You can't move RPMs that easily. You end up in more or less the pre RPM days of resolving library dependencies by hand. Sure I've used Fedora RPMs on RHEL. But you have to worry about same compiler, same library names, compatible versions.... You end up doing some hairy stuff.
I'm not the one who is applying this concept, Microsoft is. And we mostly have a situation today of computerized devices not talking because they don't share data or applications. So I think Microsoft is right in challenging the notion that lots of computing is ubiquitous computing.
But regardless it ain't my theory I'm just explaining the difference between what Microsoft proposes and what Apple proposes.
I don't know how Bruce will answer this, and it could be interesting given he took over from Ian one of the largest still ongoing collaborations.
But... when exactly do you think there was not binary incompatibility between distributions? When the system wasn't fragmented and fractured. Those have been constants.
Bruce, you were the founder of UserLinux which aimed to create binary compatibility for Linux, a simple VAR platform. Google with Android attempted something similar. How well do you believe Android fulfills the objectives you set out for UserLinux. And where they missed do you believe those misses were unavoidable given the changes in focus (desktop vs. handset) or something where a minor change of strategy could allow them to achieve those missed objectives?
I assume you mean source from the 70s not executables. I can't think of any UNIX that's been around since the 70s. Source compatibility is much easier than binary compatibility, which is why Linux targets the former not the later.
As for guarantees there are no guarantees. Microsoft owns the platform. They will do what they think is in the platform's best interest. They have a long history of being concerned with binary compatibility even at tremendous cost and effort. They on the other hand have shown a willingness to not consider it a top priority, i.e. they've never subscribed to the "bug for bug compatible" model.
All those various parties are just a frictional costs. The play the same role as broker fees its just that they are being broken out individually for you. High broker frees reduce the value of assets because they reduce liquidity, they don't increase it.
Thank you. Most users do seem to prefer the Apple model and obviously their's is sort of working already. Microsoft's will be amazing if it works but is quite likely to fail. I'm just glad to see Microsoft leading the industry and making gusty calls. I think it is exciting that we are going to have two genuinely different paradigms. And if you count Android / Windows 7 as the loose system 3 paradigms.
Well if you are right then Windows 8 fails, Microsoft has to exit the consumer market and the battle of the 2020s is over enterprise with Microsoft on Windows 7ish systems defending the enterprise space from the now much mature applications migrating from the vibrant tablet / smartphone space. Repeating how they beat DEC, IBM and Unisys with them in the opposite role.
I think what Microsoft is aiming for is possible, I agree it is more difficult than dedicated. But ultimately they lost the initial battles for the smartphone space. The battle they need to fight is consumer computing in about 5 years. Different products for different devices introduces tremendous system and data integration hassles. As iOS and Android become more important I think the downsides of separate evolution will become more apparent. iOS at least has a big cousin to act as a bridge system.
Did you click on the link: http://www.youtube.com/watch?v=a6cNdhOKwi0 ? That's what Microsoft pictures in terms of long term direction. Those things have to blur. Watch for example how her smartphone applications moves to her tablet moves to the full desktop application in the video.
Apple actually offers a spreadsheet application for tablet, Numbers of iOS. The emphasis there is: viewing + light editing. While Numbers for OSX is designed primarily around creation, heavy editing and publication. That's the sort of thing that needs to happen for applications to move to even the first step.
, but I've yet to see a single paradigm that works *well* in all of those places.
Well just to pick a few examples.
The Linux kernel which is running on $3 embedded systems and at the same time is the dominant system for super computing and 2nd most popular virtual environment for the z-series (mainframe).
Oracle which runs well on small embedded systems (say about $200) all the way up to the largest databases in the world
Apache....
Now another example for Windows is SQL Server. SQL Server's developer product runs on small X86 apps that cost $39.95. SQL Server is also giving Oracle and DB2 this decade a run for the money on data warehouses.
____
I agree though that scaling is a real issue and takes testing and development. It doesn't happen automatically. All sorts of things start to break that people never considered.
decades ago Unix, standard open API and portable C provided a superior solution to your imagined "sames applications" running on wide range of devices
No it didn't. Unix were expensive workstations generally 5 figures. They also didn't port to the large stuff. UNIX a few decades ago was a niche. Since then obviously it has expanded out to cover the entire range. But UNIXes are a family of operating systems not an operating system.
What I'd say today is that applications do not move seamlessly from UNIX to UNIX though most UNIX applications are ported via. the distribution / source methods. There is no data integration at all between UNIXes so the end user ends up having to provide all the integration for themselves. UNIXes features depend on the user / developer model which while quite empowering have failed to catch on. Certainly UNIX offers an alternative approach but my point was the distinction between Apple's approach and Microsoft's.
As for Windows embedded, Windows embedded and Windows both run Metro apps and Visual Studio is allowing for the cross application ports. They aren't there yet, but it is their direction. And this entire post was about Microsoft's direction with Windows 8 not their current state with Windows 7.
Finally as far as malware... Windows evolved from a single user OS, in a world of isolated computers. Moving an entire platform quickly is complex. I don't agree with all their choices, but in terms of ubiquitousness, it is hard to argue their choices didn't work out.
Understood. I'm an apple guy but I fully understand the appeal of Microsoft's strategy if they can get it to work.
Linux is a yet another model... The ability to construct an environment for any hardware platform. Linux long ago decided for source compatibility (and they've done a good job) not binary compatibility.
Microsoft itself has created an operating system and application platform which allows the same applications to run on a $200 throw away netbook and a $2000 workstation or a $20,000 multiprocessor array drive server. Nothing like that existed a few decades ago.
The API are going to need to mature and stabilize before their can be an abstraction layer. Right now what we want is to get these APIs to be feature rich, which means rapid changing and evolving, which is best achieved by each of these agents evolving their products as quickly as possible. Once the evolution has mostly stopped then is the time for standardization. Far from we should have put this to rest a decade ago, we are probably at least 2 decades off from this being the right time.
On the other hand for functionality that is needed cross platform the best solution is a single purpose application which ports across the systems. To use your finding friends example foursquare.
Just to point out here the assumption of the question is wrong. Apple is proposing the exact opposite of ubiquitous computing. They instead have two products iOS and OSX which evolve semi-seperately so that data can pass between similar applications but that the applications are quite different.
Microsoft conversely is proposing a shift to ubiquitous computing that applications and devices can alter themselves based on the way they are used, the form factor of the human. ( http://www.youtube.com/watch?v=a6cNdhOKwi0 ).
What the author is proposing is different from either one of these, the current situation where there are loose standards for moving application data and different applications.
And then what? With a generic OS they end up in the same position as the PC manufacturers as commodity vendors selling their phones for slightly more than the cost of parts. Where is the upside there?
BTW Jolla's MeeGo runs Android apps as does RIMs system.
Slashdot Mirror
The days of unshared code was rather common by the 70s. That's long before there was a Linux.
You can't move RPMs that easily. You end up in more or less the pre RPM days of resolving library dependencies by hand. Sure I've used Fedora RPMs on RHEL. But you have to worry about same compiler, same library names, compatible versions.... You end up doing some hairy stuff.
I'm not the one who is applying this concept, Microsoft is. And we mostly have a situation today of computerized devices not talking because they don't share data or applications. So I think Microsoft is right in challenging the notion that lots of computing is ubiquitous computing.
But regardless it ain't my theory I'm just explaining the difference between what Microsoft proposes and what Apple proposes.
I don't know how Bruce will answer this, and it could be interesting given he took over from Ian one of the largest still ongoing collaborations.
But... when exactly do you think there was not binary incompatibility between distributions? When the system wasn't fragmented and fractured. Those have been constants.
Bruce, you were the founder of UserLinux which aimed to create binary compatibility for Linux, a simple VAR platform. Google with Android attempted something similar. How well do you believe Android fulfills the objectives you set out for UserLinux. And where they missed do you believe those misses were unavoidable given the changes in focus (desktop vs. handset) or something where a minor change of strategy could allow them to achieve those missed objectives?
I assume you mean source from the 70s not executables. I can't think of any UNIX that's been around since the 70s. Source compatibility is much easier than binary compatibility, which is why Linux targets the former not the later.
As for guarantees there are no guarantees. Microsoft owns the platform. They will do what they think is in the platform's best interest. They have a long history of being concerned with binary compatibility even at tremendous cost and effort. They on the other hand have shown a willingness to not consider it a top priority, i.e. they've never subscribed to the "bug for bug compatible" model.
All those various parties are just a frictional costs. The play the same role as broker fees its just that they are being broken out individually for you. High broker frees reduce the value of assets because they reduce liquidity, they don't increase it.
Thank you. Most users do seem to prefer the Apple model and obviously their's is sort of working already. Microsoft's will be amazing if it works but is quite likely to fail. I'm just glad to see Microsoft leading the industry and making gusty calls. I think it is exciting that we are going to have two genuinely different paradigms. And if you count Android / Windows 7 as the loose system 3 paradigms.
Nokia doesn't have a lower cost of components. They would be more like DEC or SGI when they tried to move to PCs.
Well if you are right then Windows 8 fails, Microsoft has to exit the consumer market and the battle of the 2020s is over enterprise with Microsoft on Windows 7ish systems defending the enterprise space from the now much mature applications migrating from the vibrant tablet / smartphone space. Repeating how they beat DEC, IBM and Unisys with them in the opposite role.
I think what Microsoft is aiming for is possible, I agree it is more difficult than dedicated. But ultimately they lost the initial battles for the smartphone space. The battle they need to fight is consumer computing in about 5 years. Different products for different devices introduces tremendous system and data integration hassles. As iOS and Android become more important I think the downsides of separate evolution will become more apparent. iOS at least has a big cousin to act as a bridge system.
Did you click on the link: http://www.youtube.com/watch?v=a6cNdhOKwi0 ? That's what Microsoft pictures in terms of long term direction. Those things have to blur. Watch for example how her smartphone applications moves to her tablet moves to the full desktop application in the video.
Apple actually offers a spreadsheet application for tablet, Numbers of iOS. The emphasis there is: viewing + light editing. While Numbers for OSX is designed primarily around creation, heavy editing and publication. That's the sort of thing that needs to happen for applications to move to even the first step.
Reread the response, GP's claim was that these systems didn't exist.
, but I've yet to see a single paradigm that works *well* in all of those places.
Well just to pick a few examples.
The Linux kernel which is running on $3 embedded systems and at the same time is the dominant system for super computing and 2nd most popular virtual environment for the z-series (mainframe).
Oracle which runs well on small embedded systems (say about $200) all the way up to the largest databases in the world
Apache....
Now another example for Windows is SQL Server. SQL Server's developer product runs on small X86 apps that cost $39.95. SQL Server is also giving Oracle and DB2 this decade a run for the money on data warehouses.
____
I agree though that scaling is a real issue and takes testing and development. It doesn't happen automatically. All sorts of things start to break that people never considered.
There are counter arguments to ubiquity. And your criticisms are spot on. '
But GP was claiming this didn't exist, I argued it did. You are arguing it is a permanently bad idea. I'm on the fence on that one.
decades ago Unix, standard open API and portable C provided a superior solution to your imagined "sames applications" running on wide range of devices
No it didn't. Unix were expensive workstations generally 5 figures. They also didn't port to the large stuff. UNIX a few decades ago was a niche. Since then obviously it has expanded out to cover the entire range. But UNIXes are a family of operating systems not an operating system.
What I'd say today is that applications do not move seamlessly from UNIX to UNIX though most UNIX applications are ported via. the distribution / source methods. There is no data integration at all between UNIXes so the end user ends up having to provide all the integration for themselves. UNIXes features depend on the user / developer model which while quite empowering have failed to catch on. Certainly UNIX offers an alternative approach but my point was the distinction between Apple's approach and Microsoft's.
As for Windows embedded, Windows embedded and Windows both run Metro apps and Visual Studio is allowing for the cross application ports. They aren't there yet, but it is their direction. And this entire post was about Microsoft's direction with Windows 8 not their current state with Windows 7.
Finally as far as malware... Windows evolved from a single user OS, in a world of isolated computers. Moving an entire platform quickly is complex. I don't agree with all their choices, but in terms of ubiquitousness, it is hard to argue their choices didn't work out.
Hmmm.... good point.
I'd say the consumer space was more diverse price wise but with phones that isn't true either.
Understood. I'm an apple guy but I fully understand the appeal of Microsoft's strategy if they can get it to work.
Linux is a yet another model... The ability to construct an environment for any hardware platform. Linux long ago decided for source compatibility (and they've done a good job) not binary compatibility.
In 1981 you had greater diversity. Computers at the $99 price point all the way up to tens of thousands.
First off all not guaranteed. Linux lost on the desktop to Windows for example.
But the point is why would a manufacturer want to target that. They work really hard to produce lots of product and make no money.
Microsoft itself has created an operating system and application platform which allows the same applications to run on a $200 throw away netbook and a $2000 workstation or a $20,000 multiprocessor array drive server. Nothing like that existed a few decades ago.
Perfect no. Impressive progress, yes.
The API are going to need to mature and stabilize before their can be an abstraction layer. Right now what we want is to get these APIs to be feature rich, which means rapid changing and evolving, which is best achieved by each of these agents evolving their products as quickly as possible. Once the evolution has mostly stopped then is the time for standardization. Far from we should have put this to rest a decade ago, we are probably at least 2 decades off from this being the right time.
On the other hand for functionality that is needed cross platform the best solution is a single purpose application which ports across the systems. To use your finding friends example foursquare.
Just to point out here the assumption of the question is wrong. Apple is proposing the exact opposite of ubiquitous computing. They instead have two products iOS and OSX which evolve semi-seperately so that data can pass between similar applications but that the applications are quite different.
Microsoft conversely is proposing a shift to ubiquitous computing that applications and devices can alter themselves based on the way they are used, the form factor of the human. ( http://www.youtube.com/watch?v=a6cNdhOKwi0 ).
What the author is proposing is different from either one of these, the current situation where there are loose standards for moving application data and different applications.
current: loose standards
apple: seamless data portability
microsoft: seamless application portability
Apple's views and Microsoft's views shouldn't be confused.
And then what? With a generic OS they end up in the same position as the PC manufacturers as commodity vendors selling their phones for slightly more than the cost of parts. Where is the upside there?
BTW Jolla's MeeGo runs Android apps as does RIMs system.
x11 has problems with high end video cards on desktops and $2000 laptops. On a phone its not really that outdated. And I say that as a Wayland fan.
jolla is going to be continuing the Nokia MeeGo line.