[fingerprinter]

> Nowadays nobody under thirty writes anything on Microsoft developer tools unless they are demented or brain-dead.

I see most people saying he is wrong here, though I can't think why.

Are there people who thing A. writing software for windows? B. are targeting the windows phone? C. deploying to a cloud (private or public) using Windows servers?

If you are targeting a phone, it is most def either iOS or Android, probably both. If MS, a distant third (and at that point you'll probably be using PhoneGap, Titanium or HTML5).

The number of people actively developing new desktop apps for Windows has to be tiny. Maybe even smaller than tiny.

And if you are deploying to anything other than Ubuntu, you're crazy (and potentially fiscally irresponsible...BizSpark not withstanding).

I get that some people might be using the dev tools, though I would wager (no numbers on this, just gut) that the number of MS Web Devs is far, far fewer than the same open source web devs (PHP, Ruby, Python, NodeJS, Clojure etc).

So, I don't get why people say he is off base.

Frankly, the only people I can see still using MS stuff are the big corporates. IMO, MS is riding the long tail into obscurity. Though, with their financials, it would still be a long, long tail.

[skrebbel]

I agree with much of your reasoning, until

> And if you are deploying to anything other than Ubuntu, you're crazy

What? Did you ever see deploying to Windows servers? There's a reason AppHarbor's documentation pages are so much smaller than Heroku's: Deploying a .NET web app is peanuts. Through Microsoft-only-means, it's done with a single buttonclick from Visual Studio. This works really great. There's a convention as to how web applications are structured that is very widespread and supported by all relevant tooling.

Really, there may be many reasons for not choosing .NET for cloud apps, but server support is not one of them.

Secondly, you're forgetting a major category for .NET developers: devices. Office multifunctional printers, cash registers, machines in factories, any mid to high tech equipment really. Windows has a massive market share here, and lock-in is only one of the reasons why this is going to remain. For example, developing a touch screen interface for an ATM using WPF is very, very easy - definitely among the best options out there. The entire world may be moving to the web, but for devices there is no strong benefit in doing so. Why make a built-in webserver and a windowlwss browser if you can just make a decent native app in half the time, using half the resources? Also, the price of a Windows Embedded license isn't very interesting if the device you're selling costs a few thousand dollars a piece.

Admittedly though, I've no idea how small or large the dev market of machines and equipment is. But it's really pretty sizeable, much bigger than the average HN world-vision warrants.

A lot of software is never seen by consumers, don't forget that. A lot of it is never seen by humans at all.

[spamizbad]

> What? Did you ever see deploying to Windows servers? There's a reason AppHarbor's documentation pages are so much smaller than Heroku's: Deploying a .NET web app is peanuts. Through Microsoft-only-means, it's done with a single buttonclick from Visual Studio. This works really great. There's a convention as to how web applications are structured that is very widespread and supported by all relevant tooling.

Sorry but in a realistic production environment, not just a dev test environment, this is not true. Even Scott Hanselman pointed out ASP.NET has a terrible "deployment story"[1] compared to other options. I'm currently working as an ASP.NET MVC 3 developer and I've been really disappointed in Microsoft's stack in this regard. Life was a lot less stressful back when I was doing Python and PHP deployments on LAMP stacks.

To do it right you'll probably end up rolling our own Powershell scripts to do 1-shot deployments. These work great - but again - it's essentially the same story you have in the Linux world.

[1]: http://www.quora.com/Chandra-Sekhar-2/Posts/Excellent-unbias...

[kenjackson]

The number of people actively developing new desktop apps for Windows has to be tiny. Maybe even smaller than tiny.

Wrong. Windows desktop application development is still huge. Take a look at sites like download.cnet.com and look at how many apps get added each day. And this generally doesn't include the huge ecosystem of things like Office add-ins.

The Window's market isn't a growth market right now, but it's a market where there's still a lot of code being produced for it.

[kitsune_]

The corporate market is much larger than the start up market. Many corporate web sites, intranets and web applications run on either Java or .NET. This isn't exactly a secret.

That's only one part of the cake however. Think of all the companies out there. From engineering to banking to manufacturers of Q-Tips.

[jgeralnik]

(Anecdotal counterpoint) For the vast majority of the students at my university, visual studio is the reach-to solution for any problem. They would much rather figure out how to write ASP.NET because visual studio is "awesome" than go around running node.js (what's that?) from a command line.

[spamizbad]

Anecdotal counterpoint to your counterpoint - I saw the same thing 10 years ago when I was in school (Comp-Sci students gobbling up Visual Studio). It seems the more things change the more they stay the same...

Despite this:

-Qualified .NET developers continue to be difficult to find. Also, they make slightly less money than their Java counter-parts for reasons I don't fully understand.

-The 00's are regarded as Microsoft's "lost decade"

-Microsoft's tools gained little traction among startups... although MS-based ones do exist. The most notable being StackExcahnge however that was created by industry veterans not out-of-college whippersnappers.

Don't count on computer science students to create Microsoft's future. Many will drop the major for something easier.

[skrebbel]

Yeah, but many of those are going to be the lazy and uninfluential engineers anyway. I'm more interested in why people who do look further than the most obvious option would or would not choose .NET.