Tomorrow I start work on a WPF application that is supposed to be as modular as possible. Since I know almost nothing about WPF or modular applications, I started researching a little bit how it should be done. Basically I found only two ways I cared to expand my research on: Prism (patterns & practices Composite Application Guidance for WPF and Silverlight) and MEF (Managed Extensibility Framework).

Unfortunately I've only had time to do some Prism, although MEF seems to be the way to go. First of all it is a general framework, not limited to WPF/Silverlight and secondly it is used in the new Visual Studio 2010 release. What is amazing is that both frameworks come as freely available opensource.

Ok, back on Prism. The concepts are simple enough, although it takes a while to "get" the way to work with them. Basically you start with:

• a Bootstrapper class - it initializes the Shell and the Catalog
• a Shell - a visual frame where all the modules will be shown
• a Module Catalog - a class that determines which modules will be loaded
• an Inversion of Control Container - class that will determine, through Reflection usually, how to initialize the classes and what parameters they receive
• a RegionManager - class that will connect views with Regions, empty placeholders where views are supposed to be shown
• an EventAggregator - a class that is used to publish events or subscribe to events without referencing the objects that need to do that

Easy right? I don't even need to say more. But just in case you don't have a four digit IQ, better watch this four part video walkthrough:
Creating a modular application using Prism V2 - part 1
Creating a modular application using Prism V2 - part 2
Creating a modular application using Prism V2 - part 3
Creating a modular application using Prism V2 - part 4.

I did try to take the WPF Hands on Labs project and mold it with Prism and it partially worked. The problem I had was with the navigation controls. These work as a web application, where you call the XAML and it has to be a file somewhere, and you have events for the calling and returning from those "pages". I could find no way to encapsulate them so I could build no modules out of them and the whole thing collapsed.

So, for a quick walkthrough on using Prism with WPF.
Creating the core:

1. create a new WPF application project
2. reference the Prism libraries
3. create the Bootstrapper class by inheriting UnityBottstrapper that will determine the Shell (a WPF window class) and set it as the application MainWindow, as well as create the type of ModuleCatalog (either take a default one or inherit one from IModuleCatalog) you want
4. create the layout of the Shell and add region names to the controls you want to host the loaded modules (example <ContentControl Grid.Row="0" Margin="2" Regions:RegionManager.RegionName="SearchRegion"/>

.

Creating a module:

1. create a new library project
2. add a class that inherits IModule
3. the constructor of the IModule can have different parameters, like an IRegionManager, an IUnityContainer, an IEventAggregator and any other types that have been registered in the container (I know it hasn't been initialized in the core, the catalog takes care of that). The IoC container will make sure the parameters are instantiated and passed to the module
4. register views with regions and any additional types with the IoC container in the Initialize method of the module
5. create view classes - WPF controls that have nothing except the graphical layout. Any value displayed, any command bound, any color and any style are bound to the default DataContext. The views will receive a view model class as a constructor parameter which they will set as their DataContext
6. create the view model classes - they also can have any types in the contructor as long as they are registered with the IoC container, stuff like the eventAggregator or a data service or other class that provides the data in the view model.
7. provide all the information needed in the view as public properties in the view model so that they can be bound
8. subscribe or publish events with the event aggregator

As you can see, most of the work is done by the modules, as it should be. They are both communicating and displaying data using the event aggregator and the binding mechanisms of WPF. There are some differences between how WPF and Silverlight approach some issues. The Prism library brings some classes to complement the subset of functionality in Silverlight that are not needed in WPF. However, one can still use those for WPF applications, making a transition from WPF to Silverlight or a mixed project more easily maintained.

The video walkthrough (as well as my own text summary) are based on the rather new Model-View-ViewModel pattern, which many people call a flavour of MVC. It was created specifically for WPF/Silverlight in order to separate behaviour from user interface.

Expect more on this as soon as I unravel it myself.

This is only a bookmark. If you want to know how to do this, read this link from Microsoft: Walkthrough: Using ASP.NET Routing in a Web Forms Application.

Well, you might laugh out loud when you read this, but it is an achievement for me. I wanted to update a table (table A) which had a connection to another table (table B) through a many to many table (table A2B). I had changed A2B to link A to a new table C. I wanted to reflect that change in my Entity Model.

So the first step was to right click on the model and do an "Update Model from Database". Then went through all the steps and clicked Finish. Now entity A has both a collection of Bs and a collection of Cs. I wanted to delete the B collection but, to my chagrin, no delete option was available.

The epiphany came when I managed to select the graphical link from entity A to entity B. The link has the delete option! Which, I have to admit, does make sense. So whenever you want to remove an association between entities... well, select the association and delete it, not the entity properties.

I was trying to do something that is usually a drag: create a query with variable parameters. Something like Google search, where each query has a number of words and you want to search for each of them. How can one do this with Linq? More than that, Linq to Entities, which is pretty much making my life hell nowadays.

The idea would be to start with a IQueryable object and keep adding queries to it. This isn't so bad when you want to do an AND operation between your individual query strings.

var dc=new MyEntities();
var content=dc.Content;
foreach (var filter in filters) content=content.Where(c=>c.Text.Contains(s));

But what if you want to do an OR operation? Then you would want to be able to add stuff to the lambda inside a singe Where method. I won't get into the details, rather give you a link. I am not an expert on this myself, so it would be pointless. The more important thing is that you can do this a lot easier by using a library called LinqKit, which adds some very useful extension methods for Linq, enabling you to dynamically create queries, lambdas, etc.

There are other 'lighter' versions of this method on the Internet, unfortunately most of them are usable only with Linq to SQL, not the Entity Framework. For example I've read an interesting blog entry about this, tried the code, and got an ugly error: "The LINQ expression node type 'Invoke' is not supported in LINQ to Entities". LinqKit also did this in its earlier versions, but the Albahari brothers fixed it. So, as far as I can see, I recommend this library for real life linq composition.

The scenario is this: You use a ScriptManager, a TextBox and a Button and your project is at least referencing the Ajax Control Tookit library. You expect when pressing Enter in the TextBox to get to the button Click event. Instead you get the Microsoft JScript runtime error: ‘this._postBackSettings.async’ is null or not an object javascript error.

I found the solution here: JScript Exception in AJAX Control Toolkit. Basically, put the textbox and the button in a Panel and set the DefaultButton property to the button id.

Today a new AjaxControlToolkit was released. I had hoped that at least they noticed the patch I made the effort of sending them, the one that fixed the dynamic TabPanel issue. But no. Three more controls and probably most of the same old bugs.

Here are more details: New release of the Ajax Control Toolkit

I have been working on a silly little project that involved importing files and then querying the data in a very specific way. And I wanted to do it with the latest technologies so I used The Entity Framework! (imagine a little glowing halo around that name and a choir in the background).

Well, how do I do an efficient import in Linq to Entities? I can't! At most I can instantiate a lot of classes and add them to the DataModel, then SaveChanges. In the background this translates to a lot of insert statements. So it occurred to me that I don't really need Entities here. All I needed is good old fashioned ADO.Net and a SqlBulkCopy object. So I used it like that. A bit of unfortunate translation of objects to a DataTable because the SqlBulkCopy class knowns how to import only a DataTable and I was set.

Ok, now back to the querying the data. I could have used ADO.Net, of course, and in this project, I would probably have been right, but I suspected the requirements for the project will change so I used Entities. It worked like a charm and yes, the requirements did get bigger and stranger as I went. But then I had to select the users that have amassed a number of rows in two related tables (or use a value in the user table) but only if the total number of amassed rows would satisfy a formula based on a string column in the user table that mapped to a certain value stored in the web.config a complicated query. I did it (with some difficulty) in Linq, then I had to solve all kind of weird issues like not being able to compare a class variable with an enum value because the number of types that can be used in a Linq to Entities query is pretty limited at the moment.

Well, was it the best way? I don't know. The generated SQL is something containing a lot of select from select from select sometimes 6 or 7 levels deep. Even the joining is done with select from tables. Shouldn't I have used a stored procedure instead?

To top it off, I listened to a podcast today about Object Databases. They do away with ORMs because there is no relational to begin with. The guy argued that if you need to persist objects, wouldn't an Object Database be more appropriate? And, as reporting would be a bitch when having to query large amounts of tabular data, shouldn't one use a Relational Database for this particular task in the project?

So this is it. It got me thinking. Is the database/data access layer the biggest golden hammer out there? Are we trying to use a single data access model and then build our applications around it like big twisting spaghetti golden nails?

Well, this is a time of great change in my life. First I had to give my cat away, due to medical reasons. I had him for more than 5 years and I really liked him. Now he is living in the countryside, with my parents in law, trying to get some pussy (sorry, couldn't help it :) ) and getting beat up for it by sturdy country female cats.

A few months after I got the cat, I also got a new job, prompting me to write my first blog entry. I was saying then that I am starting my first real software developer job in an Italian company. Now, after almost five years, I am giving away my cat and also changing jobs.

My new company is (hopefully) a place where I can accelerate the rate of my learning and professional development and I will be working there on a Windows desktop WPF+WCF+WF+Entity Framework application. So expect a lot of blog entries about new (for me) technologies. I will be starting work there on the first of June.

On the other hand I don't know if it would be permitted (or I would have the time) to stay available for chat on the blog, so if I don't answer, it's probably because I can't.

Wish me luck, everybody!

VirtualPathUtility.ToAbsolute("~/") = /Site/
HostingEnvironment.MapPath("~/") = c:\InetPub\wwwroot\Site\.
new Uri(HttpContext.Current.Request.Url,VirtualPathUtility.ToAbsolute("~/")).AbsoluteUri = http://server/Site/

A Romanian blogger made a very information link laden post about the new features in .Net 4.0. Check it out: Learning resources for .NET 4.0 new features.

I have seen a small video presentation about the new ASP.Net 3.5 SP1 Script Combining feature. Basically you take a bunch of scripts (like the ones from AjaxControlToolkit) and you bundle them together in a single cacheable file. This decreases the number of concurrent connections on your production site.

The problem was that you had to use some component to see what script files were being loaded and then manually add them to the ScriptManager CompositeScript Scripts collection. And this applies only to correctly registered scripts, not stuff that is embedded in the HTML or what not. Isn't it easier to just parse the generated HTML and then replace the script tags, I thought?

Well, I did a small Response filter/IHttpHandler in about two hours. It would take all consecutive external file references and combine them in a single cached and cacheable call. Then I tested it with Asynchronous postbacks. Epic Fail! The main problem was that the combined scripts would re-register themselves at postback and throw all kinds of errors therefore. But how did they know if they were registered or not before?!

I vaguely remembered an old post of mine about the notifyScriptLoaded function that must be called at the end of every external javascript registrations. Examining the system a little I realised that the flow was like this:

1. register the script (either include it in the HTML in regular render or sending it to the UpdatePanel javascript engine to be registered as a new dynamically script element in the Head page section)
2. if the registration is an async dynamic one, check in an array if the script is loaded and if it is, don't register it
3. in the script call Sys.Application.notifyScriptLoaded() which would take the src of the script element and use it as a key in the above mentioned array to declare it has been loaded

Of course, that means combining all the scripts in a single file registers only that file and you get the original files registered again. Then you get errors like 'Type [something] has already been registered.' or (because there are more than one script file bundled together) 'The script [something] contains multiple calls to Sys.Application.notifyScriptLoaded(). Only one is allowed.'.

Well, I did manage to solve the problem by following these steps:

1. forcefully remove Sys.Application.notifyScriptLoaded() from bundled scripts
2. add Sys.Application.notifyScriptLoaded() at the end of all scripts
3. surround the various scripts with a code that looks like

   sb.AppendFormat(@"if (!window.Sys||!Sys._ScriptLoader||!Sys._ScriptLoader.isScriptLoaded('{0}')) {{
   {1}
   if (window.Array&&Array.add&&window.Sys&&Sys._ScriptLoader) Array.add(Sys._ScriptLoader._getLoadedScripts(), '{2}');
   }}
   ", HttpUtility.HtmlEncode(src), content.Replace("Sys.Application.notifyScriptLoaded();", ";"),src);

.

With problem solved, the AutoScriptCombiner works, but it feels wrong. I wanted to post it on Github, you see, but in the end I've decided not to. However I did learn something about how the Asp.Net Ajax framework functions internally and I wanted to share it with you.

Another great blog post from Rick Strahl made me investigate a little the new JSON support in the latest browsers. Internet Explorer 8 already has it and FireFox has it in its beta version and soon to be released 3.5 version.

Rick's article says more than I could. A short summary:

• use the native JSON object
• much faster and safer than any js library
• still has problems with types of data which have no native notation or type, like datetime

.

Update: the native JSON object is not something that Microsoft and Mozilla got together and decided is a good idea (which for me seems the natural way of deciding the direction of code versus compatibility, not overbloated committees and standards) but because the release of the latest ECMA Javascript specs which contains, among other things, the JSON thing and the getter/setter syntax.

Gödel, Escher, Bach by Douglas Hofstadter has won a Pulitzer Prize and I believe it is the only thing that ever made me want to read other Pulitzer Prize books. I have to thank Meaflux for pointing me to this book and if he ever writes anything in his blog, you can find it here. Anyway, here is my humble review:

I have just finished reading the book and, in its self-referential spirit, I am also starting reading it. A strange loop of sorts, when starting to read a book follows finishing it. It is not an easy read, but it is certainly worth it. I was instantly both in awe and full of envy on this Hofstadter guy that I have never even heard about before.

What is the book about? I believe the most basic answer is it is about the roots of consciousness, and before you run away thinking this is some sort of new age pseudo (or fully) religious crap, let me assure you it is not. The title itself shows the perspective one gains by reading it: look at the same thing from the viewpoints of a mathematician (meta-mathematician, at that), an abstract painter and a great music composer. It's a definition of abstract thought by intersecting the works of three great abstract thinkers. But it is more than that.

The most intriguing part of the book it is how self referential it is. There are portions in the book that are modeled after Bach fugues while paraphrasing Escher drawings in order to illustrate a mathematical idea of Gödel. It talks about artificial intelligence, consciousness, the workings of the brain, formal systems, computer programming, music, art, science, mathematics, quantum mechanics, biology, genetics and does so in a way that links all these things together in a reasonably easy to understand way. It does not feel like a book made out of separate chapters, but one master-single-piece linked to itself in the most imaginative and twisted ways.

I urge you to buy the book, if you find it. I have read a text OCR version of it and I know I missed a lot. If you can't afford it, there it a torrent on the net with the PDF scanned version as well as the music, paintings and other media the book talks about.

The bottom line is that it is an amazing book. For someone like me, a software programmer dreaming of AI, it was a shame I didn't read it before. I almost believe that you will see me in buses like those old ladies reading the Bible, only with GEB in hand. I can't imagine anyone over 15 years of age that shouldn't read this book. I doubt anyone under 15 can truly comprehend it and, as Frank Herbert's Dune, it must be read every 10 years or less, just to see how much more you can understand from it.

Update: I found on the Internet a full length movie based on Hofstadter's ideas. Interesting, in a geeky/goofy kind of way. Here it is: Victim of the Brain.

While working on RegexConverter I've started learning things about XML. I disliked it before, but now... I loathe it completely. It feels like the XML people are from another planet and I have to twist my brain at straight angles just to understand what they meant.

One of the problems I met was "how to declare in an XML schema a unique constraint/index on the entire document when there are nested elements of the same type". In other words, something like this:

<regex>
<group index="1">
    <group index="2">
    </group>
    <group index="3">
    </group>
</group>

Bottom line:

1. Use the declared namespace
2. Declare the index in the root element
3. Use the name of the nested element (in my case group) in the selector
4. Use the one or more levels deep selector (.//)
5. Use the declared namespace AGAIN.

.

Concrete example:

<xs:element name="regex" type="regexType">
    <xs:unique name="uniqueIndex">
      <xs:selector xpath="mstns:group|.//mstns:group"></xs:selector>
      <xs:field xpath="@index"></xs:field>
    </xs:unique>
  </xs:element>

I tried to solve this for two hours only to find the solution was this idiotic. Neither one of "group","mstns:group","mstns:group|.//group" work. The only one is "mstns:group|.//mstns:group".

Grrrr...

Update: two more days of work and a huge text file of regular expressions and I know a lot more about the syntax of .Net Regex than even MSDN :)
The third (Apr11) release of the library has a huge load of bug fixes and new features and it is now... [Tadaaaaa]... a stable version. At least I think of it that way :) Go download it.!

I've been working for two days on an idea. What can I do to make those long regular expressions that I always leave in my code more readable and easy to understand without having to compile automatons in your head?

I have first researched on Google regular expression converters. Nothing was even close to what I wanted. Also, on the more scientific side, people were talking a lot about BNFs, as a superset of regular expression syntax. I looked at BNF. Yes, it describes anything in a human readable form, it is used in RFCs but I hate it even more than XML! So XML it is. Most of the inspiration for the code came from this link: Regular expressions and regular grammars

I give you, RegexConverter. It is a library+demo that transforms a Regex into an XML and then back again. The demo application demonstrates (duh!) this by having two panes, one in which to write regex and the other in which to write XML. Changing one, also changes the other. It warns you of errors in both Regex, RegexConverter and then checks if what it got can be safely converted into your changed string!

Please tell me what you think. I believe it can be a real help in understanding regular expressions, some specific ones or regular expression syntax in general, whether one is a pro or a complete noob.

I've worked hard to design the library source in a way that is understandable, I also added comments everywhere. I tried to implement all the specifications of .NET regular expressions from the MSDN site so if you have a regex that is valid but can't be turned into XML or the conversion is not perfect, let me know.

The link is: RegexConverter on Github.

I will update this post with more science and links to my places of inspiration for this, so stay tuned.

Update: Ok, so I haven't updated this post much. Shame on me. I was reminded of this project when I got an email from the FamousWhy site which said RegexConverter "has been granted the Famous Software Award". I know it's probably an automatically generated message and went to many or all of the Codeplex people, but still: automated attention is still attention :)