each OS has their default firewalls turned on.

Ideally, we need some way of handing out IP’s and ideally having DNS name resolution to those IP’s (the most important one being the server).  To do that we first need to pick an IP subnet and a static IP for the server VM.  In my case I went with 192.168.76.x for the subnet and 192.168.76.1 for the static IP.  I then set the IP address of the “Private LAN” network adapter on the server VM to that static IP, which is done in Windows itself:

static-ip-config

Next we need to turn on DNS services on the server VM so it can resolve DNS names local to our virtual network.  After installing the service, I created a forward lookup zone for a private DNS domain called “ts-local”.  You can use whatever private DNS domain name you want as long as it isn’t something that might conflict with domain names on the internet or any other network you may connected to.

Then I added an A-host for the server VM itself whose hostname is “MCP”:

dns-server

Next we need to enabled the DHCP server service on the server VM so we can hand out the IP’s in our subnet.  Once you install the service, configure it to hand out IP’s to our subnet as well as specify the DNS server IP (same as the server VM) and the domain name (in my case “ts-local”):

dhcp-server

Finally, we should be able to release and renew the IP address on each of the client VM’s and get an IP in our subnet.  We should also be able to resolve the IP address of the server VM from the client VM’s using PING:

ping

Install VisualSVN Server on the server VM

Now for the real reason we created this server VM in the first place.  Setting up a Subversion server that uses Apache (the web server that SVN uses) by hand is no easy task (trust me, I’ve done it).  VisualSVN Server makes this a snap.  It will install Apache, configure HTTPS, and create a location for the repositories.  By default, VisualSVN Server creates the repositories directory at C:\Repositories.  I prefer C:\SVN.

visualsvn-server-manager

After the install is complete, all you have to do is create a username for your client VM’s to use to access the repositories.

Move the repository

If you started with Method 1 like I did, you need to move the file-based repository from the single VM to our shiny new Subversion server VM.  VisualSVN Server makes this easy with it’s import option:

import-repository

Simply browse to where the repository files are and perform the import.  You may want to copy them over to a temporary directory on the server VM first to make them easily accessible from the server’s file system.

Once the repository has been imported, you’ll want to set assign the user account you created earlier to have read/write access to the repository:

asign-user

Your server is now ready to service source control requests!

Point the Working Copies to the new repository

If you already have Working Copies (the term Subversion uses for directories that are under source control) on your original development VM, you can point them at the new URL using TortoiseSVN’s Relocate function.  For example, if you had a working copy connected to file:///C:/SVN/my-repository, you can now point it to http://mcp.ts-local/SVN/my-repository:

relocate-1

relocate-2

Configure repository backup

The final step to getting the multiple-VM source code repository operational is to configure the backup.  The approach is the same as with the single-VM approach – you just have to back up the repository files on the server VM instead.  In my case, that directory is C:\SVN.  I point Live Sync there which syncs the files to my MacBook which in turn backs everything up via TimeMachine.  Very nice!

The multi-VM solution also has the same disconnected benefits that the single-VM solution does.  You can perform all your source code operations anytime you want, even if your host machine is disconnected from the network.  The next time you are connected, Live Sync will sync any repository file changes to the MacBook so they can be backed up.

Saturday, July 25, 2009

Turn off Push with a Jailbroken iPhone 2G on T-Mobile with no Data Plan

So this tip probably applies to about 3 people out there besides me, but the road to a solution was about three days of pain, so I feel compelled to post something.

I've been a happy iPhone user since the beginning of 2009 when my manager at work sold me his 2G. A buddy of mine showed me the ropes of jailbreaking and unlocking the iPhone and before I knew it I was running iPhone OS 2.2.1 via QuickPwn on the T-Mobile network. Better yet, I didn't have to have a data plan (yes, I'm cheap), like you do with AT&T. I just hopped on an available WiFi network whenever I needed to use data - life was good.

Until I decided it would be a fun idea to upgrade my iPhone to the new 3.0 OS.

The actual install itself (I used redsn0w) was quite painless and relatively easy. After I got all my apps re-installed and my settings back to what I had before, I had a sweet phone that couldn't receive incoming calls or text messages. After several re-installs of the OS, attempting to use trial and error to narrow down the source of the problem, I finally stumbled upon the answer. And, yes, I tried googling the heck out of the internet and could not find another poor soul out there who was in my same predicament (hence the first sentence of this post).

The source of the problem was that I had Push turned on (which is the default when you install OS 3.0). Push is a technology that allows applications like email and calendar to receive updates from the server without having to continuously poll the sever. Polling like this is also called Pull, the opposite of Push. Push requires a data plan to work. I don't have a data plan with T-Mobile. I guess Push doesn't behave well without one. Kinda makes sense since I'm sure the iPhone developers didn't really think about this scenario since I think you're required to have a data plan with your iPhone if you're with AT&T.

Anyway, to turn Push off, here's where you go in the iPhone settings:

Settings > Mail, Contacts, Calendars > Fetch New Data > Push > Off

After I did that, everything works! And I have to say the 3.0 version of the iPhone OS is a nice upgrade, even on a 2G.

Wednesday, February 18, 2009

Using the Repository Pattern with CSLA.NET, Part 2

In my last post, we built an abstraction for a data access layer used by CSLA.NET business objects based on the Repository Pattern.  We started by creating two base interfaces, IRepository<TContext> and IContext and then used them to define a set interfaces that abstracted the data access for a simple order entry system.  Those interfaces were:
  • IOrderRepository - represented the repository for orders and line items; a factory object that can create the other repository objects
  • IOrderContext - the context that the IOrderRepository creates that performs the actual data access
  • IOrderInfoDto - a DTO for order summary entities
  • IOrderDto - a DTO for orders
  • ILineItemDto - a DTO for order line items
We then demonstrated how objects based on those interfaces would be called from the data access areas within actual CSLA.NET business objects (in the DataPortal_* methods).  We also showed how the repository object was injected in via dependency injection.  In our case, we used the Unity IOC Container but any IOC Container or dependency injection framework would do.
In this post we're going to build a concrete implementation of our abstract data access layer.  In other words, we'll build classes that implement the above interfaces.  In this case, we'll use LINQ to SQL as our actual data access technology.  However, given the high level of abstraction of our repository pattern implementation, we could conceivable build concrete implementations in many other technologies such as LINQ to Entities, <insert your favorite ORM here>, or even plain old ADO.NET.

LINQ to SQL Classes

Before we jump into the implementation of our interfaces, we need to first build our LINQ to SQL classes.  These classes will play an integral role in our concrete implementation.
LINQ to SQL requires that a physical database exist, so assume that we've started with a simple SQL Server database that contains an Orders table and a LineItem table with a one-to-many relationship between them.  We can then add an Orders.dbml file (LINQ to SQL Classes item) to a Visual Studio 2008 project and drag both tables onto the design surface:
Orders-dbml-screenshot
Visual Studio (actually LINQ to SQL) does a lot of work behind the scenes when we do this.  It code generates a LINQ to SQL DataContext class called OrdersDataContext and two entity classes called Order and LineItem that map to the database tables.
You may also notice is the screen shot above that we've added a set of stored procedures to our database that perform all the inserts, updates, and deletes, and they show up as methods on the OrdersDataContext class.  While LINQ to SQL entity classes have the ability to perform these operations on their own, they have a limitation that forces us to use a alternative mechanism, like stored procedures, instead.  We'll investigate this issue further a little later on in the post.
At this point, the OrdersDataContext, Order, and LineItem classes could be used by a set of CSLA.NET business objects to perform all the required data access.  However, the business objects would be be tightly coupled to the LINQ to SQL code and therefore there would be no easy way to abstract the LINQ to SQL code so it could be mocked for unit testing purposes.  Let's see how we can migrate this code into a concrete implementation of our repository pattern abstraction.

Concrete DTO's

Before we jump into the concrete repository and context objects, let's take a quick look at how we implement the DTO's.  LINQ to SQL makes this relatively easy since the entity classes code generated by LINQ to SQL are the DTO's.  All we need to do is make them implement our DTO interfaces.  Since the LINQ to SQL code generation is done using partial classes, this is quite easy.
First the, concrete IOrderDto class:
partial class Order
    : IOrderDto
{

    byte[] IOrderDto.Timestamp
    {
        get
        {
            return this.Timestamp.ToArray();
        }
        set
        {
            this.Timestamp = new Binary(value);
        }
    }

    IEnumerable<ILineItemDto> IOrderDto.LineItems
    {
        get
        {
            return this.LineItems.Cast<ILineItemDto>();
        }
    }

}
At a minimum, we need to define the partial class Order (which binds to the code-generated Order class at compile time) that implements the IOrderDto interface.  But we also need to add a couple explicit IOrderDto property implementations.
The first is due to the fact that the Order class that was code-generated by LINQ to SQL has a Timestamp property that is of the LINQ to SQL-specific type Binary.  However, the IOrderDto interface defines the Timestamp property as a byte array, which is not specific to a data access technology.  Therefore we need to add the IOrderDto.Timestamp property explicitly and marshal the Binary and byte array values back and forth.
The second explicit property implementation is IOrderDto.LineItems.  The Order class code-generated by LINQ to SQL also defines a LineItems property, but it's of type EntitySet<LineItem>.  Therefore, we need to convert between the two and a handy way to do it is to use the Cast extension method.
The concrete ILineItemDto class is very similar, but we only have to add an explicit implementation of its the ILineItemDto.Property:
partial class LineItem
    : ILineItemDto
{

    byte[] ILineItemDto.Timestamp
    {
        get
        {
            return this.Timestamp.ToArray();
        }
        set
        {
            this.Timestamp = new Binary(value);
        }
    }

}
Now, with the concrete DTO's defined, let's move onto the concrete repository and context classes.

Concrete Order Repository

You may recall the role of the repository object is to be a factory for all of the other objects needed by the data access layer.  It's main job is to create the associated context.  Therefore our order repository will need to create an order context.  It will also need to be able to create any DTO's required by data access methods that require DTO's as inputs. 
Given that, here's our concrete OrderRepository class:
public sealed class OrderRepository
    : IOrderRepository
{

    IOrderContext IRepository<IOrderContext>.CreateContext(bool isTransactional)
    {
        return new OrderContext(isTransactional);
    }

    IOrderDto IOrderRepository.CreateOrderDto()
    {
        return new Order();
    }

    ILineItemDto IOrderRepository.CreateLineItemDto()
    {
        return new LineItem();
    }

}
A fairly simple implementation of a factory.  The CreateContext method creates a new instance of our concrete OrderContext (which we'll see its implementation just ahead), passing a value to its constructor, telling it whether or not it needs to be transactional.  Then we have two methods for creating DTO's: CreateOrderDto and CreateLineItemDto.  Notice that what we're actually returning are instances of the two entity classes code generated by LINQ to SQL since they implement the required DTO interfaces.

Concrete Order Context

While the repository object is the factory that creates all the data access objects, the context object plays the star role in actually performing the data access operations.  Therefore, the OrderContext class is going to have the most meat of any of our concrete repository pattern classes.  Let's examine the OrderContext class in chunks since there's a lot going on. 

Basic Implementation of OrderContext

First, let's take a look at the class definition itself and its constructor that we know takes a isTransactional boolean parameter:
public sealed class OrderContext
    : IOrderContext
{

    private OrdersDataContext _db;
    private TransactionScope _ts;

    public OrderContext(bool isTransactional)
    {
        _db = new OrdersDataContext();
        if (isTransactional)
            _ts = new TransactionScope();
    }

}
As you can see, our OrderContext object wraps an instance of an OrdersDataContext object (via the _db field) which is a LINQ to SQL DataContext.  Therefore, our OrderContext object is essentially an abstraction of a LINQ to SQL data context.  When it implements the remaining IOrderContext interface members, it does this by making calls against that LINQ to SQL DataContext instance.
The OrderContext also wraps a TransactionScope object, which it only creates if the calling OrderRepository object specified that the context is transactional.  That transaction is committed in the CompleteTransaction method, which is required by the IContext base interface:
    void IContext.CompleteTransaction()
    {
        if (_ts != null)
            _ts.Complete();
    }
The last place we interact with this transaction is at the end of the order context object's lifecycle during the IDispose.Dispose implementation:
    void IDisposable.Dispose()
    {
        if (_ts != null)
            _ts.Dispose();
        _db.Dispose();
    }
We also dispose the OrdersDataContext which closes up the database connection.
So far with the OrderContext class we've implemented the creation and clean-up of the object.  Now we need to implement the methods defined by the IOrderContext interface that actually do the data access. 

IOrderContext.FetchInfoList Implementation

First, let's take a look at the implementation of the IOrderContext interface's FetchInfoList method:
    IEnumerable<IOrderInfoDto> IOrderContext.FetchInfoList()
    {
        var query =
            from o in _db.Orders
            orderby o.Date
            select new OrderInfoData
            {
                Id = o.Id,
                Customer = o.Customer,
                Date = o.Date
            };
        return query.Cast<IOrderInfoDto>();
    }

    private class OrderInfoData
        : IOrderInfoDto
    {
        public int Id { get; set; }
        public string Customer { get; set; }
        public DateTime Date { get; set; }
    }
The purpose of this method is to return a list of all the orders in the database.  They come back as the light-weight IOrderInfoDto objects.  Our implementation of this method performs a standard LINQ to SQL query against the Order entity objects in the OrdersDataContext.  However, we don't want to return all the data of each order.  The IOrderInfoDto object is only a subset of that data.  An easy solution is to perform a LINQ projection of Order objects to IOrderInfoDto objects.  This will generate only the T-SQL necessary to populate the data required by the IOrderInfoDto objects.  And of course we need a concrete IOrderInfoDto class to create and return; an easy approach is just to declare a private OrderInfoData class shown above just below the FetchInfoList method.

IOrderContext.FetchSingleWithLineItems Implementation

The next data access method required by the IOrderContext interface is FetchSingleWithLineItems:
    IOrderDto IOrderContext.FetchSingleWithLineItems(int id)
    {
        var options = new DataLoadOptions();
        options.LoadWith<Order>(o => o.LineItems);
        _db.LoadOptions = options;
        var query =
            from o in _db.Orders
            where o.Id == id
            select o;
        return query.Single();
    }
This method is similar to the previous in that it is another LINQ to SQL query.  However, it returns a single DTO (instead of a collection) and that DTO is an instance of the full Order entity class, which happens to implement the IOrderDto interface.
But we don't want to return just the data of the order itself.  We also want to return all of it's child line item data as well (which will be accessible via the LineItems property), and preferably all with one call to the database.  We can do this with a little LINQ to SQL magic by configuring the LoadOptions property of the OrdersDataContext, telling it that when it loads the data of an Order object, go ahead and load its child LineItem objects contained within the LineItems property.
The FetchInfoList and FetchSingleWithLineItems methods pretty much take care of all the data access querying.  Now we need to implement the insert, update, and delete operations. 

Insert, Update, and Delete Method Implementations

While the query method implementations simply took advantage of the built-in LINQ capabilities of the entity classes, we can't quite do the same with the insert, update, and delete methods.  Normally, with LINQ to SQL you can make whatever state changes you want to those objects and when you're ready to persist those changes back to the database, you just call the SubmitChanges method on the DataContext object, which keeps track of which entity objects have changed. 
While this approach could work in our situation, it requires us to maintain references to all of these objects between data access operations performed by our CSLA.NET business objects and this is a problem.  All CSLA.NET business objects need to be serializable so they can function properly in the CSLA.NET DataPortal.  This means that any objects contained within a CSLA.NET business object must also be serializable.  Unfortunately, LINQ to SQL objects are not.
There are some potential workarounds like one where you create a new LINQ to SQL entity object, load it with data, and then attach it to a DataContext as if it were previously fetched by the DataContext, but this is not using LINQ to SQL as it was intended and in many cases produces unexpected behavior.  The only reliable solution is to use a different mechanism than the LINQ to SQL entity objects and the DataContext to persist changes back to the database.  One of the easiest is the use of stored procedures.  This, in fact, is the same approach that Rocky uses with his LINQ to SQL data access code in his sample ProjectTracker application.  In the Orders.dbml file in our project, we simply drag those stored procedures over from the database and LINQ to SQL adds them as methods to the OrdersDataContext.
The last thing we should mention about the insert, update, and delete methods is that, unlike the query methods which returned DTO's (or collections of them), these methods take DTO's as parameters (except for delete which typically only takes an ID).  Therefore, the caller (in this case the CSLA.NET business object) needs to be able to create an empty DTO and populate it.  Which is why the OrderRepository class has the DTO creation methods. 
So, without further delay, here are the insert, update, and delete method implementations for the order entities:
    void IOrderContext.InsertOrder(IOrderDto newOrder)
    {
        int? id = null;
        Binary timestamp = null;
        _db.insert_order(
            ref id,
            newOrder.Customer,
            newOrder.Date,
            newOrder.ShippingCost, 
            ref timestamp);
        newOrder.Id = id.Value;
        newOrder.Timestamp = timestamp.ToArray();
    }

    void IOrderContext.UpdateOrder(IOrderDto existingOrder)
    {
        Binary newTimestamp = null;
        _db.update_order(
            existingOrder.Id,
            existingOrder.Customer,
            existingOrder.Date,
            existingOrder.ShippingCost,
            existingOrder.Timestamp, 
            ref newTimestamp);
        existingOrder.Timestamp = newTimestamp.ToArray();
    }

    void IOrderContext.DeleteOrder(int id)
    {
        _db.delete_order(id);
    }
We also have insert, update, and deletes for the line item entities as well, but they are very similar, so I'll save them for the sample code download at the end.

Sample Application

To make all of this really gel, I wanted to include a fairly extensive sample application that demonstrates everything we've been talking about in the last three posts: dependency injection and the Repository pattern with CSLA.NET.  However, this sample goes a bit further than what we've talked about so far so I wanted to briefly discuss it but save the details for an upcoming post.
A lot of what we've been building here are mechanisms to abstract each layer of our application so they are more loosely-coupled and more testable.  This is why dependency injection is so useful and why patterns like Repository really help.  But the Repository pattern is really just a means to abstract the data access layer.  What if you wanted to abstract the business layer?  What if you wanted to write unit tests that tested the functionality in your UI layer in isolation so that you wouldn't have to build concrete business objects in your tests?  That's the one big additional thing that the sample application sets out to do.  In short, it does this by defining abstractions (in the form of interfaces) for each business class and pulls out the static factor methods into separate factory interfaces and classes.
So here's a quick rundown of the projects that are included in the sample application solution, which is called CslaRepositoryTest:

  • DataAccess - contains the abstract data access layer interfaces
  • DataAccess.SqlServerRepository - a concrete implementation of the types in DataAccess
  • BusinessLayer - CSLA.NET business layer objects
  • BusinessLayer.Test - unit tests that test the business layer, mocking out the data access layer
  • Gui - a Windows Forms GUI that uses the business objects in BusinessLayer.  The GUI uses a Model/View/Presenter-style architecture so it can be more easily tested.
  • Gui.Test - unit tests that test the GUI layer, mocking out the business objects
  • Core - contains common types not necessarily specific to a particular layer
Some other notes:

  • The BusinessLayer is compiled against CSLA.NET 3.5.
  • The sample application uses v1.2 of the Unity Application Block as an IOC Container.  The Gui project configures uses a file called Unity.config to configure Unity.
  • The DataAccess.SqlServerRepository project uses SQL Server Express to attach to a file instance of the database, just like the CSLA.NET ProjectTracker sample application does.  The database file is Orders.mdf and is located in the same directory as the solution.
  • Both test projects use NUnit as their unit testing framework.  You should be able to run all of the unit tests in the NUnit-GUI application.
  • Both test projects use Rhino Mocks v3.5 for mocking objects.
  • The binaries for all of the 3rd party dependencies mentioned above are included in a lib folder in the download.  The only thing you need to have installed is Visual Studio 2008 and SQL Server 2005 Express (which usual comes with Visual Studio 2008).
Finally, here's the sample application.  (UPDATE: Changed link to point to my GitHub repo)

Conclusion

In this post we walked through a concrete implementation of our abstract data access layer that uses the Repository pattern.  Our implementation used LINQ to SQL, but we could have easily created one that used any other data access technology.  In an upcoming post, we'll dig a little further into how to abstract the business layer itself.

Wednesday, January 28, 2009

Using the Repository Pattern with CSLA.NET, Part 1

In a previous post I showed a way to use dependency injection with CSLA.NET.  In this post, I'll show a practical example of how to use that technique to abstract the data access layer used by your business objects using the Repository pattern.  This post is part 1.  At the end of part 2, I'll include a link to a fully-functional demo application that even includes a fully functional business layer, data access layer, UI layer, and unit tests.

Background

The default approach for data access in CSLA.NET is to write your data access code right in the business class in the DataPortal_* or Child_* methods.  These methods have direct access to the internal state of the business object so they can easily persist that state to your data store without breaking encapsulation.  While this approach obviously works, there are several reasons why one might want to move this code out of the business object and into its own layer.  Here are few good ones:

  1. Testability: If you want to be able to write tests for your business objects that test isolated units of code (i.e. unit tests), your business object shouldn't be talking to a real database at test-time.  Therefore some kind of abstraction needs to be implemented that allows the real database to be called at run-time and a mocked database to be called at test-time.
  2. Maintainability: It's never good when you cram too much code into one class.  Arguably, data access logic is very different than business logic, and therefore has a different set of concerns and pressures for change.  Moving it out to its own layer allows both the business logic and the data access logic to change independently with minimal impact on each other.  For example, an abstracted data access layer, if designed at a high enough level, might allow you to change the underlying database platform (ex: from SQL Server to Oracle) or technology (ADO.NET to LINQ to SQL) without having to alter and recompile your business layer. Abstracting data access into its own layer also jives with SRP (Single Responsibility Principal) which states that every object should only have one reason to change.  While CSLA.NET business objects may never only have just one reason to change, it's always good when you can reduce the number of reasons.
  3. Security: Some application architectures require that the data access code itself cannot exist on any public-facing machine (the client workstation in the case of a Rich UI application like WinForms, WPF, or Silverlight or the web server in the case of a web app).  The idea being if a hacker is able to compromise a public-facing machine, and that machine has the data access code on it, they then potentially have the information to access the database itself (i.e. via the connection string).  Moving the data access code to a separate layer means you have the flexibility to only deploy the data access layer to an application server, which stands in between the public-facing machine(s) and the database, behind its own firewall.

Problem

Unfortunately, abstracting the data access layer is no easy task.  There are a lot of choices and options, so you have to be careful that you don't end up limiting yourself.  Here are a few things we need to consider:

  1. We want a data access layer that is called and consumed by the business layer and not one that populates the business objects for us.  If your data access layer populates your business objects it either has to set public properties on your business objects (which fires validation rules unnecessarily) or it has to have access to the internal state of your business objects, which violates encapsulation. 
  2. We need to determine what level we want to do our abstraction.  We could create a low-level abstraction that represents a specific data access technology.  For example, you could abstract ADO.NET by only working with the various ADO.NET interfaces (IDbConnection, IDbCommand, IDataReader, etc.).  But this approach prevents you from easily switching data access technologies, since, by doing so, you would have to alter your business layer code.  Another approach to consider is making your data access layer abstraction generalized enough so it doesn't pin you to a specific technology.

Solution

The solution I chose to pursue was to follow the relatively well-known data access pattern called the Repository Pattern.  Repository is a high-level abstraction of your data access layer that allows the caller to perform any necessary data access (queries, CRUD, etc) and get back a structured set of objects that represent the data.  I like to call these objects DTO's (data transfer objects) since they are nothing more than strongly-typed data structures (just data, no logic).  And since the Repository is a high-level abstraction, you gain the flexibility of being able to change the underlying data access technology in the future as well as do other fun tricks like data caching.

Given that description of Repository, there still are a lot of ways to implement it.  Just do a Google search on "Repository Pattern" and you'll see what I mean.  I've come up with what I think is a relatively simple approach that seems to work well with CSLA.NET.  Let's take a closer look.

Repository and Context

There are two key components to my Repository design: the repository and the context.

The repository represents the gateway to the data access layer.  The repository will create all the necessary data access objects we need to perform our data access.  Essentially it is a factory object and it can be injected into our business object via dependency injection. 

When you want to actually perform data access operations, you usually want this to be done in a context of some sort that you can dispose of when you're done.  Maybe you need to execute one or more calls atomically within a transaction or maybe you just need to do some querying and you want to ensure that the connection is closed when you're done, even if the query fails.  This is what a context is for. 

The repository's primary job is to create a context when the business object needs to perform data access.  It can also create other data access objects, like DTO's, that may be needed to assist in performing data access.

So, let's take a look at what the repository and context might look like in code:

public interface IRepository<TContext>
where TContext : IContext
{
TContext CreateContext(bool isTransactional);
}


public interface IContext
: IDisposable
{
void CompleteTransaction();
}

Pretty simple.  An IRepository object can create a specific type of IContext object, and when it does, you have to specify whether or not the IContext is transactional.  If an IContext is transactional, it has a method called CompleteTransaction that is called by the business object to indicate that the transaction has succeeded and should be committed.  And regardless if the IContext object is transactional or not, it is disposable which makes it easy for the calling business object to close things like database connections.

And by the way, I personally prefer to use interfaces for my abstractions (vs abstract classes) because they're much easier to mock in a testing scenario.

Notice that neither of these interfaces has any methods that perform specific database operation (ex: CRUD).  I prefer to move those down into more specialized interfaces that abstract a specific set of entities in your data access layer.  For example, here's a repository and a context for a simple order-entry system that contains order and line item entities:

public interface IOrderRepository
: IRepository<IOrderContext>
{
IOrderDto CreateOrderDto();
ILineItemDto CreateLineItemDto();
}

public interface IOrderContext
: IContext
{
IEnumerable<IOrderInfoDto> FetchInfoList();
IOrderDto FetchSingleWithLineItems(int id);
void InsertOrder(IOrderDto newOrder);
void UpdateOrder(IOrderDto existingOrder);
void DeleteOrder(int id);
void InsertLineItem(ILineItemDto newLineItem);
void UpdateLineItem(ILineItemDto existingLineItem);
void DeleteLineItem(int id, int orderId);
}

Most of the action is down in the IOrderContext object; it contains all of the data access methods that involve interacting with the database.  There are two methods for querying and retrieving order data.  The remaining methods perform the inserts, updates, and deletes.

Notice that my query methods are explicit: "fetch info list" and "fetch single order with line items".  I've seen some Repository Pattern implementations out there that provide more generic query methods that allow you to harness the power of LINQ by returning an IQueryable<T> object.  While there's nothing wrong with this approach, it does lock you into using data access technologies in your concrete implementation that only support LINQ.  Some do, some don't (like ADO.NET).  To be flexible, I chose to not lock myself into LINQ, so my query methods return simple IEnumerable<T> results.  And even if you are confident that any data access technology you will use is LINQ-friendly, not all LINQ implementations are the same.  For example: LINQ syntax for LINQ to SQL is a little different than LINQ syntax for LINQ to Entities.

The DTO's

Notice that all data is transferred to and from these data access methods via DTO's.  The only issue is that these DTO's are interfaces so if the business object has to pass them to the context (say to the InsertOrder method), it needs a way to create one first.  That's where the DTO factory methods on the IOrderRepository object come into play.  This further cements the concept that the repository is really an object factory.

Before we move on, let's take a look at the definition of the DTO interfaces in my example.

First, the IOrderInfoDto which is returned by the FetchInfoList method:

public interface IOrderInfoDto
{
int Id { get; set; }
string Customer { get; set; }
DateTime Date { get; set; }
}

This is giving us only a portion of the fields that may be defined by the order entity in the database.  The purpose of this DTO (and the ReadOnlyBase<T> business object it populates, is to provide a simplified view of order data, perhaps to populate a list in the GUI.

When we need a single entire order, the FetchSingleWithLineItems method returns the IOrderDto object:

public interface IOrderDto
{
int Id { get; set; }
string Customer { get; set; }
DateTime Date { get; set; }
decimal ShippingCost { get; set; }
byte[] Timestamp { get; set; }
IEnumerable<ILineItemDto> LineItems { get; }
}

It contains pretty much the same data as IOrderInfoDto plus the remaining order entity data and also a collection of child line items.  Those line item objects are of type ILineItemDto:

public interface ILineItemDto
{
int Id { get; set; }
int OrderId { get; set; }
string ProductName { get; set; }
decimal Price { get; set; }
int Quantity { get; set; }
byte[] Timestamp { get; set; }
}

By the way, both the IOrderDto and ILineItemDto interfaces contain a Timestamp property which is used to manage data concurrency.  This timestamp has to be persisted into the business object itself.  I'll show how this is done in the sample code, but Rocky also does it in his ProjectTracker sample application.

Calling the Repository from the Business Object

I think to really see how this implementation of the Repository Pattern works, we should take a look at how a business object would call these data access objects.  Let's examine the simplest case of a root-level collection of read-only order business objects (we'll name it OrderInfoCollection) performing its data access while fetching the collection.

private void DataPortal_Fetch()
{
RaiseListChangedEvents = false;
using (var context = EnsureDependency(_repository).CreateContext(false))
{
IsReadOnly = false;
foreach (var dto in context.FetchInfoList())
{
var child = DataPortal.FetchChild<OrderInfo>(dto);
this.Add(child);
}
IsReadOnly = true;
}
RaiseListChangedEvents = true;
}

The code follows the standard pattern for the implementation of the DataPortal_Fetch method of a business object that inherits from ReadOnlyListBase<T,C> where we turn off list-changed events and make the collection temporarily writable while we load it.  But instead of opening up an ADO.NET database connection or creating a LINQ to SQL data context, we talk to an abstract repository instance instead and call CreateContext which returns us a IOrderContext instance.  All context objects implement IDisposable so we wrap it in a using statement which guarantees things like database connections and transactions will be closed up at the end.

Within our using block is where we make all the data access method calls against the context.  In this case, we're calling FetchInfoList, enumerating the IOrderInfoDto objects returned, and using each to create an associated child OrderInfo business object for the collection.

Dependency Injection

You may be asking: where does the instance of the _repository field get created and what's this EnsureDependency method?  This goes back to how I implement dependency injection with CSLA.NET which I detail in the previous post.  Here's the definition of that field and the method that Unity (my dependency-injection/IOC Container framework of choice) uses to inject it:

[NonSerialized]
[NotUndoable]
private IOrderRepository _repository;

[InjectionMethod]
public void Inject(IOrderRepository repository)
{
if (repository == null)
throw new ArgumentNullException("repository");
_repository = repository;
}

This is the magic that allows us to easily mock out the data access layer at test-time and use a real concrete implementation at run-time.  Showing how you would mock these objects is a bit more than we have time for now.  However, I will provide unit tests that show it in the downloadable sample code at the end of part 2.

Transactional Data Access Code

So we showed the simple example of how a read-only collection would call the data access layer to query the database and populate itself.  Let's also show an example of some data access code in a business object that's transactional. 

Let's jump over to a full BusinessBase<T> business object for an order (we'll call it Order) and take a look at what the DataPortal_Insert method implementation might look like for inserting an order:

protected override void DataPortal_Insert()
{
using (var context = EnsureDependency(_repository).CreateContext(true))
{
var dto = EnsureDependency(_repository).CreateOrderDto();
dto.Customer = ReadProperty<string>(CustomerProperty);
dto.Date = ReadProperty<DateTime>(DateProperty);
dto.ShippingCost = ReadProperty<decimal>(ShippingCostProperty);
context.InsertOrder(dto);
LoadProperty<int>(IdProperty, dto.Id);
_timestamp = dto.Timestamp;
DataPortal.UpdateChild(ReadProperty<LineItemCollection>(LineItemsProperty), this, context);
context.CompleteTransaction();
}
}

A little more complicated, but the pattern is very similar.  We create a context (this time specifying that it's transactional) and we perform our data access operations against it.  In this case we to first create an empty IOrderDto so we can populate it with the business object state and pass it into the InsertOrder data access method. When that's done we load the ID and timestamp of the newly inserted order entity back into the business object.  Finally, we cascade the update call down to any child line item business objects.

Conclusion

So this is all great and fun and with just the code we've written so far we could get our business layer up and running with passing unit tests.  But with all of that we still haven't actually hit a real database!  We need to code up a concrete implementation of our repository interfaces to get there.  I'm going to save that for the next post and with that post will come the promised full set of sample code.