Adverse conditionals

Static links for aspx pages in 20 lines of code

2009-02-24T20:08:00.001Z

Still using WebForms? Sick of magic string URLs that break when you move you ASPX pages around? Try this! I've created a class called UrlFor<TSomePage>. It examines the page class and figures out the URL by making a few assumptions about your aspx pages (the aspx file has the same name as the class, the namespaces match the folder structure, that kind of thing...)

public class UrlFor<TAspxPage>
    where TAspxPage : Page
    private static string baseUrl = typeof(TAspxPage).FullName.Substring("YourAppBaseNamespace".Length).Replace('.', '/') + ".aspx";

    public Dictionary<string, string> QueryString = new Dictionary<string, string>();

    public UrlFor<TAspxPage> Set<TValue>(Expression<Func<TAspxPage, TValue>> getProp, TValue value)
    {
        this.QueryString[((MemberExpression)getProp.Body).Member.Name] = value.ToString();
        return this;
    }
    public override string ToString()
    {
        var qs = String.Join("&", this.QueryString.Select(pair =] pair.Key + "=" + HttpUtility.UrlEncode(pair.Value)).ToArray());
        return baseUrl + (qs.Length ] 0 ? ("?" + qs) : String.Empty);
    }
}

anyway, it lets you write things like this in your code:

[%= new UrlFor<Meetings.Details>().Set(d => d.WorkspaceId, 1).Set(d => d.MeetingId, 2) %>

where Meeting.Details is a Page class with a property on it called MeetingId and one called WorkspaceId. This generates a url like

/Meetings/Details.aspx?WorkspaceId=1&MeetingId=2

Of course your pages may not have properties that correspond to the querystring parameters, but you can always write an extension like

public static class UrlForExt
{
    public static UrlFor<T> SetWorkspaceId<T>(this UrlFor<T> url, int value)
        where T:WorkspacePage
    {
        url.QueryString["workspaceid"] = value.ToString();
        return url;
    }
}

and get the same effect by writing something like

new UrlFor<Default>().SetWorkspaceId(1)

Neat huh!? Use this to build your links and you can rename and move your pages to your hearts content!

MigrationScriptGenerator

2009-02-09T01:09:00.001Z

Hi all, I thought I'd give you all a heads up on a new tool I've written for keeping your database schema in sync with your development schema
Basically its a tool for generating SQL scripts for changes (using Open DBDiff), with errors if you do a change that might lose data. It creates a script to you to review. If your changes are non destructive, then it just makes the change script for you.
Its pretty alpha at the moment, but seems functional enough for development work. Check out the sample project - it's hooked up with Fluent NHibernate. Try changing some of the entity classes and rebuilding, and have a look in the /Sql folder at the generated scripts.
You can find it at http://code.google.com/p/migrationscriptgenerator/

Automatically serializing value objects using Fluent NHibernate

2009-01-18T17:24:00.004Z

After reading James Gregory's posts on Fluent NHibernate (in particular http://blog.jagregory.com/2009/01/11/fluent-nhibernate-auto-mapping-type-conventions/) I had a quick play and came up with this.

It's very cool how FNH will generate your database for you, but sometimes you want to have a value object instead of an entity, so I have created a generic class you can use which will serialize an object into a field automatically, so add something like this to your FNH conventions.

.WithConvention(c => c.AddTypeConvention(new AutoTypeConvention<LatLng>()))

So now my class

public class Venue
{
    public virtual int Id { get; set; }
    public virtual string Name { get; set; }
    public virtual LatLng Location { get; set; }
}

Will have the following SQL generated for it by FNH/SchemaExport

create table [Venue] (
  Id INT IDENTITY NOT NULL,
   Location NVARCHAR(MAX) null,
   Name NVARCHAR(100) null,
   primary key (Id)
)

The LatLng object will be serialized into the NVARCHAR(MAX) column automatically. Neato! Warning - you should only do this with immutable types (I think) and you might want to use a version tolerant serialization strategy in case the definition of your value type changes from the version serialized in the db. I've provided an XML serializer implementation, feel free to change it!

You can get the code from here (apologies for the file hosting)

UPDATE: The xml serializer is a bit broken - get the latest from http://refactormycode.com/codes/710-polymorphic-xml-serializer#refactor_144121

Announcing SimpleScriptRunner - a very basic migrations tool

2008-10-27T17:37:00.005Z

Just thought I'd share a very basic tool called SimpleScriptRunner. Its basically just something for running a directory of numbered SQL scripts against a SQL Server database for migrations, using the syntax

SimpleScriptRunner.exe <server> <database> <path to sql scripts>

I like to call it from a post build event in a VS project like so:

SimpleScriptRunner .\sqlexpress $(SolutionName)Db $(ProjectPath)

If the database doesn't already exist, it will create it. The scripts are run in numerical order (you need to call them something like 0001 - First script.sql), and any scripts that have been already executed against the database are skipped*.

I had been using MigratorDotNet for my migrations, and its probably a better tool as it can target databases other than SQL Server, but if you're using a lot of MS SQL Server, you might find this handy.

You can find it here on google code.

*actually if you modify the most recent script, it will re-execute it. This is a good thing when you are developing.

LinqtoDBMLRunner improvements

2008-07-14T22:41:00.001Z

I've made a couple of improvements. Generally you want to make a couple of changes to an element in the xml file once you've selected it. The Apply extension method allows you to chain changes together so you don't need to keep reselecting objects e.g.

root.Table("People")
    .Apply(table => table.Member = "Persons")
    .Apply(table => table.Type
                        .Apply(type => type.Name = "Person")
    );

The next improvement is a method to automatically recognise single table inheritance from the database, if, of course, you stick to the convention.

Image you have a table defined with the following SQL. The underscores in the column names are used to indicate that the column should belong to a subclass.

CREATE TABLE [dbo].[Device](
    [Id] [int] IDENTITY(1,1) NOT NULL,
    [DeviceType] [int] NOT NULL,
    [Mouse_NumberOfButtons] [int] NULL,
    [Keyboard_NumberOfKeys] [int] NULL,
    [Keyboard_WirelessKeyboard_Frequency] [decimal](19, 5) NULL,
)

You can then add the line below to your DbmlRunner script

root.ExtrapolateSubclasses();

This will rearrange your Dbml mappings to give the correct inheritance, as in the diagram below

This lets you very quickly and easily add new subclasses without having to do any real dbml scripting. A word of warning. It generates the inheritance modifier value for the subclass based on the order the columns appear in the system. If you remove a subclass and its columns once you have data in the system, you'll probably want to manually script the inheritance modifier values yourself. I might make it run off a named enum in the future, theres an idea...

Try-catching in a single line of code

2008-06-03T17:05:00.003Z

Do you like reducing your line count at the expense of readability to others? Then you'll like this!

public static class FuncExtensions
{
    public static TResult Catch<TExc, TResult>(this Func<TResult> func, Func<TExc, TResult> handleException) where TExc : Exception
    {
        try
        {
            return func();
        }
        catch (TExc ex)
        {
            return handleException(ex);
        }
    }
}

This lets you write things like:

int x = new Func<int>(DoSomething).Catch((NullReferenceException e) => -1);

instead of

int y;
try
{
    y = DoSomething();
}
catch (NullReferenceException ex)
{
    y = -1;
}

You can of course write overloads for Funcs that take more than one parameter.

Edit: I've written a version that lets you chain catches but I'm not sure if its better or not...

int x = new Func<int>(DoSomething)
    .Catch((NullReferenceException e) => -1)
    .Catch((NotImplementedException ex) => -2)();

storing ViewState in memcached - the ultimate post

2008-06-02T08:17:00.005Z

The HttpModule for storing memcached in the ViewState is dead - I discovered that you can far more easily intercept ViewState using a custom PageAdapter. Please go to http://code.google.com/p/memcached-viewstate/ for frictionless memcached viewstate goodness! Hopefully this is the end of the matter :)

Introducing LinqToDbmlRunner

2008-05-14T13:12:00.001Z

I've created a tool based on my previous post. This should allow scripting of dbml changes in a team environment.

The project is hosted at http://code.google.com/p/linqtodbmlrunner/ Enjoy!

Scripting changes to Linq To SQL DBML files

2008-05-12T15:14:00.004Z

I've been playing with Linq To SQL and have been getting very bothered by the method MS expect us to use to develop the mappings. You can't regenerate your dbml schema without overwriting the changes you made in the designer. Not only this, but your changes won't merge well with changes made by other developers on your team when it comes to check-in time as they are all in xml files.

The process that is required for safe, team development that I've come up with is this:

1. Update database schema (using something like migratordotnet) 2. Generate just the dbml file from schema using sqlmetal.exe 3. Run a change script of some kind against the dbml file to update it to include your custom mappings. 4. Generate the classes from the dbml file using sqlmetal.exe

Step 3 is the missing link. You could use xslt, xpath or Linq to XML. You could even generate change scripts using the designer and MS XML Diff. But the easiest thing to use is a Linq To XSD program and put it in the post build events for a project that contains the generated XSD file as an embedded resource. Linq To XSD gives you typed control over the dbml file in C#.

This lets you write something like this in your script:

var dbml = XRoot.Load(@"..\..\GdwDb.dbml");
foreach (var table in dbml.Database.Table)
{
foreach (var column in table.Type.Column.Where(c => Char.IsLower(c.Name[0]))
{
column.Name = column.Name.ToUpper()[0] + column.Name.Substring(1);
}
}
dbml.Save(@"..\..\GdwDb.dbml");

This snippet will update all camelCase field names to PascalCase ones. Of course you can do anything, like creating inheritance mappings, removing prefixes etc. One you write a couple of helper methods you should be away.

I would have thought a similar process could be used for NHibernate and Entity Framework, although I know some people prefer to work backwards from code to database.

You can download the latest version of Linq To XSD from here and download the Linq classes I generated for the Dbml xsd here. Include that in a project with a reference to Microsoft.Xml.Schema.Linq.DLL and you can create your own mapping update script.

Info on using SqlMetal to generate the dbml file, and the appropriate classes can be found on Ben Hall's blog here.

I'll upload a demo project later, as this could all be a little confusing.

Storing ViewState in memcached part 2

2008-05-06T14:16:00.004Z

If you have access to the source code of the web app, you can also create inherit from Page and add the following two methods. I prefer the HttpModule myself, but this is pretty convenient. You'll still need to set up the memcached settings in web.config protected override void SavePageStateToPersistenceMedium(object state) { var key = Guid.NewGuid().ToString(); BeIT.MemCached.MemcachedClient.GetInstance("ViewStateCache").Add(key, state); base.SavePageStateToPersistenceMedium(new Pair("Key", key)); } protected override object LoadPageStateFromPersistenceMedium() { var pair = (Pair)base.LoadPageStateFromPersistenceMedium(); return BeIT.MemCached.MemcachedClient.GetInstance("ViewStateCache").Add(key, pair.Second); }

Storing ASP.NET ViewState in memcached

2008-05-04T11:40:00.017Z

This has been deprecated! Please see this post After reading various old blog posts about moving the ViewState to the end of the page using an HttpModule, I decided to create an HttpModule to store the ViewState in memcached, eliminating the need to send the data back and forth to the client and stopping prying eyes from seeing the ViewState. Halfway through writing the HttpModule, I discovered that Philip Rieck had already written a pluggable ViewState manipulation HttpModule, so I made a memcached plugin for it using the BeITMemcached client. To get this up and running on your site, download the latest source from http://code.google.com/p/memcachedviewstatehttpmodule/ and follow these steps 1. Copy all the code from the Release folder into your web sites bin directory 2. Merge the xml below into your web.config and edit the memcached server addresses <configuration> <configSections> <section name="beitmemcached" type="System.Configuration.NameValueSectionHandler" /> </configSections> <appSettings> <add key="viewRestateStore" value="Assembly"/> <add key="viewRestateStoreType" value="MemcachedViewStateStore.MemcachedViewStateStore,MemcachedViewStateStore"/> </appSettings> <beitmemcached> <add key="ViewStateCache" value="127.0.0.1" /> </beitmemcached> <system.web> <httpModules> <add name="viewrestate" type="NFission.Web.ViewRestateModule,NFission.Web.ViewRestate"/> </httpModules> </system.web> </configuration> 3. That should be it. A couple of words of warning. I haven't used this in production yet, but will hopefully do so sometime in the next week. Also, I haven't distibuted the source for Philip Riecks HttpModule as I don't know the licence for it. I had to make a couple of minor changes to get it to work correctly

I commented out the GZip stuff as I'm not using it
changed line 72 of VsInFilter from Buffer.BlockCopy( this.dataBuffer, (int)position, buffer, offset, count); to Buffer.BlockCopy( this.dataBuffer, (int)position, buffer, offset, realCount);).
changed line 16 of VsFilter from private const string ViewstateKey = "name=\"__VIEWSTATE\ value=\"";" to private const string ViewstateKey = "id=\"__VIEWSTATE\" value=\"";

I'm trying to get in touch with Philip to clear up the licence.

How To Enable TransactionScope commands for unit tests

2008-04-16T10:24:00.002Z

Something plundered from elsewhere on the internet, no doubt. Also insecure, no doubt.

Go to Administrative Tools->Component Services
In the Component Services window, go to Component Services, and right click on "my computer". Select "properties".
Click the MSDTC tab, then click the 'Security configuration...' button.
Set the settings to look like this:

Quickly navigating solution explorer in Visual Studio

2007-11-05T10:26:00.000Z

First post!11!1!!1

Often in Visual Studio I find scrolling around the solution explorer can be a bit of a hassle - if you are moving between files in different projects it can take some time to find the file you want. In order to cut out this annoyance, I have this handy tip:

Install launchy (http://www.launchy.net/), a keystoke application launcher, and get it to index .cs files in your solution directory. Now all you need to do to get to a file in Visual Studio is hit alt+space and type the file name and voila!

Its probably also worth enabling "Options -> Projects and Solutions -> Track active item in solution explorer" otherwise Visual Studio gets left behind.