First off, MS's open source credibility is super damaged every time they:

* have Steve Ballmer talk about how he's going to sue all Linux users for patent infringement (and subsequent Novell partnership)

* Create a "we won't sue you" licensing agreement for the Office Ribbon UI  that explicitly states OpenOffice/Google Apps can't use the technique

* Release propaganda referring to the GPL as "viral"

* Invest in SCO

There's nothing you can do about Microsoft's history, and there's little you can do to prevent similar incidents in the future. But just know, that Microsoft will never regain the non-MS-ecosystem developer's trust so long as they keep doing things like this.

Anyway, it's s a tough problem.

#1

BizSpark helps a bit, but the LAMX stack completely obliterates Microsoft's offering on a price per * ratio. By far the greatest problem.

#2:

The performance issue isn't perception, it's real. It doesn't matter if Windows + IIS + ASP.NET is faster than Linux + Apache + X, because 95% of your performance hit is on the DB. MySQL + Memcached in the hands of experience programmers will outperform SQL Server by orders of magnitude. MySQL has its problems, but get it properly configured for a specific application, apply memcached servers, and you're rocking.

#4:

Too much of Microsoft (and community) are still focused on vertical scaling (you know, like Jeff Atwood talking about their 32gb single DB server for StackOverflow). Even if the tools were there, MS developers don't get sharding or distributed hashtables. I'm not even sure MS gets it. Also, with your higher cost in #1, you have much worse scaling cost.

Honestly, a lot of these problems seem linked to SQL server. Not only is it hugely expensive but everything about it screams enterprise. Its models for scaling typically involve buying more expensive versions of the software, and host OS and hardware. Every new release adds a bunch of features that don't help me at all, while the features that would are never implemented. This will only get worse if the move to CouchDB (or the friendster model) catch on.

What can Microsoft do? Well, Microsoft is building Velocity to compete against memcached (and it won't work because memcached will be either cheaper or faster (or both)), instead of, you know, making memcached work better on windows, with .NET and SQL server. Or making MySQL run decently on Windows (I posted old benchmarks a while ago, and InnoDB on Windows was brutal). Microsoft is adding SMTP support to SQL Server (yay!), while Google built Hibernate Shards.

Rob,

I would really like to see Microsoft putting people to support (ie, developers, documentors) open source projects that are not owned by Microsoft.

It can be as simple as putting MS employees on the projects, paying the committers to work on that, etc.

Karl,

regarding #2 - apples & oranges?

regarding #4 - I built a few NMemcached servers, both with compatible binary protocol and WCF protocols.

Easy, cheap and takes almost no time.

I happened upon your post on your blog and posted there, but I think I'll expand on it a bit here.  One problem I see is that up until this point Microsoft seems to feel that OSS projects using your own technology are the competition, which seems odd to me.  Instead of saying "NHibernate is good", you build EF.  Instead of just adding better hooks for N/Mb/XUnit into Visual Studio, you come out with an integrated testing solution.  Instead of playing up the benefits of StructureMap, Castle Windsor and/or Ninject, you come out with Unity.

The main thing that these have in common is that all the MS products are significantly worse than the OSS versions.

Granted, I'm sure developing these "little" projects barely put a dent in MS's overall development, competing against products already built using your own technology seems like a complete waste of time.

Throw in the fact that a large majority of .Net Developers out there only do what Microsoft tells them (due to their vocal bashing of OSS instead of support), I think what has been done is actually a disservice  to your customers.

Not to mention that spending resources (and more importantly time) to build what already exists means MS always seems to be lacking way behind when it comes to technology.

#1 is still absolutely and utterly true.

Now, let me qualify that statement a bit.  When I got out of college I had no money.  None at all.  Not even a penny. I couldn't afford even a single copy of Windows.

Even if I believed that Windows was better and that in the long run the TCO of Windows would be cheaper, at that moment it would not have mattered since I didn't have *any* money.  I went with what was cheaper at the time, Linux.

Until Microsoft matches that price, FREE at that very instant, they will continue losing people to their competition, and even that may not be enough.  Of course, I could have downloaded software illegally and developed on and for Windows, but I didn't.  I winder which of the two is better for Microsoft?

Fast forward seven years.  I'm now a professional software developer working in a largely Linux shop. I don't believe most of the things that you have listed are true, but it doesn't matter now; I have a job I love working with Linux.  Had I started developing on Windows, the story would probably be reversed.

Microsoft has come a long way.  They now offer programs to let students have software at no cost (but only under a restrictive license), or offer software such as their Xpress versions of Visual Studio products.  But, until Microsoft lets every graduate walk away from school with a free and unrestricted copy of Windows, instantaneous cost will drive some developers to lockin to Linux.

Of course, aside from instantaneous cost, there is no such thing as lockin. It's a choice... and I think I'll go back to my jail now ;).

PS: Windows, .NET and their companions have some great features now.  I'm excited to learn some of them (C# and F# look very nice, I can get a Windows domain host for far cheaper than most Ruby-on-Rails ones, etc.), but unless we get a large Windows-specific contract, it won't happen.  I'll have to do it on my own time or at my next job.

I think it would help tremendously if MS would officially sanction and/or promote existing OSS projects.  Forget money or resources, the simple act of MS publicly embracing the tools would be huge.

I do think the MVP awards should be given out for people who make substantial contributions to OSS to help foster the tools and community by putting the all important MSDN licenses in the hands of the OSS devs.

1. Make visual studio (or is it time for a new IDE) run on linux/mac. Visual studio is the one and only piece of software where I need vmware for. Developers using linux are not your enemy. We create software that walks(runs) on the windows boxes of our customers.

2. Visual studio integration for various opensource tools: vcs (svn, git, mercurial, etc.), test runner for all popular testing frameworks.

3. Stop creating worse variants of frameworks that are already available as open source. Instead of that, contribute to the open source! I don't understand why stuff as Visual Source Safe and MS test are developed by microsoft.

4. Work together with other commercial companies building on dotnet, like Sun did with ibm and oracle in the beginning of java. Java developers have still more choice and better tools then .net developers.

@Ayende

#2 - it might not be fair to compare MySQL + memcached to SQL Server, but I think that's how people see it (its largely how I see it).

#4 - I don't know much about your NMemcached project, but (a) I don't want to build my own (I probably couldn't if I wanted to) and (b) if I'm looking at building the next facebook/digg, I want to use *memcached* not something different than what they use (I'm more willing to experiment with a stack than I am with some highly specialized and highly proven C socket server doing crazy mallocs)

Jeremy's right about the MSDN license. Today Martin Fowler mentioned he lost his MVP and thus is MSDN Subscription. Doh! I can't believe I focused on the production licensing cost (which is still significant) and overlooked the development cost.

I think it's not entirely about Open Source.  I think the whole Open Source argument is a bit of a cargo cult.   There are certain attributes about OSS that developers benefit from and the logic becomes that the problems they don't get with OSS would go away if Microsoft had more OSS projects.  Things like openness, etc. are all nice; but there are existing Microsoft products that we currently get that from; like the .NET framework.  But, changes to the .NET framework are fraught with problems; it's slow to change, it's hard to change, change is avoided to avoid breaking existing code, etc, etc.

If Microsoft suddenly moved several projects over to OSS, that wouldn't solve any of these problems.  The only benefit that that would produce would be the ability to fork a new revision.  The problems with inability and difficulty to change would still be at issue.  The process of adding or changing anything at Microsoft is what most people have issue with.  If Mirosoft was more agile and could respond to change and respond to evolution in the development community in a much faster fashion, whether or not a project within Microsoft was OSS wouldn't be at issue.

If Microsoft did want to get into more OSS, they will have to be able to respond to changes and be more agile with these projects or they will simply be failures. *And* support it afterwards.  Don't just throw it over the wall into the community then never be involved ever again.

I'm with Kaleb on this one and I'll go a bit further.  It can't just be free.  It has to be easily available and people have to KNOW about it.

I've always had easy access to their server products  legally/not-so-legally and I think that made the difference.  I've learned enough LAM* to get by when I needed to, but it has always been much easier to learn on Windows.  The products are integrated-enough/friendly-enough so that I can focus on what I am trying to learn without being bothered by unimportant-at-the-time details.

Cost of entry and staying current is M$ biggest problem. Linux is gaining ground and I figure it's only a matter of time before I make the transition to Ruby, Java, or maybe back to straight C++ and drop M$ once and for all.

Coming out of school I took advantage of some "ok" "educational"  pricing for development tools and such, and also took advantage of cross-over "upgrade pricing later on. But it still wasn't free. I worked for one of the early phase "gold" partners back when M$ was practically giving away NFR licenses where employees could buy one copy of just about anything M$ sold, cheap as chips, but alas that didn't last long.

Even today I can't justify spending the kind of money asked for MSDN subscriptions, certification, or even some of the tools to try and keep current. I keep current with VS2008 at the grace of my employer's  MSDN license. I still use an antiquated Office 2000 installation. (For writing the odd doco, and updating the resume.) Plus I plan to remain on XP until something at least "no worse" comes around. Hell, if Windows 7 blows even half as much as Vista that will probably be the last nudge I need to drop M$ once and for all.

I like .Net. I like it a lot, But it's damned expensive to keep up with. Even know I work for companies that struggle to keep up. They're still maintaining .Net 1.1 applications and aim to "trick" new talent in by porting the 1.1 apps into .Net 3.5, ArrayLists, Datasets and all,  just so they can advertise to attract for .Net 3.5 devs. Sadly, I can't say I blame 'em for trying.

It would be nice if MS put their weight behind existing projects instead of having a severe case of "not invented here" syndrom.  All the other big companies pay people to work on open source projects.  This allows those companies to sell services supporting those projects.

I work in a financial institution. We have a great deal of heterogenous systems and database platforms. From c#, VB, MSSQL to oracle, vsamm files, and DB2.

We have java, websphere, asp.net webservices, wcf, and a host of many different types of technologies all working together.

I've seen consistently that our microsoft technology based teams have more secure code, easier to maintain code, and put together more complexity on higher visibility applications more quickly than other teams.

We do have more outages than anyone else but usually they're caused by one of our java or mainframe dependencies. Rarely is our MSSQL a problem or any windows server have any problems. They're easy to update, patch, maintain and have considerable uptime. Our java apps have memory leaks, have considerable IO problems and security vulnerabilities that are constantly being unearthed.

I can see why business people say that open systems and MySql are always bright looking choices.

Because on paper they look cheaper.

To me sitting in the trenches though there is something to be said for keeping your employees happy and keeping maintenence at extremely low costs. When I think of the ultra geeky smart c++ or mainframe guys, I am simply in awe at how great they are, however... i'm not... and i'm lazy... and i hate debugging... and i hate troubleshooting and dealing with crappy tools.

As a developer, I just like the simplicity and snap together-ness that microsoft stuff gives me. I've had to deal with enough thinking and hard work in school to get through C++, oracle, and linux classes. Now i just wanna chill and write c# and kick my feet up.

hah hah. Ok so i didn't answer the question. Microsoft should do what they always do, keep stealing open source project ideas and doing them half assed but marketing them better. :)

It is not secret that .NET is taking from the best of breed technologies in the Java and Rails spaces and making them their own. This is a good thing, since it brings proven technology into the .NET world. M$ needs to embrace these open source project ideas and concepts earlier in their lifecycle and provide more of their own technology as open source as well. This will mature the .NET technology stack faster and make it more agile.

Rob, I think the direction is good. Supporting OSS projects is the way to go. But you have to find out a way to support OSS without endorsing it so that it doesn't become anti-competitive.

Thank you for submitting this cool story - Trackback from progg.ru

I'm a developer. In my "principal work" I use C# but I'm very happy when I can work in a open source enviroment. I think it's better ... and I think it's not very difficult ... If people can learn some arguments in M$ enviroment is not a problem ... but ... why in open source context is a problem? I don't understand it!

Why not support open source?  What happened NAnt, NDoc, Nunit?  It is great that MS Open Sources their stuff, but they have never embraced something out of their control and in the community's control.  They could have embraced Nhibernate, etc.. but they always choose to build something else.  I understand sometimes that building something new is necessary to support customers, but always coming to the same conclusion is a bit suspect.

A couple people have mentioned that MS should sanction/support/etc OSS projects that are outside of Microsoft.

I think they've already done that in a significant way with jQuery.

@mgroves

jQuery is a great example of the way they should act. People just get upset because of the NIH thing that inspired MSTest, MSBuild, Unity, L2S, EF, WF, etc.

As a small startup using the .NET platform, I agree with many of the points from Scoble's post. The three points that matter most to me:

1. From a web applications perspective, scaling up with the Microsoft platform is prohibitively expensive. It would only be worthwhile if IIS and SQL Server were significantly better than the free and open source competition.

2. From a client applications perspective, Microsoft needs to back Mono or some other technology that allows .NET applications to run on Mac OS and Linux. Most companies go with Java or C if they've want broad reach across these different operating systems.

3. Developers, developers, developers. Practically every CS grad leaves school knowing some Java. How many colleges and universities teach their CS curriculum in .NET? By ignoring the academic community -- and especially the pool of students eager to get practical experience on OSS -- Microsoft ensures that it will always play second fiddle to Java.

As a non-MVP who does full-time .NET consulting, I have to buy all of my own tools and the cheapest I've been able to get MSDN is $600/yr, plus $299/yr for the Action Pack (to get production Vista/Server 2008/Office 2007 licenses). Throw in ReSharper, CodeSmith and a few others and I'm generally budgeting $1500/yr for software licensing. That's not exactly a low barrier of entry and is only for MSDN Pro, not Universal or Premium.

I actually have gotten free copies of, for instance, Visual Studio from attending a conference or MS event, only to find out things like that Visual Studio Standard doesn't support unit testing, etc.

I'm trying to figure out how to budget in a copy of Expression Blend to jump start my WPF learning, but can't justify yet another $500 on software licensing this year.

I think this is an interesting question.  I would like to see Microsoft move towards an open commercial style of development like IBM is doing with its Jazz platform.  

Making the source code available while retaining licensing would greatly improve Microsoft's image.  Third-party users would be able to develop better integration with Microsoft products in the future.

If they were to additionally provide support to the open source community (true, free support) that would be great too.  Why write Unity when Ninject/StructureMap/etc. do a better job? Why write Enterprise logging stuff when log4net does most of what you need and more? Why not simply provide support to these open source projects and possibly improve them rather than spend a lot of time and effort reinventing a worse wheel!

@J Wynia: Ah, that's another good point about Expression Blend. I was just looking to kick-start my WPF introduction as well on the weekend, but found the exercise very painful with just VS2k8.

But more on topic about M$ & OSS, one reason I personally don't plan to start/participate in an OSS project with .Net is simply because M$ has a good reputation for "stealing" (for lack of a better word) ideas from OSS, badging it as their own, and bundling it in with VS. (At the expense of a half-baked WPF editor)  I mean seriously, who believes M$'s take on unit testing is nearly as good as NUnit, or their re-factoring support comes close to R#?  How much have they ripped off when they could have just contributed to bettering those technologies? Not to mention recent efforts to possibly put a stake through the heart of Castle & Autofac by putting their leads on the M$ payroll for EF... Time will tell.

The whole crux of the story is - does really MS wants to change? The whole point behind Linq to Sql or EF as I see it, is this - if more and more people use NHibernate, then there are chances that they'll start using other databases - so come out with frameworks which work only with SQL server, so that even if minuscule number of people use it, MS has created some lock-in. And so many others who've never used NHibernate and / or EF, or never bothered to check which one is better of the two, will start arguing why we shouldn't use EF (because it comes from MS)?

Yes, jQuery support was like a shock to me, but again the % of applications / projects which require JavaScript as compared to some RDBMS data access code, is quite small.  So it seems pretty unthinkable that MS advertises NHibernate (or some other OSS projects like Castle, Spring.Net etc.) as the DEFAULT and PREFERRED way to build applications, and of course, Ado.Net and other low-level APIs are still there (for edge cases and those who prefer to steal from employers pockets).

I really feel the opinionated way of RoR community is lot better than non-opinionated way of building apps in MS or Java world. Ever wondered why .Net development is a lot easier than Java (at least for beginners and in some way for experienced guys too) - ease of development environment setup. Install VS.Net and you are done - one can build desktop, thin-clients, web services etc. any kind of application without having to download anything extra.

The similar analogy applies to development experience - IoC, AOP, ORM should be baked into the platform - why do I have to study and then download OSS projects to use it in my project - MS should give me the default choices and those choices have to be the best as voted / used by the community - which means non MS choices, for so much of the stuff.

So that's it Rob - please let there be more Fluent NHibernate first and then Fluent Hibernate instead of the usual - Jxxx followed by Nxxx. Make the lives of those who write such stuff simpler, so that they don't have to convince those who are in power to decide what methodology / framework to use. Please let there be more companies who by default use IoC, AOP in their day-to-day development work, so that people (like Davy Brion) don't have to write that (at this point) in general Java developers are better than .Net developers and I don't have to blurt out in every interview that MS Asp.Net MVC is based on Castle Monorail which in turn is based on RoR and that MS hired the guy who wrote Castle.

And yes, by all means, improve MSDN documentation by including more examples to show this usage -

IList<T> list = new List<T>(); instead of

List<T> list = new List<T>();

It's interesting to see how much hate some people have for MS. It's also very interesting how all those arguments can be true when I see how .NET thrives and often surpasses JAVA and others even during recession. Is it just marketing? Or is .NET simply better?

Steve Py,

Not to come off as a defender of Microsoft, but  I think you are wearing a pretty big pair of anti-Rose Coloured glasses there.

R# isn't a free product, so even though Jetbrains is one of the most open companies around, they're not OSS.  Companies steal from other companies all the time, if they didn't, we'd live in a world with a lot of monopolies and nothing would ever get done.

Also, I imagine one of the many reasons the leads of Castle and Autofac did work on those OSS projects was to bring visibility to themselves (see Jeremy's post about revitalizing your career), so getting hired by Microsoft is definitely something they wanted or they wouldn't have taken the job.  Don't blame the other man when your girl friend dumps you, it was her decision.

Also, to everyone using the term, can we cut the "M$" stuff?  To me it just comes across as unprofessional, petty and immature.  This world is full of companies that act just like Microsoft, yet we don't get so angry with them as we do Microsoft.

Stop the waste! :)

Here is something very simple to do.  Help the Mono team by making core .Net libraries more portable.  For example, Parallel.Net, CCR should not have to be reimplemented by the Mono folks.  That makes no sense.  Instead, MS should embrace feedback and criticism by letting others see how they are implementing core libraries.  At the end, everyone will benefit.  MS may get free, valuable feedback to make things better.  And the .Net ecosystem will benefit from new work being build on top of the open libraries by the Mono engineers, rather than duplicating the effort just so that we have the freedom of running C# on any platform they like.

Specifically Microsoft could do the following:

1. Build a Visual studio design tool for NHibernate (similar to Linq to Sqls).

2. Do more of the MVC project creation model where you can add things like Xunit tests to your project by default.  Add drop downs for logging, IOC, and mocking tools to the MVC project creation tool.

Specifically Microsoft could do the following:

1. Build a Visual studio design tool for NHibernate (similar to Linq to Sqls).

2. Do more of the MVC project creation model where you can add things like Xunit tests to your project by default.  Add drop downs for logging, IOC, and mocking tools to the MVC project creation tool.

An interesting lesson I've learnt over the last year is when you find yourself violently disagreeing with someone, or worse still, going along with an angry mob who violently disagree with someone, instead of ignoring that person, or trying to find fault

1. Give away all developer tools for free.

   Developers developers developers developers...

2. Try to get leading web frameworks (open source frameworks in general) run top notch on windows.

3. Open source or give MS SQL for free. Small to medium businesses don't need all the enterprise features and Microsoft can still make money from these big boys.

4. Integrate better with the *nix world. A *nix flavor of command line instead of old and tired dos-prompt and power shell is welcomed.

5. I don't mind Microsoft comes up with their own framework however, supporting other open source ones to integrate better is essential.

6. Lower your licensing fees.  BizSpark is a good initiative but more is needed.

If you have a huge developer base there's no doubt that it generates revenue.

Definitely not create another mocking framework without listening to the community:

social.msdn.microsoft.com/.../bbbb26c3-f474-4273-b486-f7c35fe1476d

damianm.com/AninterestinglessonI&

Best online pharmacies for buying soma medication. Soma online sales. Buy soma online. Soma online. Soma overnight shipping online.