0 comments suggest edit

This past weekend Akumi and I headed over to the downtown flower mart to purchase some plants. We came home with a couple money trees and some bamboo for my office.

Bamboo

There’s something about Bamboo that is very soothing for me. Sometimes if I look at it long enough, I can swear that the leaves are moving gracefully, twisting upwards.

[Listening to: Undo - Björk - Vespertine (5:38)]

0 comments suggest edit

As you’ve probably heard, Google is testing AdSense within RSS on a single site.

I love the idea of being able to include a Google AdSense ad within my RSS feed and watching the money roll in. At least in theory. Once ads proliferate in RSS feeds, ad-blockers within RSS aggregators will proliferate. Just look at this prototyped ad-blocker for RSS Bandit that Torsten developed.

Another issue that comes to mind is that the audience for web ads is not the same audience for rss feeds. I have a theory that those who find my site via a Google search are more likely to click on an ad than someone who is subscribed to my RSS feed?

Why is that? It boils down to the fact that a subscriber via RSS isn’t likely to be in the middle of a search for a particular piece of information while reading my blog as would a Google user. Rather, I suspect that he or she is a friend or a geek wondering if I’ll finally be inspired someday to write something worthwhile because I once wrote something interesting enough some time ago that encouraged this person to add me to his or her aggregator and now only inertia keeps this person from unsubscribing. But I digress.

Unfortunately, I don’t have any hard (or even soft) numbers to back this claim up because I don’t have ads in my RSS feed. Recently I was linked to on Scoble’s link blog which jumped up the number of aggregator views, while my web views remained constant. Likewise, my tiny advertising revenue for that day stayed constant as would be expected because those who read my blog via an aggregator aren’t likely to take their time to look at the actual site and click on an ad. I’d love to know what the effect would have been had my feed included an ad.

When I post something that users find via Google, my web views jump as does my advertising revenue, which makes perfect sense. I’ll be watching with interest to see how Google’s experiment with Longhorn blogs does and if they end up rolling it out to everyone. I’ll certainly experiment with it to see if my conjecture is correct unless I get a backlash from commenters saying they’d rather not see it in personal blogs. Thoughts?

As for my advertising revenue in case you were wondering, I’m strictly a small fry, though I do make more than enough to pay for hosting.

[Listening to: 3-2-1 Fire! - Fat Boy Slim - Fatboy Slim: Live On Brighton Beach (4:34)]

company culture 0 comments suggest edit

I saw this post on Don Park’s blog…

My perspective on this latest hot topic is that I think companies should act only in interest of itself, the Microsoft shareholders in this case, and not anyone else, including its employees. So I think the discussion should have been about whether leaders of Microsoft made the right decision in the interest of its shareholders.

I don’t understand this mode of thinking. Should people in general act purely on self interest and not for the common good? Why should corporations be any different? After all, they are made up of people, both shareholders and employees. It seems not all companies agree as Boeing, Nike, Coors, HP, and others supported the bill.

I suppose Microsoft would have continued to support the bill if it was rewritten to state that gays are not protected under anti-discrimination laws to own stocks in major corporations. Oh then you’d see some major rallying around gay rights.

I understand that having a corporation get behind a political cause is a touchy situation. Whose values do you support? Your employees? Your shareholders? America at large? But it’s clear that this bill supports values Microsoft already holds dear. Microsoft has in the past supported this bill and has shown itself to be progressive in its own policies.

And Ballmer’s worry of discriminating against anti-gay bigots is a real piece of work. Here’s a snippet.

What message does the company taking a position send to its employees who have strongly-held beliefs on the opposite side of the issue?

As Shelley eloquently points out, this bill doesn’t deny rights to others. And the message would have already been sent by the fact that Microsoft’s own policies are reflected in this bill. So if you’re worried about the bigots feelings Mr. Ballmer, are you going to change your policies to reflect that as well?

We’ve all seen what acting purely on corporate self-interest produces. Oh, Enron springs to mind, sweat shops, destruction of the environment, and countless other examples. Corporations have all the rights of a person without many of the implicit responsibilities. Corporations are members of the community. It’s time they start acting like it.

0 comments suggest edit

UPDATE: Mea culpa! Maurice pointed out that (except for the casing) my original expression WAS correct. I only needed the RegexOptions.SingleLine option. I didn’t need to add the (\s|\n) everywhere. Here’s a corrected post. Thanks Maurice!

Last time I talked about matching HTML with regular expressions, I published a regular expression with a couple small bugs. The first bug was not my fault, but rather the fault of the rich text editor that comes with .TEXT. It was being overly “helpful” when I tried to edit the post and uppercased some of the code. As you know, “\S” is much different than “\s” to a regular expression.

The second bug is entirely my fault and I write this as a confession and to provide a fix. You see, I assumed (and you know what happens when we assume) that complete tags tend to be on a single line. Well that’s not the always the case. You might encounter something ugly like this:

<div     id = "blah" alt=" man    this is ugly html "     >     fire this guy... </div> 

The expression I had posted wouldn’t have matched the div tag sitting in plain sight there so I went in there and corrected that sucker all by itself. It requires using the RegexOptions.SingleLine option so that the . character matches \n. Here’s the expression reprinted (with correct casing) for your reference.

</?\w+((\s+\w+(\s*=\s*(?:”.*?” ’.*?’ [\^’”>\s]+))?)+\s* \s*)/?>

The main difference is now I’m including \n anywhere I’m matching whitespace (via \s). In order for this to work, you need to use the RegexOption SingleLine. Here’s a code snippet that uses this expression to match the above html.

string html = “<div\n\tid=\“blah\” alt=\” man\n”

    + “\tthis is ugly html \”\n”

    + “\t>”

    + “fire this guy…\n”

    + “</div>”;

 

Regex regex = new Regex(@”</?\w+((\s+\w+(\s*=\s*(?:””.*?””|’.*?’|[\^’””>\s]+))?)+\s*|\s*)/?>”, RegexOptions.Singleline);

 

MatchCollection matches = regex.Matches(html);

Console.WriteLine(“…..Original Html……”);

Console.WriteLine(html + Environment.NewLine);

Console.WriteLine(“…..Each Tag……”);

foreach(Match match in matches)

{

    Console.WriteLine(“TAG: “ + match.Value.Replace(“\n”, “ “));

}

This produces the output:

.....Original Html......<div    id="blah" alt=" man   this is ugly html "   >fire this guy...</div>.....Each Tag......TAG: <div     id="blah" alt=" man     this is ugly html "     >TAG: </div>

So sorry about that. Hope this one treats you better.

tech 0 comments suggest edit

I’m chatting with my buddy Micah who will ask me a question and then annoyingly answer it himself immediately after. For example, he asked me how FireFox picks up an icon from web pages. I replied that there’s a favicon protocol but that I didn’t know the details. He then proceeded to explain how to format the link tag properly.

Apparently the latest version of Trillian will highlight words with links to their definition in Wikipedia. So as you chat with your friends, you become a living breathing dictionary. That’s a rather nice feature for IM conversations.

0 comments suggest edit

Yesterday I wrote about a neat Trillianfeature that highlights words with their Wikipedia definition.

Today Dimitri told me via MSN Messenger that I caused him to buy some music online. While I was out for a walk, I had left my iTunes playing. I’m using the new 7.0 version of MSN Messenger which allows you to automatically publish what you are currently playing as your status message. Apparently something I was listening to (as I have fantastic taste in music) caught his attention enough that he clicked on the song and bought it.

You have to love these small details that make software more interesting to use. Trillian’s wikipedia integration puts knowledge at your fingertips (to use a much bandied about Microsoft phrase), while Messenger connects us via the universal language of music. I love checking Messenger to see what people are playing at any given moment.

So Dare (currently playing something by, whoelse?, 50 cent), much kudos to you and your team on the latest messenger as well as for helping unleash the inner consumer whore within each of us. :)

[Listening to: Athena - Tiësto - Parade of the Athletes (6:17)]

0 comments suggest edit

CancunPlanning a vacation is proving to be a challenge now that I’m independent. Last year, Akumi and I decided that we were going to England for two weeks this April. Well April has come and is about to go, and we’re no closer to Jolly old England (and Old Trafford).

With my project ramping up, I just don’t think we can do two weeks of vacation in a row. Of course, this isn’t really fair to my wife as I’m still going to Burning Man at the end of August. We need a shorter vacation for the two of us, so I thought of our good neighbors next door. Mexico!

Having surveyed her Mexican expatriate coworkers, my wife reported that they unanimously declare Cancun as the place to go. Armed with that information, we nearly jumped on an all-inclusive deal that would have us flying out of Los Angeles this Saturday. But good sense (or something resembling it) took hold and we stopped ourselves (for the moment).

I’m skeptical of all-inclusive deals because of the simple fact that once they have you there, they don’t have much incentive to step up the quality of food. Imagine travelling all the way to Mexico for bad Mexican food. ¡Aye Carumba!

So now we’re looking at November as a good time to enjoy the turquoise beaches of Cancun and Tulum. However, if you’ve been there and had a great time (or not), I’d love to hear about it. ¡Muchas Gracias!

[Listening to: Dreaming (Percussion Mix) “Vamos a Jugar en el Sol!” - Ruff Driverz Pres. Arrola - Trance: A State of Altered Consciousness (5:25)]

0 comments suggest edit

One thing I’ve found annoying at times with the DataGrid control is there’s no way to specify a title to be displayed above the headers. Being lazy, I often resorted to adding a label above the data grid followed by a br tag.

But no longer! I wanted to have the title display in its own row within the data grid structure so I created a custom data grid control that does just that. See the example below for how it renders.

This is the Title
Column 1 Column 2 Column 3
See Spot Run
Run Spot Run!
Flee the Maniacs

The key to this is to override the OnItemCreated method and set my own rendering method for the header item.

/// <summary>
/// Assigns our own render method for the header item.
/// </summary>
/// <param name="e"> E. </param>
protected override void OnItemCreated(DataGridItemEventArgs e)
{
    if (ListItemType.Header == e.Item.ItemType && Title != null &&
Title.Length > 0)
    {
        e.Item.SetRenderMethodDelegate( new RenderMethod(RenderTitle));
    }
    else
    {
        base .OnItemCreated(e);
    }
}

When ASP.NET is ready to render the Header, it’ll call my method instead which is named RenderTitle.

/// <summary>
/// Renders the title as its own row.
/// </summary>
/// <param name="writer"> Writer. </param>
/// <param name="ctl"> CTL. </param>
protected virtual void RenderTitle(HtmlTextWriter writer, Control ctl)
{
    // TR is on the stack writer's stack at this point...
    writer.AddAttribute(
        "colspan",
        this.Columns.Count.ToString(CultureInfo.InvariantCulture));

    writer.AddAttribute("align", "center");
    writer.RenderBeginTag("TD");
    writer.Write(Title);
    writer.RenderEndTag(); // Writes </TD>
    writer.RenderEndTag(); // Writes </TR>

    // Now we add the header attributes we
    // copied.
    this .HeaderStyle.AddAttributesToRender(writer);
    writer.RenderBeginTag("TR");

    //Render the cells for the header row.
    foreach (Control control in ctl.Controls)
    {
        control.RenderControl(writer);
    }

    // We don't need to write the </TR>.
    // The grid will do that for us.
}

The snippet shown here will style the title row the same as the header style. In my actual control, I defined a property named TitleCssClass to enable you to define your own Css class to use to style the title row. This required me to do a bit of hacking so that the HeaderStyle gets removed from the render stack and then gets added back later when rendering the Header row. If that makes no sense, you’ll see what I mean. I’ve put the code for the control here.

0 comments suggest edit

Elbow Yesterday Akumi pointed at my elbow in shock and asked me what happened.

“What happened to what?” I replied?

“You’re elbow. It’s badly bruised.”

Sure enough my entire elbow is covered in a dark shade of reddish purple. “Huh. How about that. I have no idea.” I did go to a party the night before, but I’m pretty sure I managed to stay on my feet the entire night. I don’t remember getting injured during soccer. It just showed up.

I’ll be sure to install a web-cam in the bedroom tonight. I have a sneaking suspicion that Akumi beats me while I sleep. If she does, it’s probably well deserved as I’ve been known to elbow her in the head while we sleep from time to time…. Hmmm. I better check her head.

0 comments suggest edit

Since I mentioned that the Family Guy was no longer cancelled on my blog, I’ve noticed that Google Ad Sense has been displaying a Family Guy banner ad on my homepage.

Family Guy In case it is gone by the time you read this, the image at left is a snippet from the full banner ad. It’s colorful and looks rather nice at the top of my blog.

Now I agree that it’s a bad idea to try to game the google AdSense system for profit, but what about gaming the system for pure fun? For example, coaxing the Google Ad-Sense system to display the most interesting (or offensive) ad possible.

Well I won’t condone such behavior lest I’m removed from the program. I will merely mention the idea.

0 comments suggest edit

Before you upgrade to Community Server, be aware that they’ve turned it into a commercial product. Yes, there is a non-commercial license available, but it requires you to display the following image at the bottom of every page.

CS EULA I have no problem giving credit where credit’s due, but this is in stark contrast to the BSD license employed by .TEXT and you should be aware of this if you plan to upgrade to CS.

For example, competitors to Telligent would probably want to purchase the commerical license rather than have to prominently display their logo on your blog.

In any case, I’ll probably stick with .TEXT and start looking into what it will take to switch to DasBlog. I think I prefer the more focused approach DasBlog takes.

0 comments suggest edit

Jayson Knight has been cutting his teeth on the latest version of .TEXT now called Community Server trying to get it’s RSS and CommentAPI support to work.

Thankfully he posts his fix here for others to use.

I’m pretty sure he’d agree with me that given the chance, he wouldn’t have upgraded to Community Server so soon after its release. To be fair, it’s an ambitious release intending to integrate Forums, Blogs, and Photo Galleries all in one package. But for me personally, and I’d guess for a lot of .TEXT users, this wasn’t a necessary feature. Especially not at the cost of having existing features break.

It seems to me that the Telligent guys have released this puppy a bit too early. It’s really bothersome how much the RSS and commenting features regressed in this release. It’s not a case of simply introducing a bug that accidentally disables a feature, but instead appears to be a case of dropping the code for existing functionality. Who, besides the ever patient beta community, tested this?

Installation is still a tricky beast with this software as I’ve had difficulty getting it up on my development machine. Understand that I was able to get .TEXT installed and CS is supposed to have simplified the installation process. As many of you know, installing .TEXT is a trip to the dentist without N2O.

The reason I had chosen .TEXT in the first place was for its SQL Server support, and its the primary reason I’ll stick with it as running ad-hoc reports against my blog is quite convenient. I’ve considered DasBlog, but probably won’t switch until someone (or myself) writes a SQL provider.

One thing I like about DasBlog is that it’s focused on being a good blog engine. I think its that focus that will keep it very tight and without the annoyances of CS. So far, CS is a big disappointment.

0 comments suggest edit

What will this code write to the console?

public void SomeMethod()

{

    SqlInt32 x = 1;

 

    if(x == SqlInt32.Null)

    {

        Console.WriteLine(“Null.”);

    }

    else if(x != SqlInt32.Null)

    {

        Console.WriteLine(“Not Null.”);

    }

    else

    {

        Console.WriteLine(“Neither? WTF!?”);

    }

}

If you said “Neither? WTF!?” then you’d be correct.

Doesn’t that seem odd since x is either SqlInt32.Null or it isn’t? Is this a case of fuzzy logic? Well not exactly. This is one of those gotchas with operator overloading. The == operator is overloaded by SqlInt32 to return a SqlBoolean instead of a boolean.

This caused me a few minutes of pain this week as I was stepping through code like this and examining the values in the debugger and I thought perhaps someone had slipped me a very strong hallucinogen because there I was with a value that was not null, but was null at the same time. It was one of those mind warping “This sentence is not true” moments.

As typical, I was thinking about rebooting when on a whim I decided to use intermediate variables and realized that the == was returning something neither true nor false. Honestly, I think this is a very poor and unnecessary use of operator overloading (correct me if I’m wrong) because it hides a real gotcha underneath a very common paradigm. If you overload the == operator, you should most definitely return a bool.

In this case, it’s easy enough to fix. I should have been checking the IsNull property anyways like so:

public void SomeMethod()

{

    SqlInt32 x = 1;

 

    if(x.IsNull)

    {

        Console.WriteLine(“Null.”);

    }

    else

    {

        Console.WriteLine(“Not Null.”);

    }

}

This post by Cyrus motivated me to write about this.

0 comments suggest edit

Looking through my statistics today and I noticed that I received a lot of hits from this post about Google Maps using satellite photography.

Looking in my referrer logs, I noticed that most of the visits were rolling off of Google form users who were searching for the term “Google Satelite”. I just happened to be the fourth search result on the page.

“Cool” I thought to myself, “but they’re all misspelling “satellite”. Then it occurred to me that they weren’t the only ones who were spelling challenged. I had inadvertently misspelled “satellite” in the title of my post.

I’m probably dredging up yesterday’s news here, but it occurred to me that one could really get great search results for hot search words by misspelling them in ways that people commonly misspell them. This is the same thinking behind someone who tries to register “amazn.com” or “micrsoft.com” hoping that millions of typos will send users their way.

So if you want to be THE blog about “onomatopoeia”, start writing about your love of “onomatopeia” or “onomatopoea”. That’ll create some buzz.

I’ll have to figure out a way to register “hacked.com”.

0 comments suggest edit

UPDATE: As a commenter pointed out, the original code example did not properly demonstrate the problem with locking on the this keyword within a normal method. I have corrected this example and wrote a better example that demonstrates that this problem still exists even in a “normal” method.

Take a look at this code:

private bool isDisposed = false;
 
//... code...
~MyClass()
{
    lock(isDisposed)
    {
        if(!isDisposed)
        {
            //Do Stuff...
        }
    }
}

Hopefully you can see the problem here right away. The lock statement takes an object instance as a parameter. So what happens to the boolean isDisposed within the lock statement? That’s right! It gets boxed, meaning a new object instance is allocated and passed to the lock statement. Thus every time you lock on a value type, you’re locking on a new object.

Ok, so let’s try to fix this up a bit.

private bool isDisposed = false;
 
//... code...
~MyClass()
{
    lock(this)
    {
        if(!isDisposed)
        {
            //Do Stuff...
        }
    }
}

So is there anything wrong with this? You’ve probably seen the Microsoft examples locking on this. Well never give full trust to example code (especially as it’s unlikely you’ll add the code to the GAC) ;). Suppose this snippet is from a class MyClass. What do you think will happen with the following code:

MyClass instance = new MyClass();

Monitor.Enter(instance);
instance = null;

GC.Collect();
GC.WaitForPendingFinalizers();

You guessed it! Deadlock.

Every time you lock a .NET object, the runtime associates a SyncBlock structure to that object. Locking works by checking who owns an object’s SyncBlock when attempting to acquire a lock. Thus in the code sample above, the client code and the Dispose() Method are attempting to lock on the same object.

For a more in-depth discussion, I highly recommend Jeffrey Richter’s article Safe Thread Synchronization which is where I first learned about this subtle threading issue.

Likewise you might also take a look at Dr Gui’s Don’t Lock Type Objects post.

sql 0 comments suggest edit

In T-SQL, you can use the @@IDENTITY keyword to obtain the value of the identity column when you insert a new record. For example, the following query inserts a record into an imaginary table and returns a result set containing the ID of the inserted column.

INSERT INTO SomeTable
    SELECT Value1, Value2
   

SELECT @@IDENTITY -- LAST INSERTED IDENTITY VALUE

There’s the potential for a subtle bug here. Suppose later on, a coworker realizes that any time a record is inserted into [SomeTable] a record should also be inserted into the table [SomeTableAudit]. The simplest solution would be to add a trigger to [SomeTable] that inserts a record to [SomeTableAudit]. But in doing so, your coworker introduces a case of @@IDENTITY theft. Your original query will now return the value inserted into the IDENTITY column of the tabel [SomeTableAudit] instead of the IDENTITY value of [SomeTable] as you intended.

At this point some of you are shaking your heads muttering  

Well I never use Triggers. Triggers are bad umkaaaay.  

That’s beside the point, you never know when someone else is going to apply that trigger resulting in this unintended consequence. It pays to program defensively. The issue here is that although @@IDENTITY is constrained to the current session, it is not constrained to the current scope. Instead, use the SCOPE_IDENTITY() function which will return the last IDENTITY column value inserted in the current scope, in this case the value inserted into [SomeTable].

INSERT INTO SomeTable
    SELECT Value1, Value2
   

SELECT SCOPE\_IDENTITY() -- LAST INSERTED IDENTITY VALUE

As an aside, I’m fine with triggers in certain cases. One of the primary complaints about triggers mirrors the complaints about Aspects in AOP. Namely that triggers provide for unintended consequences that aren’t visible when examining a stored procedure. However when used sparingly for cross-cutting functionality, I think they can add a lot of benefit. Much like Aspects.

UPDATE: Steven Campbell adds a great tip.

 

Another tip I can offer is that you should not use ADO in combination with SQLOLEDB to retrieve IDENTITY values. I refer specifically to the technique of: \ myRS.AddNew\ …\ myRS.Update\ myId = myRS(“ID”)\ \ This fails to retrieve the correct ID, because (internally) the SQLOLEDB driver issues a SELECT @@IDENTITY statement to retrieve the newly created ID.