Todd Anglin's Code Campground

Azure’s Free Ride almost over, Get your credit cards ready

Consider this a public service announcement for those of you that may not be receiving (or paying attention to) the details about the upcoming changes to Microsoft’s Azure pricing. Up to now, Microsoft’s Azure cloud hosting has been a “CTP” preview and as such it has also been free. Come PDC and November 17th, though, your continued usage of Azure as a free playground for cloud development will nearing its end.

Here’s what will happen:

1. At PDC, Microsoft will introduce a number of new features to Azure and more-or-less start the transition of Azure from a CTP to a paid service
2. Azure will remain a CTP until the end of the year (December 31st) and you can continue to experiment and use it freely during that time. (I think MSFT will allow new CTP sign-ups during that period, but I’m not 100% certain.)
3. In January, you have to make a commitment to keep using Azure. You’ll be able to sign-up for the service, and as a way to help you make the transition from free to paid, your first month of usage will be free.
4. Finally, on February 1st, your credit card will start paying for your Azure time.

So, at this point, you’ve got about 3 more months to play with Azure for free. After that, if you use a bare minimum Azure setup with SQL Azure hosting, you can expect to pay about $100 per month (as I outlined in a previous blog post).

Microsoft’s January “grace period” is also a nice gesture. While a basic Azure setup will only cost you $100/month, costs can be much higher if you’re using many nodes. The free January “bill” will be a good reality check for how much your testing setup will cost in the real world.

Finally, don’t forget that Microsoft also recently announced that some versions of Visual Studio 2010 will be shipping with “free” Azure hosting. That means you may still have an option for experimenting with Azure without incurring extra charges if you’re using VS 2010 Pro, Premium or Ultimate with MSDN. Specifically:

• Ultimate gets you 250 hours/month of Azure cloud computing
• Premium gets you 100 hours/month
• Pro gets you 50 hours/month

While cool, don’t forget that the average month has 720 hours. So if you upload a project to Azure and expect it to be there day-in and day-out, you’re still stuck paying for about 500 hours of computing time (about $60/month at $.12/hr). It’s also unclear if the MSDN packages with Visual Studio will include any provisions for Azure bandwidth and SQL usage, and if they don’t that’ll be another $5 to $15 per month.

Now that you’ve had a year to play with and test Azure, are you ready to pay for it? Is it worth money or will you be sticking with other tried and true hosting providers?

Are you a digital pack rat?

With the recent acquisition of my new Mac Pro, I knew it was time to face the music and clean-up years of accumulated digital files. I have procrastinated the task for years, and now I fear further delay could lead to tragedy, with old, loud hard drives just daring me to let them guard my data until their platters spin no more. It is time to stare my digital pack rat behavior in the face and do some house cleaning.

But how did I get to this point? How did I become a digital pack rat?

The problem begins when you buy (or build) a new computer. The thrill of a new computer- a computer that runs fast and clean- overrides all reason. You know your old, raggedy documents need to move to the new computer with you, but you can’t be troubled at a time like this to go digging through your old hard drive making “life or death” decisions about old documents. And you definitely don’t want to just copy everything to your nice, new, clean hard drive.

What do you do?

If you’re a digital pack rat, you just add the old drive to your new computer. IDE (for really old drives) and SATA cables are cheap and chainable, so you just slap the old drive in the new system and your problem is solved. Not enough room in the new computer case for more drives? Then your drive finds an external HDD enclosure and leeches on to your new computer, like a suckerfish attached to a shark. You can worry about your old documents later when you have more time.

And, of course, that time never comes.

Years go by and systems get upgraded. Each time you make yourself the same promise: I’ll bring my old files with me now, but as soon as I have more time, I’ll go through my old files and consolidate them on one drive. This builds, until one upgrade you’ve had enough.

That’s where I found myself this weekend. Staring at five dusty hard drives, a mix of IDE and SATA, accumulated over years of upgrades. With my shiny- and most importantly, quiet- new Mac Pro begging me to not ruin the silence with my bevy of bytes, I set-off to finally make good on my promise and consolidate my digital life.

It was actually more fun than I thought. You’d be amazed by how much old files can tell you about your life. I found everything from old college class work to saved “IM fights” with my wife from our dating years. I found random sound clips, funny short video clips (reminding of the days before YouTube), and pictures of forgotten events.

There were three important pieces of technology that helped me unpack my digital pack rat nest:

1. External HDD to USB Dock – This little dock allows you to treat a full-size HDD like a USB stick. With the convenience of plug-and-play, and the transfer speed of eSATA, I was able to make quick work of my GB of old data.
2. 1TB Internal HDD – To consolidate your old data, you need a big drive to hold everything. My old drives ranged from 60GB to 320GB, so a 1TB drive was plenty for my needs. I opted for the “green, low power” Western Digital primarily because A) this drive is for storage, not active work, and B) noise is very important to me, and this drive is whisper quite!
3. Mac OS X – For quickly digging through old files you don’t recognize, you can’t beat OS X’s “Quick Look” functionality. You can literally hit the spacebar for almost any selected file- Office document, Adobe document (including Photoshop and Illustrator), videos- and OS X will instantly preview the contents for you without opening any new programs. I love Windows 7, but it has nothing on this OS X feature.

I now have an organized collection of digital files on a single drive, and a stack of old drives on my desk to format and reuse for other projects. Once a digital pack rat, I’m now organized and clean…at least until the next computer upgrade!

What you about you? Are you a digital pack rat? How many years of hard drives or “archived” files are you carrying around with every computer upgrade? Share you pack rat stories in the comments.

Unexpected changes from Silverlight 3 Beta to RTW, The Good and The Confusing

By now, everyone should know that Silverlight 3 is official. Born on July 10th at a big launch event in San Francisco, Silverlight 3 completes the 9 month cycle of improvement that kicked-off immediately after the official release of Silverlight 2 in October of 2008.

If you are at all interested in Silverlight, you probably participated in the Silverlight 3 beta. It was a great beta program that gave developers a solid look at many of SL3’s new features, but as with all betas, there were changes in the final Release To Web (RTW). Actually, there were a lot of changes, many of them “breaking,” though I’ll be quick to acknowledge that you’re “allowed” to make breaking changes during a beta cycle. You can find a complete listing of changes and how you should address them on MSDN:

http://msdn.microsoft.com/en-us/library/cc645049(VS.95).aspx

So far, so good.

In fact, some of the changes in the RTW are good breaking changes.

• Out-of-Browser API Breaking Changes
  During the beta, this API had some less than intuitive names that lead to some awkward programming. For instance, during the beta, the idea of taking a Silverlight application OOB was considered “detaching” it from the browser. When running in a “detached” state, the Silverlight app was said to be “RunningOffline.” Confusing, since an OOB Silverlight app can still be connected to the Internet and running “online.”
  
  In the RTW, the API names are much better. You now “Install” a Silverlight app to take it out of browser (instead of “Detach”), and you can detect OOB apps by checking the “RunningOutOfBrowser” property (instead of “RunningOffline”). Much better…
• Assembly Caching Changes
  During the beta, there were two problems with Assembly Caching (a.k.a Transparent Platform Extensions): it only worked with Microsoft assemblies and it required that the assemblies come from Microsoft.com. With the RTW, both problems are fixed. Now any assembly (including 3rd party assemblies, like Telerik’s RadControls) can be cached and by default the assemblies will come from your own domain (though apparently you can still use the Microsoft.com downloads if you want them).
• URI Mapper Changes
  Less breaking change, more RTW improvement, the URI Mapper (used in Silverlight Navigation to define URL routes) is now part of the default “Silverlight Navigation Application” Visual Studio template. In RTW, the Navigation routes include a “pattern matching” convention by default (instead of hard coding all links), which I think is going to make it a lot easier for people to discover the path to “clean” Silverlight deep-linking URLs.

Not all breaking changes in the RTW are good, though. And I’m not complaining about changes that are just frustrating because they break code. I’m talking changes that actually made me pause in confusion and wonder if Microsoft made the right choice in their design decision.

• The ASP.NET Silverlight Controls Are Dead
  This is the most shocking and confusing change to me. In the Silverlight 3 RTW, Microsoft has officially “killed” the Silverlight ASP.NET server controls (<asp:silverlight… and <asp:mediaplayer…). These weren’t complicated controls. They were simple wrappers that automated the somewhat tedious process of adding all of the HTML object embed tag code required to place a Silverlight application on a page. I, for one, found them convenient.
  
  To make matters more confusing, Microsoft isn’t making very clear why they killed them- only that the new guidance is to write the Silverlight embed HTML yourself (or use the new VS templates). You can continue to use the “old” version of these server controls, and you can even download the source and enhance them. But I’m still left wondering why these controls were abandoned.
• More Controls Removed from the Silverlight SDK
  I get that Microsoft wants (needs?) to keep the Silverlight plug-in size small. I was happy to hear at MIX 09 that after all the new features in Silverlight 3 that the plug-in size got smaller. But with the Silverlight 3 RTW, Microsoft has moved 10 controls back out of the SDK and in to the Toolkit. And what does that mean…
  
  It means your application XAPs just got bigger if you use any of these controls. Frankly, I’d rather have a 1% bigger core install and save all apps from delivering these controls.
  
  And should we really have to use the Toolkit for basic layout controls like WrapPanel and DockPanel? I know the Toolkit helps controls get updated faster, but these seem like staples that belong in the core- where they started in Silverlight 3 beta!

Silverlight 3 is a huge improvement over Silverlight 2. I think it is the version that is going to really make more people attempt serious application development on the RIA platform. The feedback from the Beta program obviously had a real and noticeable impact on the RTW- I’m just not sure all of the impacts are good.

What do you think?

Azure Cloud Pricing Falls Somewhere Between VPS and Dedicated

As you may have heard by now, Microsoft has just unveiled the official pricing for its forthcoming Azure cloud computing platform. Ever since its introduction at PDC08, Microsoft has been very coy about the pricing for Azure. They’ve always defaulted to the party line that Azure pricing will be “competitive” with other cloud offerings.

Today we learn exactly what “competitive” means.

According to the Azure Team Blog, pricing will be about $.12/compute hour, $.15/GB/month for storage, and $.10in/$.15out/GB for bandwidth. There will be additional flat monthly charges for using Microsoft’s “SQL Azure” (or SQL Server in the cloud). I won’t repeat all of the pricing details- they are easy to find on the web now- but I will provide some analysis of how Azure seems to compare to other “traditional” hosting options.

In my experience, there are 3 basic hosting types:

1. Shared – You get a single website on server hosting many other people’s web sites. (Cheapest option, least control – pricing usually $5 - $15/month)
2. Virtual Private Server (VPS) – You get a single Windows VM on a server hosting other people’s VMs. (More expensive than shared, less expensive than dedicated, complete control over OS configuration – pricing usually $99/month and up depending on server config)
3. Dedicated – You get a single dedicated server that only hosts your web sites (Most expensive and most configurable option – pricing usually $200/month and up depending on power of server, quality of service)

With the introduction of Azure, there is now a 4th hosting type for .NET developers to consider:

• Cloud – You get a single website on a “cloud fabric” that can be configured to run on 1 or more virtual “instances” (Most scalable and (perhaps) reliable hosting, little control over environment)

How does the cloud pricing compare to other hosting types? Let’s do the math for a “typical” website running a single Azure instance:

Hosting:  about $90/month ($.12/hr * 24 hours * 30 days)
Bandwidth: about $3/month (assuming 10 – 15 GB bandwidth)
Storage: < $1/month (assuming up to 6GB of storage)
SQL: $10/month (assuming 1GB “SQL Azure Web Edition”)

So that brings that the grand total for Azure hosting to roughly $100/month.

Of course, with Azure, there is no “flat rate.” Everything is based on actual consumption (except for SQL Azure), so if you use less bandwidth or less storage, your bill will be…relatively unchanged. That’s because the real bulk of your bill is tied-up in the “computing hour” charges, which will be at least $90 per month for a single Azure instance.

Where Azure’s “consumption pricing” starts to be an advantage is if you have “peak” processing times when you need additional instances of your application running for a short period of time. With Azure, you’ll only be billed for the additional instances for the time they are running, unlike traditional hosting that usually charges you for (at the very least) a full month’s service for additional hosting instances.

Azure’s Threat to Shared Hosting

Soon after the announcement of Azure at PDC08, the halls were buzzing with people wondering how this “hosting in the cloud” would affect “traditional” .NET hosting providers. Would Microsoft put great hosting companies like DiscountASP.NET out of business? How could shared hosting compete with all that cloud computing has to offer?

Now that we have pricing, I think it’s clear that shared hosting providers are safe. In fact, all existing hosting providers are probably safe from losing much business to the cloud as Azure seems to fit somewhere in-between Shared and VPS services:

• Azure provides little control over the hosting environment, like Shared Hosting
• Azure runs your “instance” in a dedicated virtual environment, like VPS Hosting
• Azure is priced competitively with VPS, but it is significantly more expensive than Shared Hosting

What do you think of Azure’s pricing model? Is $100/month the price you would expect to pay to have your site run in a cloud environment? Or for that price, would you rather have the control that a VPS offers? Sound-off and then prepare for the official Azure launch this November at PDC 2009.

Another Solution for Defeating the SQL Server 2008 Installer with Visual Studio 2008

Here’s the problem:

I have a clean Win7 RC box with VS 2008 SP1 installed and a working SQL 2008 Express “Data Base Engine” (installed via the Web Platform Installer). I discovered that SQL Management Studio Express was missing, and there began my long and frustrating journey to install Microsoft’s SQL management tools…

I’m not sure why the tools were missing in the first place, but it turns-out that there is no simple way to install them. There is no separate installer for the tools as there was (so conveniently) for SQL Server 2005. [UPDATE: After fighting a long battle, I rediscovered an option in the new Web Platform Installer to add the SQL Management Tools to SQL Server 2008. A little late now to know if that would have worked, but hopefully you can give that a try before getting too far on your own painful SQL Server 2008 journey.]

Enter the SQL Server 2008 installer. After countless attempts to “add” the tools to my existing install, I could never get the installer to provide me with the desired options. Not one to waste time “fighting software,” I uninstalled SQL Server 2008 and started fresh. To my dismay, the SQL Server 2008 Installer started throwing this install-blocking error at me when trying to re-install:

Setup will not complete because the rule "Previous releases of Microsoft Visual Studio 2008" failed.

This message further asserts that you must install Visual Studio 2008 SP1 before the installer will continue. But…like many others, I already installed SP1!

So, I began looking for hacks around the installer. There are many options scattered across the web, but the tip that finally made things clear is this:

The SQL Server 2008 installer is checking the HKLM\SOFTWARE\Microsoft\DevDiv\XXX\Servicing\9.0\SP reg keys to tell if SP1 is installed. If you inspect those keys and find any SP keys with the value “0,” that is your road block.

As it turns out, every installed version of Visual Studio 2008 must have SP1 installed for the SQL Server 2008 installer to work. This leads to the other main source of confusion:

If you installed SP1 for your “main” VS 2008 version (such as Pro or VSTS)- and you’re sure of it- there is probably some other version of VS installed that does not have SP1 installed (such as Express or- as in my case- Shell).

Inspecting your reg keys is the fastest way to discover if such a version is installed. In my case, I discovered that there was an entry for both “VSTS” and “IDE” in my Visual Studio reg keys. VSTS was my main Visual Studio version, and its SP reg key correctly indicated that SP1 was installed. For the IDE entry, though, the version was listed as RTM with no SP.

What is the “IDE” version of Visual Studio?

Thanks to the info on Heath Stewart’s blog, we learn that this reg key entry is for Visual Studio Shell edition. It’s not clear when this version of VS2008 is being installed (or how important it is for other programs), but uninstalling it or upgrading it to SP1 is yet another way to solve your SQL Server 2008 install problems.

One more handy tip.

If you really want to take a blunt-force approach to any of the SQL Server 2008 “warnings,” there are command line switches that enable you to bypass some of the rule checks. Originally revealed by “Paul” in the comments of another blog post, you can use the following switch to launch your SQL Server 2008 installer with the VS version check and “reboot required” rules disabled:

setup /ACTION=install /SkipRules=VSShellInstalledRule RebootRequiredCheck

Based the sheer volume of comments and blog posts already on the Interwebs about this very topic, it would seem that Microsoft may have regressed a bit with the release of SQL Server 2008. I’m sure there are plenty of “good reasons” for why SQL Server 2008 “has” to be so dependent on VS 2008, but the end result is clearly lots of frustration for users. I hope MSFT takes note and makes things simpler for the next version. I know many hours of frustration will be saved if they do!

Windows 7 Beta – Application Compatibility List (update)

With Windows 7 officially in beta and millions of people apparently clamoring to try it, I expect many people will be playing the “does it/doesn’t it” work game with their favorite software and the new beta OS. I recently made the jump to start running Win7 full-time and have already installed a number of applications with varying degrees of success. To help others traveling along this path, I plan to log my experiences here for my app compatibility experiences with Win7 Beta.

Before I begin listing apps, though, let me provide my general advice when it comes to installing software on Windows 7 beta. It is very possible that you will hit problems when trying to install your software, but in almost all cases, you can solve your problems with these steps:

1. Re-run the installer as “admin”
   With elevated privileges, most installs recover and work- even installs that should have worked b/c they worked in Vista.
   
2. Re-boot, re-run the installer
   When the simple “Run as admin” fails to solve the problem, try rebooting, then re-running the install. In cases where run as admin failed, rebooting solved most of my remaining install problems.

With that out of the way, here are the specific apps I’ve installed and the current status (last updated: 20 Feb 2009):

UPDATE: I just got tired of needing to elevate everything to “Run as Admin” to get anything done, so I disabled UAC completely. After a restart, low and behold, Live Mesh started working! And, the files that previously wouldn’t work with Windows DVD burner suddenly were available for burning! Guess UAC needs some more “tweaking.” For now, if you’re a power user, you can probably save some headaches by turning it off. </update>

That’s all for now. As you can see, the only things really giving me trouble are Live Mesh, Virtual Clone Drive, and Window’s own built-in DVD authoring. Otherwise, things are working well. If I have any other major installs, I’ll be sure to update this post, so bookmark this for future reference. Also, feel free to share your own experiences in the comments to help others. If there are enough comments, I’ll update the post with the community compatibility updates.

Hope this helps all of you joining me on the bleeding edge!

Chrome still needs some polish

As pretty much the entire developer world knows by now, Google recently made a splash by introducing their own "home grown" browser called Google Chrome. The browser was introduced as "the browser that does everything right because it doesn't have legacy baggage," and it has largely lived up to its billing. The browser is fast. It's easy on the memory footprint. And it's definitely got Google's minimalist signature all over it.

That said, as I sit here typing this, I'm waiting for Google Chrome to respond. It seems one of my tabs has introduced some nasty JavaScript and completely hosed the entire, well, chrome. That's right. The sin which Chrome was supposed slay (one tab crashing the whole browser) still lives. And that's not the only sin that lives.

In the weeks since Chrome's introduction, I've been using it as my new primary browser (move over Firefox 3 and your continued memory bloat). Over that period, I have come to appreciate the good and the bad with Google's freshman browser effort, and I can definitely report that it has its issues. Not enough to make me stop using Chrome as my primary browsing browser, but enough take some of the shine off Google's glistening browser.

Among the observed problems, here are my top "complaints" (or "ways it could be improved" for you optimists out there):

1. DNS Resolver is Weak: Don't know why or what this means, but on multiple occasions, sites that wouldn't open in Chrome could be successfully loaded in Firefox. It seems like Chrome has trouble establishing a connection with some sites and just quits trying (providing "UNKNOWN ERROR" as the most helpful of error explanations). This is a pretty big deal since it could mean users hitting this error think your site is inaccessible when, in reality, it may just be inaccessible to Chrome. The problem is magnified by the fact that the "error page" looks just like a normal Chrome 404 page, so it isn't entirely clear Chrome has encountered a problem. Hopefully this is on the top of Chrome's "must be improved" list.
2. Rendering Not Perfect (or at least not same): This isn't a big deal to users per se, but as a web developer (and a web component vendor) it is rather annoying to have yet another slightly different rendering engine to support. True, Chrome is most like Safari, but it's not identical. That means another set of CSS hacks, JavaScript tricks, and browser testing scripts to manage. And yes, hacks and tricks have already been identified that are required to support Chrome. Thanks Google...
3. Weak Developer Tools: Again, more of a developer complaint, but Chrome still lacks the powerhouse developer tools Firefox provides (albeit via its add-on market). Chrome does pack some interesting developer oriented tools, but they're not nearly powerful enough to replace Firebug or the Web Developer toolbar.
4. Inconsistent Plug-in Experience: While this burden really falls on the plug-in vendors, it is annoying that plug-ins like Silverlight and Java don't consistently work in Chrome. Eventually I'm sure these plug-ins will catch-up, but for now, it's hit or miss if the plug-ins you need to browse the web work in Chrome.

What does this all mean? It means Chrome is a "really good" browsing browser, but it's not good enough to be your only browser (yet). You still need Firefox or IE around to load specific sites or perform serious analysis on web pages. A pretty substantial accomplishment for a "brand new" (its roots are quite established, really) browser, so Google should be proud. If I were Mozilla, I'd probably start to reconsider those lunches at Mountain View - Google is definitely poised to kill Firefox popularity in a few years time.

I still plan to use Chrome as my primary browsing browser going forward. It loads faster than any browser out there (from a cold start), and that alone keeps me coming back. I'm also a fan of the new "intelligent" start page and the super-effective "everything bar." In fact, I've grown so used to using the address bar for both search and navigation in Chrome that I cringe when I'm forced to use a separate UI for those tasks in IE, Firefox, and Safari. With a little more time and a little more polish, I'm sure Chrome will really shine and quickly become my primary browser for browsing and developing.

What do you think of Chrome? Have you become a Chrome convert or do think Google's missing the mark?

XAML : Developers :: Modern CG : Spielberg & Lucas

For anyone having trouble understanding this post's title, it's old analogy syntax: XAML is to Developers as Modern CG is to Steven Spielberg and George Lucas. Still not making sense? Let me provide some more explanation.

How great directors lost their way with CG

Few people would disagree that the great movie directors of yore- namely Steven Spielberg and George Lucas- have passed their great movie making prime. If we look at the movies these directors have produced in the last 5 years (or so), like AI, War of the Worlds, or the new Star Wars, they pale in greatness to early hits. Movies like Jaws, Close Encounters of the Third Kind, ET, Indiana Jones, Star Wars, and American Graffiti are universal classics that make recent flicks look like uninspired, freshman projects.

So what happened? Did these creative minds become less creative with success? Did they lose their ability to make a great movie? I would argue that modern CG happened. Technology opened these director's brains for the world to, enabling them to put every "creative" idea they could conceive on screen. And that, as it turns out, is not a good thing.

Think about it. The practically fully animated Star Wars prequels with their armies of robots and everybody's favorite Jar Jar Binks. The latest Indiana Jones movies and it's cliche aliens and UFOs- and Indiana Jones flying through the air in a CG refrigerator. Even War of the Worlds, for all of its production value, didn't come close to capturing the magic of Spielberg's more successful alien movies like ET and Close Encounters. In each case, these movies were loaded with CG, "enabling" the directors to tells stories and put things on screen they couldn't have done due to technical limits in the 70's, 80's, and even early 90's.

And in each case, these movies have failed to live up to the greatness of their directors. Without technical constraints, these great creative minds aren't making great movies. It seems that creative minds like Spielberg and Lucas produce better results when thy have technical limits. Their movies actually benefitted in the past from some built-in editing. Technical constraints forced them to be more creative, to keep things simple. The result: better stories and better movies.

What's this got to do with XAML and Developers?

Just as great film makers made better movies when they had to be creative within limiting technical boundaries, I see the emergence of new "freeing" technologies like Silverlight and WPF (both based on XAML) creating a new challenge for developers. Now that developers can easily integrate animation, media, 3D, and a whole array of "wiz-bang" features in their applications, I fear developers will make the same missteps as the movies. Previously impossible features can now be easily included in apps, and in many cases developers and designers will probably see this as a good thing. "Finally we can put our ideas in to our apps," they'll say.

But will this lead to better applications for users? I'd argue that in many cases it won't. As application developers, I think we've benefited from a little technical constraint to keep our application's in-check and to force us to be more creative, more resourceful. With flexible technologies like XAML, great developers will be those that can be good "editors" of their own applications.

Are you saying new technology is bad?

Of course, not! I'm all in favor of new technology and moving the quality of movies and software forward. It just needs to be done appropriately to enhance traditional experiences, not completely re-invent them. We don't need Indiana Jones flying through the air in a refrigerator, but putting him on a rocket powered sled (again CG) is great! It fits the story and the CG makes the experience more believable. So it goes with software. Do we need to replace our ASP.NET AJAX apps with Silverlight? No, but we can definitely enhance them with technologies like Silverlight to make the experience more enjoyable where appropriate.

I suppose this all boils down to the old-ish adage: Just because you can, doesn't mean you should. As you start developing with new platforms like WPF and Silverlight, don't fall prey to "unlimited freedom." Be good editors of your applications. Create constraints. Be creative. If you do, I think you'll avoid going the way Spielberg and Lucas and you'll continue to create great software hits on these new platforms!

Potluck Programming (or UWWFY)

Have you ever found yourself spending more time trying to learn patterns than actually designing or coding your application? It's easy to do. With the maturation of .NET, suddenly everywhere you look you see presentations, books, and blog posts talking about this pattern or that pattern, each asserting that the other is inferior for some obscure reason. And you, being the mature .NET developer that you are, follow these conversations, desperately trying to find the "magic" patterns that will make your next project perfectly maintainable, easily extendable, and consistently scalable from the get-go. I know, because I've been there.

Now, there's nothing wrong with patterns. In fact, they're very helpful in moderation. They allow developers that implement code to benefit from the long winded thinking done by developers that have time to think about the absolute best way to do things. The problem, though, is that there is no "absolute" right or wrong way to do things. There are better  and best ways to do things for specific scenarios (as learned from collective experience), but no one can tell you the "right" way to write your application. And left unchecked, it's easy to wander down the rabbit hole of pattern research and get lost in "Patternland."

So, I'm coining (or re-coining, as the case may be) the term "Potluck Programming." Or if potluck's aren't your thing, UWWFY™- "Use What Works For You."

The concept is simple, but essential. When setting-out to improve your development with patterns, find patterns that help solve problems you have (not problems you think you'll have- YAGTNI), find patterns that make sense for your project (not every project is a huge enterprise, super complex application with the need for IoC containers), learn the patterns, and then start writing some code!

And don't feel guilty if you don't follow the pattern 100%. I like to quote the wonderful piece of 21st century cinema Pirates of Caribbean to explain this. In the words of the digitally-enhanced (or is it degraded?) Captain Barbosa:

"The patterns are more what you'd call guidelines than actual rules. Arrr..." (loosely quoted)

Patterns are supposed to help you write code in a way that avoids the problems other developers have encountered, but they're not supposed to be a paint-by-numbers guide to coding your application. Use what works for you and don't let patterns bog you down. And if you find yourself getting bogged down, step back and write some code. If nothing else, that will help provide more context for the pattern so you can learn it more quickly.

Hopefully this simple advice will help you avoid the productivity drain that pattern research can become. Have you ever found yourself crawling down the pattern rabbit hole? Sound-off in the comments.

What "cloud computing" means to you (it's important)

It is amazing to me how quickly the the noise surrounding cloud computing has grown. Overnight cloud computing has become a mainstream talking point with everyone postulating how enterprising start-ups and lumbering enterprises will use the emerging technology in software projects. InformationWeek is running cover stories on cloud computing, Microsoft is touting their new cloud services, and even Apple is getting-in on the game with MobileMe. It seems cloud computing has replaced software as a service as the most popular water cooler talk.

And if you ask me, cloud computing is much more interesting and much more able to be a "game changer" than SaaS ever was. SaaS always seemed like a fancy way to describe hosted applications. It really wasn't anything new- eBay has been providing SaaS auction services for years- just a change in focus and maturity. Even as a young entrepreneur, I built career fair management software that I sold to customers from my hosted environment- who knew I'd "stumbled" on the SaaS business model.

Cloud computing, on the other hand, is substantive. It is bringing real, game changing technology to the table that is going to enable a new generation of entrepreneurs to capitalize on never before possible access to cost-effective, low-risk scaling. In other words, cloud computing is going to give little ideas the ability to reach big audiences without millions of dollars in financing. That's an exciting development if you're like me and you're always looking for ways to bring your ideas to life without investing your wife's life's (Freudian slip) savings.

The impact of cloud computing on .NET developers should be big, too. I fully expect to see cloud service integration become standard practice in .NET applications and a new crop of tools to emerge that will help developers with the challenges cloud computing presents (network latency, dynamic scaling, etc.). At the component level, I wouldn't be surprised to see new data controls that enable you to easily bind to cloud data or ORM tools that make managing scalable data in the cloud as easy as managing a single instance of SQL Server. Unlike SaaS, cloud computing will have an impact on your daily developer life.

For those of you less interested in the business impact of cloud computing and still struggling to see its purpose, there are clear consumer driven applications of cloud computing sprouting-up at the world's major software producers that should provide some clarity. Take, for instance, the "could drive."  The long awaited drive that lives on the Internet, enabling users to save their files on their computer and access them anywhere, anytime.

Microsoft and Apple are prepping services that bring this dream to life. Microsoft's LiveMesh is a software + services approach that uses a small client install to enable seamless file synchronization between all of your devices (PC, Mac, and Mobile). Similarly, Apple's iDrive, previously part of .Mac and now a member of MobileMe, enables you to save files on your Mac and easily access them on the web or on another PC. Apple provides its software integration at the OS level, but end result is very similar to Microsoft's and both offerings rely on cloud services and storage to deliver the anywhere, anytime data access.

With Microsoft's offering in particular, developer's have a unique solution to one of the biggest problems facing Silverlight: no local file system access. How do you build web-based replacements for client-apps if you can't save files to the local file system? Live Mesh. Using the Live Mesh APIs, you simply save an user's file to the cloud and the cloud in-turn syncs the file to the desktop. It's roundabout, sure, but it's also compelling since this approach should mean the file is always available to your online app, even if the user travels to another computer.

But where's Google? Long before LiveMesh and MobileMe were twinkles in a marketer's eye, Google was rumored to be building "GDrive," a Google-powered cloud storage service that you could use to backup and access your files. Since the early rumors, though, Google has been mum on the GDrive topic, allowing both Apple and Microsoft to beat the perennial cloud proponent on what should be their own turf. It is bizarre that Google appears to be a late mover in the cloud computing space, but I think it's safe to bet Google will eventually make their move.

Until then, the work by companies like Microsoft, Amazon, EMC, and others make it clear cloud computing is here to stay, and in a big way. And while I don't think the term "cloud computing" will last forever, the use of the Internet as an applications platform will.