I have to admit, this post is going to be lame. I won’t be offended if you stop reading now because it is mostly for my own benefit.

Recently we moved our YouTrack server to a publicly accessible server and got it all setup with an URL like http://dev.acme.com. We then wanted to move out TeamCity server to the same box so moved both TeamCity and YouTrack into a Tomcat app server and gave them URLs like dev.acme.com/teamCity and dev.acme.com/youTrack. Sweet. Everything is going great.  We have them talking to each other.

Let’s flip them over to talk on only SSL. This wasn’t was easy as I hoped. Actually it was super simple, but finding the steps was painful. If you look for how to setup SSL with Tomcat you find a load of articles using tools and certificates we don’t have handy. We do have our wild card certificate from IIS that we can export and then the steps are easy.

Sweet. While I was in that file I commented out the Tomcat connector on port 80 so IIS could listen on that port. I’m more familiar with IIS and setting up the HTTP to HTTPS redirect was easier for me in IIS…which I did next.

All done. Everything is golden right? It works. It redirects to HTTPS if you request HTTP. Awesome…except we cannot attach files. Bummer.

We get asked questions like:

Are you using a proxy?
Please specifiy procy if you are using one, otherwise it is impossible to post the attachment.

Since we are not using a proxy I said “no” and then got another error message.

Sorry, cannot attach the image.

Luckily this was an easy fix as well. This was a fairly easy issue to find since JetBrains publicly tracks the issues with YouTrack. That was exactly out problem. We had YouTrack setup to listen on http://dev.acme.com/youTrack instead of https://dev.acme.com/youTrack. Switching that in the settings fixed our problem.

We all like color don’t we? We also love the command line right? Well, if you don’t, you should love it. There is nothing I hate more than seeing a screen full of gray text. You can provide a lot of context via color, especially in your prompt.

Take my git prompt for example.  I easily know if I have any changes in the branch, if I’m staging changes and an overview of the changes.

This is done in PowerShell. If you’re still using cmd as your shell, you should stop immediately. Especially if you use git. I highly recommend checking out posh-git. Auto-complete on git commands and branch names is awesome.

Also, you can checkout the source for this prompt. Figuring out all the changes takes some time on a large codebase, but for me it’s worth it.

We have a moderately large WebForms web application, but we are trying to transition to MVC. On the surface it sounds pretty simple. We can create controllers to do small functions of our page. we can be AJAXy if we so desire and all will be great.

Let’s give it a try.  We want to add a customer search to our master page.  Simple enough right?

All I really need is a controller that redirects to our existing search page.  No problem. Seems simple enough. Let me code up the controller in a simpler, reduced test case.

Done in 15 minutes.

OK. Now we just need to drop a tiny form that posts to our controller on our master page…wait  a minute. WebForms has a HUGE form that takes up the entire page! I can’t nest a form inside a form. This means I can’t post inside the form to anywhere but the same form.

I know  what you’re thinking. I shouldn’t post to search anyway. I should get instead. Yes. I agree, but this example is contrived. I actually came across this problem when I wanted to create a controller that merged two customers together and redirect to the surviving customer. In that case we do want to post…so bear with my invalid semantics.

Anyway. We can’t POST. hmm..I guess we have to do something like this (remember in real life I’m merging and not searching"):

[HttpGet] public ActionResult SearchViaGet(string searchTerm, bool isQuickSearch){}

It feels really nasty. What makes it even worse is we have to use Ajax to pull the values from the inputs and build the request URL. If they don’t have JavaScript enabled…oh well. No search for you!

I can’t really think of a way around this. With MVC, and without the giant WebForms form, this page would comprised of several tiny forms. For now, the no JS problem isn’t that big of a deal. The app is used by internal users only so if they don’t have JS enabled we can just laugh at them until they either turn it on, or we feel bad that they don’t know how to turn it on, and turn it on for them.

Where we do have a problem is when we try to use pages like this via our ancient mobile devices. We can’t trust JS on those bricks and if I had to bet, there would be no search for them either.

Google gets it right a lot of the time, including bringing back what I’m looking for as the first result. I might be one of the few people who actually use the I’m Feeling Lucky button on the Google Search page, but I do it in a sneaky way.  I do it from my omnibox in Google Chrome.

If you’re not using the omnibox to search in Chrome, you’re not one of the cool kids in my book.  Just type what you want and hit enter to search your default search engine.

What you might not know is you can search other sites as well. Type the domain, press tab and if you’ve already searched that domain, chances are Chrome will help you out.

Type your search results and press enter and abracadabra, you’re searching Wikipedia. This is very useful as I usually want to search a specific site for my information.

What I also like to do is skip my search results page and follow the I’m Feeling Lucky first result. A lot of times I’ve already searched for it, or am pretty sure the first hit will get me what I want. To do that setup a new search engine with whatever keyword you want. I picked “L”, but it could easily be “lucky” or whatever.

Here is the full text of the URL.

http://www.google.com/search?hl=en&q=%s&btnI=I%27m+Feeling+Lucky&aq=f&oq=

Now just press “L”, plus maybe a space, then tab.

True, the page you’re going to end up on is the link below , but it takes less time to just type it and press enter than to see if that is what you want, press down and then enter. Also that link isn’t always the same location.

Anyway, I hope you’re feeling lucky punk…Are ya…feeling lucky. (you know that was coming)

I’m really excited about using Cucumber. There has been a desire to get to that level of BDD-ish style tests ever since I first saw how readable they were. We’re ready to make that next step. Let’s pretend we have the following set of pages.

Pretty straight forward. We should be able to write a pretty simple Cucumber feature.

Feature: Add Foos
  In order to keep track of foos
  As a person who enters foos
  I want to be able to add a foo 

  Scenario: Adding a foo without searching
	Given I view the welcome page
	When I click the Add New Foo button
	Then I'm taken to the Add Foo page

  Scenario: Adding a Foo
    Given I've searched for a foo that doesn't exist
    When I click the "save" ...that should say "Add"
    Then I'm taken to the Add Foo page

Simple enough. I’m not going to walk you through how to write the steps for that. I’m more interested in how you organize the pages into page modules inside a static language. In Ruby you could do something like

class WelcomePage
  def intialize(browser)
    @browser = browser
    self.goto
  end

  def goto
  end

  def addFoo
    @browser.click '#add'
    AddFooPage.new(@browser)
  end

  def searchForFoo(query)
    @browser.type(query).into('#t') # made up syntax
    @browser.click '#b'
    SearchResultsPage.new(@browser)
  end

end

class SearchResultsPaage
  def initialize(browser)
    @browser = browser
  end
  def addFoo
    @browser.click '#add'
  end
end

That isn’t WatiR. It’s totally made up. I’m a Ruby Newby too so told shoot me for any syntax errors.

You can drive it with something like this.

Given /I view the welcome page/ do
  @page = WelcomePage.new(browser)
end

Given /I search for (.*)/ do |q|
  @page = page.searchFor(q)
end

Given /I click on the add button/ do
  @page = page.addFoo()
end

Pretty straight forward huh? Well how I do that in C#? All of those Navigate methods will either return a specific page type or a base type that will make it impossible to call the subsequent specific steps.

public class WelcomePage{
  public SearchResultsPage SearchFor(query){
    _browser.Type(query).into('#q');
    _browser.Click('#btn');
    return new SearchResultsPage(_browser);
  }
}

public class SearchResultsPage{
  public AddFooPage AddFoo(){
    return new AddFooPage(_browser);
  }
}

Then drive it with this.

var welcome = new WelcomePage(_browser);
var results = welcome.SearchFor('bar');
var add = results.AddFoo();

Doesn’t that feel messy? Luckily, I think a feature of .Net 4.0 might save the day!

dynamic page = new WelcomePage(_browser);
page = page.SearchFor('bar');
page = page.AddFoo();

Three cheers for dynamic!

I really like learning new languages. It’s a challenge and the change in perspective is refreshing. We spend all day trying to solve a problem with the tools our employers pay us to use. Sometimes we have tunnel vision on the means to solve the problem. Some of the things we pick up along the way are language features that give us a new understanding of what is possible.

	5.upto(10) { |i| print i, " " }

Some language features, such as Ruby’s method_missing features that aren’t possible in my benefactors language of choice. It’s not only new language features, but new ways to solve problems. For example, the way Ruby on Rails versions the database and provides the ability to version up and version down the database is impressive.

My newest journey is Google App Engine and Python. GAE supports both Java and Python, but being familiar with Java and it being so similar to C#, Python sounds more fun. Python is a dynamic language, like Ruby, and offers value as a scripting language outside GAE.

So far both Python and GAE have proven very approachable. The app engine is very well documented and the examples are clear and useful. Python is sexy. It’s all that you really need to say. I am not that familiar with it yet, but its string manipulation is very powerful. Take this example.

"something cool"[-4:]

In C# it would look like this.

var s = "something cool";
var c = s.SubString(s.Length - 4)

There’s no denying Python is sexy. You could argue that it is cryptic, but even in this simple example, I believe it’s more readable than the C#.

I have an idea for a pet project in GAE. Hopefully in the next few weeks I’ll have some updates. I’m excited to see how things go.It will not only be an exercise in Python, but also HTML, CSS, usability, UI design. On a day to day basis I rely heavily on our talented designers to make my markup looking great. This is an area of my expertise that needs to improve.

Anyone whom has read any of my recent code knows I love lexical syntax. I really love it when anyone, even our HTML designers, can read our code and understand exactly what is going on. I’m not the only one either.

@MarkBorcherding amen, fellow believer in functionNamesShouldSayWhatTheyDo(AndParameterNamesShouldReflectWhyTheyAreNeededByTheFunction)!

@noelweichbrodt

A great example is the terrible Selenium API. It is confusing to even me, at first…only for a second. Take a look at the following example.

_selenium.Open("some.url");
_selenium.Type("css=id$=_username","homer");
_selenium.Type("css=id$=_password","beer.is.yummy");
_selenium.Click("css=id$=_submit","LoginButton");
_selenium.WaitForPageToLoad("30000")
Assert.Equal("Homer Simpson",_selenium.GetText("css=id$=_userFullName"));

Now that’s not terrible. I would even lean to say anyone would know what that does, but what is the “css=id$=_username”? That looks goofy.  Why do we do that? Take a look at our extensions on top of the Selenium API.

_theBrowser.Types("Homer").IntoAspNetControl("username");
_theBrowser.Types("beer.is.yummy").IntoAspNetControl("password");
_theBrowser.ClicksOnAspNetControl("LoginButton");
_theBrowser.WaitsForThePageToLoad();
"userFullName".AspNetControl().TextIn(_theBrowser).ShouldBe("Homer Simpson");

It wouldn’t be hard to argue the second example is not much more clear. You don’t have to remember that CSS selector or why we did the “$=_” instead of just “=”. Overall, I’m pretty happy with our API. There is a valid argument that this syntax will take time to learn since any developer reading it will have zero experience with it or know what calls are available, but over time, I think the benefit of its readability out weight its learning curve.

So here is where I am today. I don’t like the following code.

<input
    type="text"
    id="<%=Html.IdFor(x=>x.CustomDomainName) %>"
    style='<%= !Model.HasCustomDomainName ? string.Empty : "display:none" %>'
    />

It’s a pretty typical ternary operator. In this case though, I have to explain to our designers what that actually means and how to read it. I’m really tempted to create the extension methods I need to have this syntax.

<input
    type="text"
    id="<%=Html.IdFor(x=>x.CustomDomainName) %>"
    style='<%= Model.HasCustomDomainName.IsFalse().Then(()=>"display:none") %>'
    />

Again, much more clear, but it feels like I’m starting to abuse the extension methods and making a mess of the language…but it reads so much nicer.

Let’s say you were designing a tool to generate code from a diagram. If you were try to generate code for a situation where you expect there to be a primary key on a table, and there isn’t, would you:

a) assume you don’t need the PK and work around it.

b) throw an error dialog to the screen with a nice error message

c) just skip generating the functionality they were expecting and go on like they really never needed it

If you were the LINQ to SQL team, you would have picked “c”. I realize tables need PK’s and that trying to find an row when there is no PK might be a tad difficult, but at the same time a little help would be nice. A simple dialog “No PK found on table Moron” would have been all I needed.

So I’m a n00b in some areas and it is painful finding this out sometimes. I spent most of the morning trying to figure out why my ASP.Net Web Site application wasn’t routing to my MVC controller that is inside of my ASP.Net MVC Area.

Granted, that is a weird setup so there are a lot of things I thought could possibly go wrong. The real reason was embarrassingly obvious, but not embarrassing enough that I wouldn’t post it for the world to see.

What do you see wrong with this route:

context.MapRoute(
    "Default Route",
    "ServerManagement/{controller}/{action}/{id}",
    new {   controller = "farms",
            action = "index"
        });

Give up? You have to give a default value if you have a placeholder in the route. In this case we have “/{id}” in the route, but we never a default value for the id. The following route worked fine.

context.MapRoute(
    "Default Route",
    "ServerManagement/{controller}/{action}/{id}",
     new {   controller = "Farms",
             action = "index",
             id=""
         });

Lesson learned!

It’s nice when fixing a problem is easy.  It’s nice when you scratch your head for a bit and then smack your forehead because someone in front of you laid out exactly what to do, and exactly what needs to be done.  I just had such a moment.

While working MVC into out existing ASP.Net Web Site project we hit, what we thought was a brick wall.

HTTP Error 500.22 - Internal Server Error
An ASP.NET setting has been detected that does not apply in Integrated managed pipeline mode.

Well the Visual Studio engineers deserve a hand for helping me get this going.  After spending a minute looking into the web.config to find out what was wrong, I ran into this little bit of commented out configuration.

<!--
        The system.webServer section is required for running ASP.NET AJAX under Internet
        Information Services 7.0.  It is not necessary for previous version of IIS.

	<system.webServer>

        ...then some more config

That is exactly what was needed to take us into integrated mode in IIS, which is needed for MVC.