Scott Hanselman

VS2008 and .Net 3.5 SP1 Beta - Should You Fear This Release?

May 12, '08 Comments [23] Posted in ASP.NET | ASP.NET Dynamic Data | ASP.NET MVC | Learning .NET | Musings | Screencasts
Sponsored By

The Beta of .NET 3.5 and VS2008 SP1 is out. I'm sure everyone is blogging the heck out of it, so I'll try to add my own specific kind of value. There's fixes, many improvements (some subtle, some dramatic), and some new technology.

Should You Fear This (Beta) Release?

Maybe a little bit. Don't be afraid of the new assemblies or the bug fixes, I have found them to be very good and have no stability problems, but this Service Pack Installer might cause you some trouble in this beta, especially if you already have beta stuff installed over the top of VS2008RTM (the original version). It'll be correct when it releases later this summer.

Now, if you're going to decide to install a Beta of a Service Pack, do read the ReadMe and ScottGu's post. Go ahead, I'll wait here. OK, if you're already running some beta stuff, like the Silverlight Tools for Visual Studio, you'll want to uninstall that FIRST, otherwise you'll get an error and be upset. This will also be figured out when the SP actually releases, but be warned.

In this release, I have found the Progress Bar is wonky on XP and 2k3. It might say failed, and then succeed. Watch for the final message, not the bar. This will be fixed. Be aware also that it can be a long-ass install so get a coffee and pillow.

On the developer side, this is much more than just a "tightening of the screws" release, as it also kind of breaks a few rules and introduces some new stuff I've been personally waiting for.

What's New?

video-292ASP.NET Dynamic Data, ADO.NET Data Services ("Astoria"), ADO.NET Entity Framework (LINQ to Entities) are all additive. ScottGu has an extensive post detailing all the interesting bits.

Here's my explanation on ASP.NET Dynamic Data and why you might care. I've been doing talks around on ADO.NET Data Services, sharing my thoughts on how its RESTfulness relates to ASMX/SCF Web Services and WS-*.*, and I'm going to dig deeper into it this year because I think it's pretty sweet, especially if Ayende and I (mostly him) get it running over NHibernate. The ADO.NET Entity Framework can be plugged in underneath both ASP.NET Dynamic Data and ADO.NET Data Services, so taken as a whole, the data story is starting to be a lot clearer and more useful and I'm continuing to work/push for all this stuff over POCO (Plain ol' CLR Objects) whenever possible and/or appropriate.

For this release I've personally recorded seven (whew!) How-To Videos on ASP.NET Dynamic Data and the Dynamic Data Samples can be downloaded here. They videos and samples are short, sweet and to the point. I'm also hoping to do more integration-style videos showing all this stuff working together in something of a larger context.

There's also a bunch of fixes like VS2008 starts faster and the IDE is snappier in spots, the WPF Designers are beefed up, and a bunch of editor/designer bugs have been squashed.

Don't Mess With My GAC, Dude!

UPDATE: Patrick Smacchia, who certainly knows his product better than I, has posted a follow up with more detailed information about what's changed in .NET 3.5 SP1.

On the .NET 3.5 side of things, since this is an SP (Service Pack), yes, some stuff goes in your GAC and gets changed. However, the changes are completely additive. That means if an API's method signature has changed, that's a bug that we need to fix. It will be a fully compatible service pack release. It shouldn't break any of your existing code.

There's roughly 50 new or modified types throughout or about 80 to 100 if you count Dynamic Data. On the WPF side, there's a bunch of new stuff in System.Windows.Effects that will make WPF people happy. There's a new WebBrowserControl in System.Windows.Controls that's like "whew! about time!" as well as new DataBinding support in System.Windows.Data.

When folks tell me that things have changed, I personally don't always take their word for it, and would rather dig in and find out for myself what's changed in what assemblies.

As far as I know there are four major tools for doing Assembly Differences (some have been added since I blogged about Assembly Differences last year):

  • Libcheck - Over three years old, obscure, command-line, confusing, but still useful. I used this when diffing .NET 2.0 and 3.5 a while back.
  • Framework Design Studio - A recent tool by Krzysztof Cwalina, a personal hero of mine, along with his helpers Hongping Lim and David Fowler, it has both a GUI and a command-line option.
  • BitDiffer - This new tool from Greg Ennis is very polished and includes both a GUI and Command Line version. If it supported drag-and-drop it'd be even better. Unfortunately it crashed while I was trying to diff these assemblies.
  • NDepend - Last, but not least, a favorite tool of mine that I've blogged about and podcast about has been dramatically updated to do Assembly Diffs as well in a very cool way. The tool is very polished and lets you compare projects, dlls, supports drag and drop and is generally polished and wonderful, so I ended up using this tool.

So, Libcheck is frustrating, FDS is close, but only creates a text-based report on what's changed, BitDiffer crashed, and while NDepend does have some weird painting issues, but I was able to create this diagram with NDepend:


Here's a close up (you can get the whole picture by clicking on it). In this example I was hovering over a method that was blue, and got a tooltip indicating that a method has changed in that type.


I created the graphic by getting the DLLs from the GAC on a .NET 3.5 machine and the DLLs from a .NET 3.5 SP1 machine. Then I put them in two different folders on a machine that *did not have .NET 3.5 on it* to avoid any wackiness. Then I ran this Code Query Language (CQL) query:


The results of that query were displayed in a number of ways, the most interesting one being the Metrics View. I set the metric to show the # of methods. In this example, 1 method = 104 pixels. Blue methods are CHANGED methods. Then I took 2 screenshots while scrolling horizontally and stitched them together in Paint.NET.

You'll see when you look at the diagram that there's some fixes here and there, but from a visual 20,000 foot view, perspective, nothing major that looks like more than bug fixes or smells of a systemic issue. Focusing on the Bug Fixes, it's onesy-twosy, bug fixes within a method here and there. The complete list of fixes is here and here.

BTW: I'm a huge NDepend fan, but I'm not associated with company at all, other than being a fanboy.

My Conclusion

Now that you've been both educated and warned, if you're not willing to wait a bit, you can download the VS2008 SP1 BETA here. Remember, this is a BETA of a Service Pack, so I'm personally continuing to install only on Virtual Machines that have VS2008 RTM.

If you have anything Beta that's layered on top of VS2008, like the Silverlight Tools, or something like that, back those off first, before installing this SP, and again, read the readme, as most known to-be-fixed issues are called out there already.

As ScottGu says:

"It will be a fully compatible service pack release.  We plan to ship the final release of both .NET 3.5 SP1 and VS 2008 SP1 this summer as free updates. "

So I'm feeling good about the final release later this year. But, for this beta, I don't want to take any chances with installer issues so it just seems prudent that I keep my primary machine on the initial version of VS2008/.NET3.5 until this SP1 is released. Have fun!

Related Links

Technorati Tags: ,,

About Scott

Scott Hanselman is a former professor, former Chief Architect in finance, now speaker, consultant, father, diabetic, and Microsoft employee. He is a failed stand-up comic, a cornrower, and a book author.

facebook twitter subscribe
About   Newsletter
Sponsored By
Hosting By
Dedicated Windows Server Hosting by SherWeb

The Weekly Source Code 26 - LINQ to Regular Expressions and Processing in Javascript

May 10, '08 Comments [11] Posted in ASP.NET | Javascript | LINQ | Silverlight | Source Code
Sponsored By

I've been getting more and more interested in how folks extend their applications using plugins and things. In my new ongoing quest to read source code to be a better developer, Dear Reader, I present to you twenty-sixth (half a year!) in a infinite number of posts of "The Weekly Source Code."

Sometimes when I read code, I kick myself (mentally) and say "Man, I should have thought of that!" Then I realize I'm not nearly as good a programmer as I think I am, and then I just let the source just wash over my brain.

Here's some source by smart people I've been reading this week that I should have thought of. ;) Coincidentally they are both examples of languages ported or re-imagined in another language.

"Processing" in JavaScript

When I say "Processing" I mean the open-source Java-based visualization language from Jeff calls it "more akin to sketching than coding" while I say it's sketching with code! Jeff yearns: "for the day when web pages are regularly illustrated with the kind of beautiful, dynamic visualizations that Ben Fry creates."


Well, John Resig, arguably already considered one of the best JavaScript coders on the planet after he gave us the tour de force that is JQuery, has ported Processing to Javascript and gives us Processing.js.

You can interact with it in two ways. First, as an elegant and tight Javascript API:

var p = Processing(CanvasElement);
p.size(100, 100);
p.ellipse(50, 50, 50, 50);

Or, you can tunnel the actual Processing language like this:

Processing(CanvasElement, "size(100, 100); background(0);" + "fill(255); ellipse(50, 50, 50, 50);");

clockThis release is specifically targeted to Firefox3, Opera 9.5 and the Webkit Nightlies (Safari) - all unreleased, beta browsers. I'm going to try it under the DLR with Javascript in Silverlight. Heh heh.

Here are his demos. Remember, these don't work in IE7.

There's a load of demos, but here's a powerful one. A working clock in 17 lines of code.

void setup() {
size(200, 200);
void draw() {
// Angles for sin() and cos() start at 3 o'clock;
// subtract HALF_PI to make them start at the top
ellipse(100, 100, 160, 160);
float s = map(second(), 0, 60, 0, TWO_PI) - HALF_PI;
float m = map(minute(), 0, 60, 0, TWO_PI) - HALF_PI;
float h = map(hour() % 12, 0, 12, 0, TWO_PI) - HALF_PI;
line(100, 100, cos(s) * 72 + 100, sin(s) * 72 + 100);
line(100, 100, cos(m) * 60 + 100, sin(m) * 60 + 100);
line(100, 100, cos(h) * 50 + 100, sin(h) * 50 + 100);

His code leans heavily on the Canvas which is why IE7 doesn't work. Much of the processing.js file is mapping from one API (the processing API) to Javascript constructs, usually canvas ones. For example, making a point(x,y) is:

  p.point = function point( x, y )
var oldFill = curContext.fillStyle;
curContext.fillStyle = curContext.strokeStyle;
curContext.fillRect( Math.round( x ), Math.round( y ), 1, 1 );
curContext.fillStyle = oldFill;

Note the rectangle that is 1 by 1. That's funny, but that's the life an API mapper. Remind me someday to tell you, Dear Reader, how I got filled pie charts working on an Original Palm Pilot that not only didn't support Put/GetPixel but didn't have floating point math. That was a hoot.

Anyway, one really good example of this guy's clean cleverness is the triangle function. Remember, this is a processing function and he's not only got to implement it, but also make the building blocks for doing it cleanly.

To start:

  p.triangle = function triangle( x1, y1, x2, y2, x3, y3 )
p.vertex( x1, y1 );
p.vertex( x2, y2 );
p.vertex( x3, y3 );

Obvious, right? Well, not really, considering that the 2D Canvas doesn't have any of those three higher-level methods. Begin and EndShape are fairly clean. However, he had to implement a nice Fill, Stroke and ClosePath to do this cleanly.

  p.beginShape = function beginShape( type )
curShape = type;
curShapeCount = 0;

p.endShape = function endShape( close )
if ( curShapeCount != 0 )
curContext.lineTo( firstX, firstY );

if ( doFill )

if ( doStroke )

curShapeCount = 0;
pathOpen = false;

if ( pathOpen )

It's about four layers deep, each primitive building on the next until he gets a nice clean triangle implementation, but then he can use it for quad() and it the same method handles bezierVertex as well. It would do you well to FireBug your way through his code. It's a wonderful fun way to re-learn Javascript from a gentleman who knows what he's doing.

LINQ to RegEx and Fluent Regular Expressions

I was trying to re-re-re-learn Regular Expressions again this week for a small task. It's funny how Regular Expressions are the first thing to leave my brain even though there are a bunch of Regular Expression Tools out there. Josh Flanagan came up with a Fluent Interface for Regular Expressions like:

Regex socialSecurityNumberCheck = new Regex(@"^\d{3}-?\d{2}-?\d{4}$");

would look like this:

Regex socialSecurityNumberCheck = new Regex(Pattern.With.AtBeginning 

It took me a second to like this. OK, it took me a while. Breathe for a minute, and read it out loud. It kind of makes sense, actually, although there is a reasonable argument against in the comments of Josh's post:

[It] strikes me that a user of this library needs to learn a fairly complex syntax which is almost as far from "plain english" as regex, when they could simply learn how to do regex.

Sure, but it's fun to try new things. If you look a his source, it's really just a really smart string concatenator. I think it would actually be a very interesting way to teach or learn regular expressions, especially if you're a casual RegEx'er like me.

Krzysztof Ko┼║mic created a similar API in 2007. His fluent interface over RegEx looks like this:

Pattern pattern = Pattern.Define().

Then Roy Osherove took Josh's API further and took Josh's Fluent Interface to RegEx from 2006 and applied a LINQ query syntax , creating in the process, LINQ to Regex.

Here's Roy's example:

public void FindEmailUsingPattern()
var query = from match in
where match.Word.Repeat.AtLeast(1)
select match;
foreach (var match in query)

After the "from match in", the simple heart of it is Roy's static Against() call that returns a RegexQuery that is IEnumerable of Match, thereby supporting the foreach later on:

namespace Osherove.LinqToRegex 
public class RegexQuery : IEnumerable
private readonly string input;
private object lastPatternRetVal;
private RegexQuery(string input)
this.input = input;
public static RegexQuery Against(string input)
return new RegexQuery(input);

private string _regex;
public RegexQuery Where(Expression<func><pattern,bool> predicate)
_regex = new PatternVisitor().VisitExpression(predicate).ToString();
return this;

public RegexQuery Select<t>(Expression<func><pattern,t> selector)
return this;
#region IEnumerable Members

IEnumerator IEnumerable.GetEnumerator()
return ((IEnumerable)this).GetEnumerator();

public IEnumerator GetEnumerator()
MatchCollection matches = Regex.Matches(input, _regex);
foreach (Match found in matches)
yield return found;

You can find all the source for Roy's project up at his project site and Josh's source is on his blog. It is worth noting, though that you can combine LINQ Queries with Regular Expressions without any tricks because Matches are returned in a MatchCollection an LINQ loves things that are IEnumerable.

You can use LINQ projections to pull objects out of a collection of matches like:

    List<yourType> = (from Match m in matches
select new YourType
Id = m.Groups[1].Value,
Something = m.Groups[2].Value

So, we've got two sides of the coin here. First, the creation of the Regular Expression. That can be the standard way, or with a fluent interface. Either way, you end up with a string. Second, you've got the extraction of the information. Most often you'll care about the MatchCollection that comes back. You'll usually want to pull information out, so while you're foreach'ing your way over the collection, you can use LINQ to create an object projection that's chopped up and sorted and grouped all with one query, regardless of how you created the query in the first place.

Choice is good.

About Scott

Scott Hanselman is a former professor, former Chief Architect in finance, now speaker, consultant, father, diabetic, and Microsoft employee. He is a failed stand-up comic, a cornrower, and a book author.

facebook twitter subscribe
About   Newsletter
Sponsored By
Hosting By
Dedicated Windows Server Hosting by SherWeb

TypeConverters: There's not enough TypeDescripter.GetConverter in the world

May 10, '08 Comments [12] Posted in Source Code
Sponsored By

While reading source today, I saw some code in the wild today that looked roughly like this (not the actual code):

type = typeof(T);
if (type == typeof(Boolean))
returnValue = (T)((object)Convert.ToBoolean(value));
else if (type == typeof(String))
returnValue = (T)((object)value);
else if (type == typeof(Int16))
returnValue = (T)((object)Convert.ToInt16(value));
else if (type == typeof(Int32))
returnValue = (T)((object)Convert.ToInt32(value));
//...and on and on for a dozen+ types

You get the idea. This isn't that uncommon, I've seen it more of than not. The person who is writing it usually knows it's bad 50% of the time, but it's always a trade-off between figuring out the right search query (a tough one, in this case!) and knowing what to look for in MSDN.

This brings up one of my favorite classes in the BCL, the TypeDescriptor class. A method like this:

public static T GetTfromString<T>(string mystring)
var foo = TypeDescriptor.GetConverter(typeof(T));
return (T)(foo.ConvertFromInvariantString(mystring));

...would allow all that switch/if/else to go away replaced by:

bool b = GetTfromString("true");

You'd probably want to expand the method with checks to see if T was in fact, System.Type, or System.String, but you get the idea.

image There are lots of standard converters like EnumConvertor, ColorConvertor, and more, all waiting for you in System.ComponentModel as seen in the image at right.

Not only that, but you can make your own TypeConverters and spread the love. They are used extensively when displaying types in Property Grids, in WinForms, or in XAML. You can also use them in PowerShell as a better way to present your objects as strings when "casting" in PowerShell.

Jesse Liberty has a fine example showing how to use Type Convertors in Silverlight to enable you to put complex types in XAML attributes.

From MSDN:

To implement a simple type converter that can translate a string to a [Type]

  1. Define a class that derives from TypeConverter.

  2. Override the CanConvertFrom method that specifies which type the converter can convert from. This method is overloaded.

  3. Override the ConvertFrom method that implements the conversion. This method is overloaded.

  4. Override the CanConvertTo method that specifies which type the converter can convert to. It is not necessary to override this method for conversion to a string type. This method is overloaded.

  5. Override the ConvertTo method that implements the conversion. This method is overloaded.

  6. Override the IsValid method that performs validation. This method is overloaded.

That's it, and when that Type was decorated with an attribute like:


Then that new TypeConverter would be picked up and used by the line of code I showed before. Much better than a switch or pile of if/elses, I think.

About Scott

Scott Hanselman is a former professor, former Chief Architect in finance, now speaker, consultant, father, diabetic, and Microsoft employee. He is a failed stand-up comic, a cornrower, and a book author.

facebook twitter subscribe
About   Newsletter
Sponsored By
Hosting By
Dedicated Windows Server Hosting by SherWeb

Introducing RockScroll

May 9, '08 Comments [64] Posted in Musings | Tools
Sponsored By

When I came to Microsoft I saw a really cool tool being used internally and immediately wanted to work with the author, Rocky Downs (who is blogless, but exceeding talented), to get it released. Fast forward to now, plus an installer, and here it is.

The basic (as in "only") idea is that RockScroll extends the scrollbar in Visual Studio to show a syntax highlighted thumbnail view of your source. This is really useful for those excessively long source code files you know you have. It's just one DLL and you can turn it off from Tools|AddIns just by un-checking the checkbox.



works-on-my-machine-starburst Works On My Machine Disclaimer: This is released with exactly zero warranty or support. If it deletes files or kills your family pet, you have been warned. It might work great, and it might not. It hasn't been tested against the myriad of other VS Add-Ins, but it works on my machine in both VS2005 and VS2008. It does look a little odd next to Resharper, a tool that also adds a scrollbar. Good luck.

A reader, Hans, says: "I have found that (not suprisingly) RockScroll works very well with split windows, if you bind a shortcut key to the "Window.Split" command." for those of you having trouble with split windows.

About Scott

Scott Hanselman is a former professor, former Chief Architect in finance, now speaker, consultant, father, diabetic, and Microsoft employee. He is a failed stand-up comic, a cornrower, and a book author.

facebook twitter subscribe
About   Newsletter
Sponsored By
Hosting By
Dedicated Windows Server Hosting by SherWeb

Hanselminutes Podcast 110 - Microsoft Research: Spec#

May 8, '08 Comments [3] Posted in Podcast | Programming
Sponsored By

My one-hundred-and-tenth podcast is up. This episode was recorded at the ALT.NET Open Spaces Conference in Seattle a few weeks back. I got to sit down with two gentlemen from Microsoft Research, Mike and Rustan, and talk about Spec# after their presentation at the conference.

What's Spec#? Currently it's a new language that is essentially extensions to C# 2.0. For example:

class ArrayList { public virtual void Insert(int index , object value)
requires 0 <= index && index <= Count;
requires !IsReadOnly && !IsFixedSize;
{ . . . }

See all that new stuff BEFORE the first curly brace? That's where you specify preconditions that you want enforced. You can also specify postconditions:

ensures Count == old(Count) + 1;
ensures value == this[index ];
ensures Forall{int i in 0 : index ; old(this[i]) == this[i]};
ensures Forall{int i in index : old(Count); old(this[i]) == this[i + 1]};

There's LOTS more, including a pretty extraordinary static analysis that runs in the background, kind of like the grammar checker in Word. Word has the spelling checker that turns errors red with squiggly underlines, and the grammar checker that turns grammatical errors green with squiggly underlines. Spec# will turn static errors into squiggles while you type.

They've even added specifications for the Base Class Library (BCL) that have previously only existed in documentation.

UseMscorlibContracts - Microsoft Visual Studio

BTW, if you want to see horribly shaky cellphone video of their presentation, you can see it at Greg Young's blog. He's posted recently about Spec# as well. If you don't like videos or podcasts, you can also read the Spec# Research Paper (PDF).

How can you get involved? We need to convince that folks on the C# team that Spec# is a valuable thing and that we'd like it sooner than later. Go over to Charlie Calvert's blog and email him that you want Spec# and why!

Subscribe: Subscribe to Hanselminutes Subscribe to my Podcast in iTunes

If you have trouble downloading, or your download is slow, do try the torrent with µtorrent or another BitTorrent Downloader.

Do also remember the complete archives are always up and they have PDF Transcripts, a little known feature that show up a few weeks after each show.

Telerik is our sponsor for this show.

Check out their UI Suite of controls for ASP.NET. It's very hardcore stuff. One of the things I appreciate about Telerik is their commitment to completeness. For example, they have a page about their Right-to-Left support while some vendors have zero support, or don't bother testing. They also are committed to XHTML compliance and publish their roadmap. It's nice when your controls vendor is very transparent.

As I've said before this show comes to you with the audio expertise and stewardship of Carl Franklin. The name comes from Travis Illig, but the goal of the show is simple. Avoid wasting the listener's time. (and make the commute less boring)

Enjoy. Who knows what'll happen in the next show?

About Scott

Scott Hanselman is a former professor, former Chief Architect in finance, now speaker, consultant, father, diabetic, and Microsoft employee. He is a failed stand-up comic, a cornrower, and a book author.

facebook twitter subscribe
About   Newsletter
Sponsored By
Hosting By
Dedicated Windows Server Hosting by SherWeb

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