Hacker News Reader

Comment by shanselman | Hacker News Reader

It's worth noting that under Satya (in my org, Cloud and Enterprise) we open sourced ASP.NET, use 50+ OSS libraries in Visual Studio, have all the Azure cloud SDKs on GitHub, and on and on. We made Portable Libraries happen and now share code between iOS, Android, and Windows. This is not your grandfather's MSFT, and now the dude who helped us (Azure) change things in a fundamentally non-MSFT and totally awesome way is in charge. I'm stoked - big things coming, I think.

Disclaimer: I work in Azure.

View on HN

I suggest open sourcing Windows. Put it on GitHub with a permissive license and accept pull requests. The internet will blow up, you'll instantly win back all of the developer mindshare that has been lost over the years.

That would be the most awesome thing that could ever happen, when they can just keep offering the windows 8 for the customers in boxed but allowing developers to contribute to the project.. I cannot think of percentage of people who will be lost customers, it will not matter when everyone in another 10 years will have their own linux systems on cloud..

Would it not be a major security problem .. Since so many people use windows and hackers would know loop more easily and attack

To roughly the same extent as Linux and *BSD, yes.

Linux and BSD have the security advantage of having their ugliness shown to the world for decades. When you push a change that everyone can see, you don't have the option of hiding your ugliness behind a compiler.

Is next decade finally going to be the decade of Linux?

The next decade will be the decade of Linux, and always will be ;-)

I think his next move will be to only accept payment in bitcoin.

And an API!

You obviously have no concept of what the windows source looks like. Putting it on github would be a massive technical challenge alone.

Can you explain?

Let's just say that comparing Windows to the linux kernel would be a mistake. Hosting the windows source would be a challenge for any source control system out there. Git / github is not exactly optimized for projects of that size.

Yea, look up "Source Depot"

No. Don't. Please

Sure, Windows is more than a kernel. I imagine the NT kernel vs the Linux kernel would be roughly comparable. But to compare 'Linux' the OS to Windows would mean Windows vs. GNU/Linux/X.org/Gnome/KDE... or say, the source for every package in the base Debian distribution.

Exactly. Now imagine that every part of it was just as bloated and convoluted as you'd imagine.

I doubt that Windows is monolithic/bloated. Convoluted sure, backwards compatibility is a bitch, but Windows is probably broken into neat little modules by now.

But wouldn't it be nice if Windows could be modularized in the process of such an open-sourcing effort, such that that comparison wouldn't be a mistake? For example, if just the NT kernel could be separated into its own project, with a clean linkage-point, there'd probably be all sorts of interesting efforts to, for example, create a Debian/NT distribution, or cross-pollinate driver code between Windows and Linux. (Actually, having two major open-source consumer-PC OSes would probably finally force a standardization on a HAL to allow a single driver-pool code base to be shared between the two systems.)

Apparently you don't know what you're talking about. Source Depot at Microsoft (rebranded Perforce) can do it, while TFS cannot, therefore half of Microsoft stores its source code in TFS and half in SD. Git scales really well, at least well enough for the Linux kernel which is not much less than Windows kernel.

What Perforce is good at and what Git is good at don't overlap too much.

I was under the impression what he posted was a joke.

Isn't Microsoft in the business of solving technical challenges?

I think you'd be surprised.

If Microsoft released the source code to Windows 98 OSR2, which is effectively abandonware as far as I understand, it would be great. A lot of multimedia applications and old games would instantly become easier to preserve.

Is there still any value for Microsoft in Win9x?

Why not XP? The have no interest in supporting it. And millions of people still want to use it.

Because this would allow these millions of people to continue using this fork of XP rather than eventually buying a MS operating system.

But this is the attitude Microsoft needs to eliminate. The days of making money on operating systems is rapidly fading away. They already gave away Windows 8.1 as a free upgrade, so they're aware of it.

An open source XP would get many hackers interested in Windows, something they desperately need.

Citation needed.

They are definitely making money on closed-source operating systems and operating system tools:

http://www.microsoft.com/investor/EarningsAndFinancials/Earn...

Umm, I don't buy Apple computers for their backlit keyboards...

> They already gave away Windows 8.1 as a free upgrade

It wasn't free--it took a couple of hours of my time!

I for one, would welcome a day where I can put install MS Office and VS on my Linux system without any hacks.

I think Microsoft would do very well to go the IBM route, and a FOSS Windows would really shake up the market.

Anyhow, the new MS CEO seems like a good guy, hopefully it's the beginning of a less evil, nicer Microsoft.

This is an amazing idea. I am sure there is a way to charge for licenses for completely open source software. Will Satya do this? Can he do this?

Microsoft won't be able to charge much for Windows anymore. They are already squeezing the price, and losing market share at the same time.

Maybe open sourcing Windows will keep their other products at the positive for longer.

It's pretty hard to do it with a conventional OSI certified license. Anyone can rebuild your product and resell it or give it away. Red Hat tried something similar but CentOS still existed.

Microsoft could come up with their own license that was "open source" but not "free software" by Stallman's definitions (ignoring here the people who've attempted to make these terms synonymous), which would allow people to change code and distribute freely among license holders, but not allow people to resell or just give out copies of Windows. I'd actually be really excited about this, because I think the FSF is overboard with their demand that everyone essentially make their software unsellable.

They could go the Red Hat route, go GPL, accept that they'll no longer get money for software licenses but instead support contracts (they could even try selling these to consumers; Dell would pay MS $50/PC not for a copy of Windows, but for a subscription to Windows Update). This would shift the dynamics of the company a bit, but probably not as much as one may think, since enterprise keeps MS afloat anyway, and it may be a good solution.

This remind me of Unix source licenses.

a FSF licensed product is not unsellable..RMS sold copies of GNU software for years to pay for his living expense

He still does. GO to one of his meetings in berkeley, he sells emacs, linux CDs. RMS is not against selling software. He just doesn't like restrictions that all. btw, who likes restrictions?

> Microsoft could come up with their own license that was "open source" but not "free software"

The important question then would be: Could you legally port directx to *nix from this? If not, then what'd be the point?

Yes, and you could distribute it, as long as you had a checkbox prior to the download that said, "I assert I am a Windows license holder and entitled to utilize derivatives of its code", or a banner that printed a similar message after installation, instructing you to uninstall if you aren't a license holder.

It's great to hear from somebody who worked in Satya's division at Microsoft. Are people at Microsoft generally upbeat about Satya's CEO appointment?

Totally. He's super nice, wicked smart, very approachable, listens more than he talks. He has the Enterprise and Tech chops.

He promptly promoted Scott Guthrie in his first official move (that I know about), so things are off to a great start in my book!

And I'm totally stoked to learn that Scott Guthrie has been promoted to take Satya's place. Both of you guys are awesome! Now when are you going to open up C#, CLR and buy Xamarin ;)?

Things are pretty open now, but we're going to open things up as much as we can.

As for buying them, why mess up a good thing?

Ah just so they have access to the vast resources and become 1st party libraries. Mono is not known in most of the linux world and Microsoft support would be a significant boost. Businesses would also be hesitant to support it adopt it unless Microsoft is backing the platform.

Microsoft supporting Mono would not be a good boost in the eyes of most Linux people, as far as I can tell. In fact, Miguel de Icaza connections to Microsoft have been cited as an argument against Mono.

The fact that Mono is a reimplementation of something Microsoft did actually went against it for a lot of *nix people. Which is dumb and stupid, but hey, after Embrace Extend Extinguish, everyone got kinda jumpy. So, Microsoft purchasing it would be a hard sell, IMO.

Because we'd love to use C# across more platforms, and the 2nd class citizen status of Mono in the *nix world makes us sad :(

Would love to see Windows Phone open up for hobbyist devs. Love the Windows Phone experience out of the box, but having to jump through hoops to write apps for my own personal use pushed me to android.

OSS is nice, but why do you guys keep pushing the shitty HTML5? It is totally crap for enterprise. Also EF is always trying to catch up with NHibernate, why not simply contribute to the latter?

What's your suggested alternative to html5?

WPF

You forgot the sarcasm tags, I see. Or you don't have much experience with WPF. It might be nice in theory, but I've had to work around enough bugs marked "wontfix" that I recommend everyone stay away.

I have years of experience with WPF & Silverlight and agree that it has high learning curve, but once you know it there really isn't much of anything you can't to simpler, faster, and more elegantly than in HTML5, if you have a back end you need to interact with.

Except ...

Some things only work in Silverlight, some only in WPF, and mixing both at the same time is a royal pain.

There are a LOT of bugs and regressions in WPF4 that are "wontfix", were supposed to be fixed in 4.5 or 5.0, but that's never going to happen. (Google "wpf 3.5 4.0 bug wontfix" to see a very incomplete list of regressions. I've met some of these myself, and some that are not on the first two pages).

And lastly, unlike Silverlight (Mac or Win, no Android, no iOS, no Linux), HTML5 is much more widely available (and with an easier graceful degradation path). The only place where using WPF or Silverlight makes sense today is when your target audience is using a locked down system - e.g. an enterprise with tightly controlled client machines. Otherwise, it's a no-go just from this perspective.

I didn't find the learning curve intimidating, FWIW. But the experience with using and deploying me scarred me.

I agree that HTML5 is much more available. I wouldn't recommend a silver light app for any consumer based product. But I would have for enterprise applications if it was going to continue to be supported.

The lack of support killed off the language, not the language itself, IMO.

What about using it on anything that isn't windows or mac os?

Isn't it kind of the point of web browsers to render webpages in a common open language instead of being a host for proprietary vendor specific plugins? What's the big advantage over standalone applications in that case anyway?

We are talking about enterprise which is 99% Windows.

Some enterprises are, perhaps most. But tablets and phones HAVE made a dent, and some places are BYOD even at the computer level.

e.g. Google is very far from 99% Windows - and though it is not your standard "enterprise", it is far from unique.

Sure, but why make the commitment to build your whole thing windows-only when you are only a tiny bit away from actually doing it platform independent. Like, for having options in the future, and stuff.

It would be adding several more months to project completion, and then recurring maintenance costs that overtime costs much more money.

Turning HMTL 5 into a competent application platform = million of man hours.

Convincing Google, Apple, Mozilla, the rest of the universe to adopt Microsoft's closed platform (even if it were Open Sourced) = billions of billions of man hours :)

We are talking about enterprise. There is a chance to build something nice there instead of cramming badly-designed document-oriented languages into GUI space. Apple is closed AFAIK.

Also: You guys contribute a lot of good work to Hadoop and related projects.

Disclaimer: I contribute to some ASF projects.

scott, what about the operating system engineering group? Do you see it will lean towards open source under Satya's leadership?

also, what do you see MSFT to do to regain talent from Facebook and Google?

Seems like it may be an interesting place to work in the coming years!

We'll see, Gates and Ballmer are going to stay on Microsoft board. I'm wondering how much freedom Nadella is going to have in making decisions.

Reply on news.ycombinator.com