Ben Reichelt’s Weblog

Sorry to disappoint people, but the hype that is starting to build around the Origami projec that MS is supposed to be announcing on March 2nd is going to be a letdown.  Know how you can tell?  If it was really something cool, we would have heard about it already, especially if its a new device as the rumors are speculating. Also, Robert Scoble, king of hype, has started with a teaser blog post, a sure sign that it probably won’t be as cool as you’re lead to believe. (Think Segway)

I’d like to be wrong on this one, and I will be pleasantly surprised if I am.

Just had to point out this FANTASTIC article about realistically scheduling software projects by Christoper Hawkins.  The problems that he mentions about scheduling, and about appeasing the client towards the end of the article are right on!

So I finally got around to watching some Ruby on Rails videos this past weekend, I watch the How to Create a Blog Engine in 15 Minutes demo, and the Evolving your db Schema Without a Sweat demo. After watching both of these demos, I was definitely interested in trying out Rails, so I set out to install it on my home machine, which runs XP Pro.

On the Rails website, it mentions that Rails works best on a *nix OS, and in particular, does not work well with IIS.  I didn’t really want to install Apache on my server, so I set out to try and Rails running without Apache.

In order to get Rails working with IIS, there are a couple of pieces you are going to need.

• Ruby (link to download)
• Rails (link to download)
• Ionic Url Rewriter (link to download)
• FastCGI for IIS (link to download)

Now, after you have all of this stuff, there are some steps you need to follow in order to get Rails installed and running properly. The best tutorial I came across in my searching is located here.  I should point out that I have not gotten Rails to run on IIS yet. On either WinXP or Win2003, its got something to do with FastCGI, every time I try to route a page through FastCGI I get the error message “Sorry, unable to connect to FastCGI server”.  Of course, googling for this error message is completely useless.

So now I’m going to bite the bullet and install Apache, which, in the end, isn’t really a bad thing, cause I figure having a working knowledge of Apache is a decent skill to have, so I’ll take the plunge.  Especially if it means I can stop screwing around with all this junk and just starting hacking my first Rails app, which is the whole point of this anyways.

When I come back for round 3 in the fight against Rails on Windows, I’m going to use this tutorial that appears to be detailed enough to help me get Apache and Rails running together on Windows, I will report back after I try that, hopefully running Apache on windows doesn’t bring my system to its knees.

While this rumored GMail service sounds pretty sweet, I wonder how many companies would be willing to route all of their email through Google?  Can you say “privacy concerns.”  I think the most use of this would be hobbyists who run their own website, bloggers who host their own blog domain, or startups. Which isn’t to say the service wouldn’t be successful, because GMail is a great client and if it takes the headache out of managing email, then thats great, but I don’t think the perception of Google’s privacy policies is the greatest right now.

I really don’t foresee myself using any version of IE 7 until its finally released. I usually don’t mind downloading beta, or even alpha, products like Avalon, Indigo, WWF, or Atlas, but with Internet Explorer, its a slightly different story.

Since so many system components rely on the Internet Explorer libraries, when you replace Internet Explorer, you aren’t just getting a buggy, beta browser; you could potentially be crippling an unknown number of other system functionality. I tried googling for a list of stuff that depends on Internet Explorer to give a list, but I couldn’t find what I was looking for, maybe I wasn’t using the right terms, but in any case, upgrading IE to a beta (beta preview, really) just isn’t worth it, especially when I’m not really interested in their new features anyways, and I only use IE to view sites that need it (read: Sharepoint sites).

As soon as the final version is released, I’ll definitely install it, I bet it’ll even turn up on Windows Update at some point, but until then, I’ll stick with Firefox.

I’ve been wondering lately about best practices for installing and updating an asp.net web site (or any web site, for that matter).  Suppose I have a web site that I’ve been working on, and I’m now ready to deploy it to the production server for the first time.  I can create a setup project in VS that will bundle all my files together, install the database, do any custom actions I want, and create an .msi file to install my web app for the first time.

Thats all well and good until the site has been running for awhile and you’ve been fixing bugs, now you want to roll out an update of the web site, which would include, for instance, changes to the assemblies, changes to some of the .aspx files, and some database changes as well.  Whats the best way to roll these changes out?  I would like to make another installer that detects what needs to be installed, and only installs the updated bits and updates the database, but I also want the new installer to be able to install the entire updated site if I was installing it on a clean server.  What I don’t want, is something that uninstalls the current site, and then reinstalls the entire thing.  Does this functionality exist out of the box with .net setup projects?  Or is this something I would have to create on my own? Or am I missing a better way to deploy updates to web sites? So many questions……….

I was reading Eric Sink’s post about his custom blog tool that he, and only he, uses. The reason no one else could use it even if they wanted to, is that its full of bugs and usability quirks that only he knows about, hell, he only runs it by firing up Visual Studio and running the project in debug mode!

That got me to thinking about the MeWare that I’ve got. I’ve got a server at home that I use for my version control, I host Subversion on it, so I can have all my little projects in versioning and use it as a weak web/database server.  The server is a PIII 667 MHz Dell with 128 MB of RAM, so its not exactly a powerhouse. It seems like everytime I remote desktop into the server from work or my other machine at home, the server ends up dropping the wireless connection, so I get booted off when I’m trying to setup a new Subversion repository.  So, to circumvent the problem of remote desktop, I made a small Subversion admin web page.  Basically the page allows me to create a new Subversion repository by specifying the name of the repository. Heres a little screenshot:

I just enter in the name of repository and check the box if I want my tool to automatically create a tree structure in the new repository and I’m all set!

The tool makes some assumptions about the server setup, namely the location of the svn.exe and svnadmin.exe files, default repository usernames and password to create the initial tree setup, and the permissions necessary to create folders, files and execute commands on the server.

While this isn’t as ‘custom’ as Eric’s app, it just reminded me of my most recent custom app that I made for myself. Whats the most custom app you’ve made for your own use? I would think it’d be hard to beat the restrictions that Eric’s has

I just got an email recently from an online tax preparing service. Its the same one that I used last year to do my taxes, and they were emailing me to get me to sign up again this year because they’ve got all my 2004 info on hand to make my tax prep easier this year.

I gotta say though, the web apps that these tax companies use are pretty cool, and the UI is usually really slick and nice. I just wish they would send me a friendly url in the email. Instead of a nice email that wont get line broken or anything like “http://www.example.com/ABCDEFG”, they sent me one with a bunch of querystring parameters and special characters.  This is bad for a couple reasons:

• The line breaks in some email clients make the link unclickable, because the link spans 2 lines, so half the text is a link, and the other half is lost as regular text.
• It looks crappy. This isn’t terribly important, but I like url’s to be clean and tidy.  Its not too hard to make them that way, and given how much they’ve spent making the tax prep web app, you’d think they could invest in a little url rewriting.

Whats even worse, is that they don’t even tell you how to get to the url in the case that the link goes across two lines.  Non-web savvy users aren’t always going to realize that the link is half broken, so they will just think they got a dead link, not too useful, and bad for business.  At the very least they could have a little message above the link that says “If this link does not work for you, try copying and pasting the link into your browser window…..”.  Granted, non-web savvy users probably don’t file their taxes online, but the problem of messy email url’s goes beyond tax preparation emails, this really applies to anyone sending an email.

UPDATE: Took the apostrophe out of the title, the incorrectness was bugging me (son of a teacher, what can I say)

Recently I was developing a little one-off application for use on our home computer that I share with Erin. I was working on a library that would enable you to easily add items to the shell context menu in windows, so you could put your own custom context menu options onto specific file types.

In order to test the library out, I had to register the assembly using regasm.exe, then copy the assemblies to the GAC so explorer.exe could find them, and then I needed to restart explorer.exe, so that it would refresh the code.  The problem came when I wanted to restart explorer, because we use fast user switching, so it was possible for Erin to have explorer.exe running, as well as myself, and I didn’t want to kill her instance of explorer, I wanted to kill mine. So I needed to know how to get the username of a process, because there isn’t a convenient Process.UserName property on the System.Diagnostics.Process class (don’t ask me why, I don’t get it either).

Anyways, a quick google search brought up some results, and I thought I would share them here. I found a vb solution on experts-exchange, and I ported this to c#, heres the code to get the username of a process, by process ID:

static string GetProcessOwner(int processId)
{

string query = “Select * From Win32_Process Where ProcessID = “ + processId;
ManagementObjectSearcher searcher = new ManagementObjectSearcher(query);
ManagementObjectCollection processList = searcher.Get();

foreach (ManagementObject obj in processList)
{
     string[] argList = new string[] { string.Empty };
     int returnVal = Convert.ToInt32(obj.InvokeMethod(“GetOwner”, argList));
     if (returnVal == 0)
          return argList[0];
}

return “NO OWNER”;

}

Of course, I come to find out later on that its a bad idea to write shell extensions in managed code, which makes sense when you think about it.  Oh well, it was interesting work while it lasted

Ah, finally, Sql Server has an easy way to page through the result set without getting the entire set. In MySql there is a LIMIT keyword that you can append to your select statements to select only a subset of the query, for instance:

SELECT * FROM Customer LIMIT 0, 50

This query will return 50 rows, starting at row 0.

SELECT * FROM Customer LIMIT 1000, 50

This query will return 50 rows, starting at row 1000.

Getting data like this is nice because with large result sets, you can select just the rows you want to display. Sql Server didn’t have this functionality until now with Sql Server 2005.

In Sql Server 2005, there is a Row_Number() function. You can specify this Row_Number function in your column list of your select statements, like this:

SELECT *, Row_Number() over (ORDER BY FirstName) AS RowNumber 
FROM Customer
ORDER BY FirstName

This will add a column called ‘RowNumber’ to the result set, and all RowNumber will be is a sequential number that numbers the rows, and it will number them using the FirstName column.  So even if your query doesn’t order by FirstName, you can still number the rows by FirstName, make sense?

So now, what you can do, is add to your query to get a subset like this:

SELECT *, Row_Number() over (ORDER BY FirstName) AS RowNumber 
FROM Customer
WHERE RowNumber > 1000 AND RowNumber < 1051
ORDER BY FirstName

This query is going to retrieve the rows numbered between 1000 – 1051, allowing us to page through the data just like we could in MySql.  This is awesome, I’m so glad they put this into Sql Server, its something that I’ve been wanting ever since I’ve used MySql.