Microsoft Could One Day Potentially Open-Source Windows

[Luke]

I've never had Internet fast enough to need a/c

And the both of you conveniently choose to ignore the fact that I am talking about 80211ac.

I have a subscription to 1Gbps fibre broadband. It will be a cold day in hell if you expect me to waste all that speed and bandwidth with obsolete N-grade WiFi adapters.

having to wait almost 3 years with no drivers in the kernel is not acceptable under any means.

I've never had access to that kind of network speed. The fastest I've ever seen an Internet connection run was a bit slower than what I was able to get between a pair of computers connected by b/g grade wifi cards, and I've rarely had driver issues with those in recent years. A Tenda N compact USB device did not work when it came out, but with later kernels now works fine.

It's hard to write kernel drivers if the manufacturers won't play ball. Suppose I was doing all my mate-panel hacking with no access to the MATE source code, having not to learn C but instead having to learn to either decompile and edit or worse, figure out what everything is on a hex editor? Suppose in order to distribute the code I could not even use a decompilation of the upstream binary but had to start from scratch like the Nouveau team?

[drSeehas]

I've never had access to that kind of network speed. The fastest I've ever seen an Internet connection run was a bit slower than what I was able to get between a pair of computers connected by b/g grade wifi cards, ...

Can you imagine there are use cases for wifi cards other than WAN access?

[Luke]

Been there, done that

Can you imagine there are use cases for wifi cards other than WAN access?

Like I said I have used them peer to peer-but in that case you are again bottlenecked by the slowest card on all the machines you are networking except for networks with three or more nodes and a file transfer job bypassing the slow one. Had I not used them peer to peer I would not know that my b/g devices were good for about 2MB/sec, fastest ever seen on a WAN about 3/4 that, fastest I've connected two machines by an Ethernet cable about ten times that.

Sure as hell not going to run Windows or a closed driver just to get more network speed!

[birdie]

What does this mean, really? Sure, a program links to certain library versions, but on what systems does it not?

In reality there is no problem with binaries. If you have all the libraries they need, they run. For example I wrote this post in a browser I did not want to compile myself on archlinux, so I used one from an ubuntu ppa: https://i.imgur.com/evbcyqH.png
There are no problems with doing this.

For instance try to run Fedora version X+1 binary in Fedora version X (forward compatibility). In far too many cases it won't be possible (glibc/libglib/libgtk+/etc incompatibilities).

In certain cases you won't be able to run distro version X binary in distro version X+1 (backward compatibility).

That's absolutely unacceptable. Linux must at least provide 5+ years compatibility between all major distros so that any ISV could create software for Linux without bringing tons of libraries (like e.g. VMWare does that). Don't get me started that everything must be open source. Even if it were, new APIs mean you have to recompile everything against them, which means the behavior of your application becomes unpredictable.

And don't get me started on the Linux kernel where internal APIs aren't even defined as APIs (google for stable api nonsense).

[akincer]

For instance try to run Fedora version X+1 binary in Fedora version X (forward compatibility). In far too many cases it won't be possible (glibc/libglib/libgtk+/etc incompatibilities).

In certain cases you won't be able to run distro version X binary in distro version X+1 (backward compatibility).

That's absolutely unacceptable. Linux must at least provide 5+ years compatibility between all major distros so that any ISV could create software for Linux without bringing tons of libraries (like e.g. VMWare does that). Don't get me started that everything must be open source. Even if it were, new APIs mean you have to recompile everything against them, which means the behavior of your application becomes unpredictable.

And don't get me started on the Linux kernel where internal APIs aren't even defined as APIs (google for stable api nonsense).

What are you talking about? You write to the LSB specifications and anything outside of that you package specifically with your app which is essentially what app developers on Windows do.

[birdie]

What are you talking about? You write to the LSB specifications and anything outside of that you package specifically with your app which is essentially what app developers on Windows do.

LSB? Lol. You're talking funny.

[DanLamb]

As an accountant, I can tell you most professional accounting software runs almost exclusively in a Windows environment, unless you are dealing with an enterprise-level package from a group like Oracle or SAP. But even then, the client programs usually run in a Windows environment.

That's just one major profession out of many similar professions where nearly all of our end-user software tools are tied to Windows environments. Try document processing - do you have any idea how limited the tools are for someone who needs to process huge batches of PDF files every day? Medical billing, medical records, medical imaging - a lot of those systems are still completely tied to Windows.

Face it - if you are working in a cubicle like hundreds of millions of corporate workers or government workers worldwide, you are still very likely to be using a Wintel box out of necessity. And that's not likely to change anytime in the near future.

Sure, some specific industries are tied to Windows. I know electrical engineers heavily invested in National Instruments software, which is completely Windows centric... There are many niches like that.

Those things do change. Ten to twenty years ago, the market for custom business software development was overwhelmingly Windows+Microsoft centric. It was a rare fringe that didn't use Windows. Visual Studio, SQL Server, Visual Source Safe, and Windows were very dominant. Today, that's really changed. Things like Ruby on Rails, Python, Java, Scala, and Clojure are way more popular and databases like Postgres are considered commodity. Mac is the most popular workstation OS by a long shot, and Linux is the most popular server OS.

Another example is point-of-sale systems. Ten years ago, if you were a restaurant, you got a Windows box to run your point-of-sale system. Today, iOS has taken a lot of that market.

If I was serious about a specific industry that was super Windows centric, I'd use Windows. My dedication to my career overrides my preference for Linux/Mac. In the data science world I live in, Linux/Mac is a better choice than Windows.

[Luke_Wolf]

Those things do change. Ten to twenty years ago, the market for custom business software development was overwhelmingly Windows+Microsoft centric. It was a rare fringe that didn't use Windows. Visual Studio, SQL Server, Visual Source Safe, and Windows were very dominant. Today, that's really changed. Things like Ruby on Rails, Python, Java, Scala, and Clojure are way more popular and databases like Postgres are considered commodity.

the only language you listed that's "way more popular" is Java, yes python and friends are enjoying some mild popularity but the market for enterprise LOB falls squarely into Java vs C#, and C# is set up to consume Java's Enterprise and potentially other parts of Java's share due to the mix of .NET being open sourced and Oracle continuing to be an ass.

Mac is the most popular workstation OS by a long shot,

[citation needed]

Another example is point-of-sale systems. Ten years ago, if you were a restaurant, you got a Windows box to run your point-of-sale system. Today, iOS has taken a lot of that market.

That's funny given I see Android not iOS systems at restaurants like Chili's that are doing tablets, and dedicated systems usually running some form of windows embedded for self-service PoS at places like Jack in the Box, and for that matter places like Walmart, or the local library. in fact I've never seen an iOS device used for this purpose, and what PoS systems I do see transitioning I see going from a mix of things (Usually IBM based actually) to self-service kiosks.

Honestly you're kinda coming off as one of the Apple Loyal...

[Passso]

The biggest problem with Windows is MicroSoft desire to maintain 100% compatibility way back to god knows when. with respect to commercial software, if one look at what Apple has done over the years we can see that just maybe that obsession with the past is a bad thing. the Mac OS of today has nothing in common with the original Macs other than where the menu bar is. They then took what was Mac OS and pared and tried to make IOS which is again a successful product.

Just get a look on forums, even Apple fanboys are exhausted about buying the same softwares/hardware again and again because of this lack of compatibility. Basically the fact that Apple attracts "rich clients" make this very bankable because they can buy.

On Windows you have a wide range of budget and people are very happy to install a 10 years old little software they still use/like. I am pretty sure a lot of them would have left Windows if they did not have this compatibility. This is their real strengh.

[droidhacker]

That's funny given I see Android not iOS systems at restaurants like Chili's that are doing tablets, and dedicated systems usually running some form of windows embedded for self-service PoS at places like Jack in the Box, and for that matter places like Walmart, or the local library. in fact I've never seen an iOS device used for this purpose, and what PoS systems I do see transitioning I see going from a mix of things (Usually IBM based actually) to self-service kiosks.

Honestly you're kinda coming off as one of the Apple Loyal...

Same.
But from the developer point of view. And I work as a developer supporting the hospitality (read: restaurant) business.
Company I am working for (for now, I'm working on a new venture that should break the shackles shortly) is VERY highly into the apple junk... but even they understand that since the customer is going to be using dedicated hardware anyway (means that they need to buy equipment for the task and not just re-use existing hardware), that is makes a WHOLE lot more sense to allow the customer the option of choosing between a $60 chinese Android tablet and a $400-$500 Nexus 9 (or anything in between) rather than forcing them into a $900 apple tablet that won't even allow the customer to install the software unless we agree to the apple software distribution policies and take the risk of apple saying "no".

Note: in some applications, there is a much higher chance of the device being *physically* damaged, such as when your servers are carrying them around a busy restaurant, or even worse, when the restaurant has tablet devices that they hand over to customers for input (i.e., customer surveys). In these applications, you don't really want to force the restaurant into expensive devices that will have to be replaced frequently. A $900 apple tablet will break just as easily as a $60 chinese when dropped, but which one hurts more to replace? Actually, in this kind of application, the $900 apple tablet will likely break MORE easily than the cheap-o, since it is all tight and rigid with a glass screen, whereas the cheap-o chinese tablet has a lot more (volume-wise) flexible plastic and air space (buffer), plastic screen, etc.

So (1)... the customer is a lot happier when they can save some money, especially when you are talking about investing in a LOT of hardware and not just a single-device. A dozen + chinese tablets for the price of one apple. Two Nexus tablets for the price of one apple (and the Nexus is a much better device altogether). (2)... we save ourselves the RISK associated with trusting our business to apple. (3)... we don't LOCK the customer in to a single hardware vendor. (4)... our development costs are a lot lower for Android than they would be for apple. (5)... if you want to build software that ALLOWS the customer to re-use hardware they already have, then it makes sense to target the platform that they already have, which gauging market share over the last few years.... is Android at ~85% market share (current). Android broke over apple market share all the way back in first quarter of 2010, market leader since 4th quarter 2010 (when broke past Symbian on its death spiral), and broke over into MAJORITY market share in 2nd quarter 2011 -- that means that more Android devices sold in EVERY quarter since then than the sum of ALL competitors.