There issue with #2 is not so much the speed of Java as the fact that Java developers by their very nature are cross platform developers. Apple wants to avoid cross platform applications because such applications tend not to take maximum advantage of the underlying hardware and OS. I don't think they would object to a 3rd party "Apple Java", like the old Microsoft Java, designed to link directly with Cocoa and tuned for the iPhone / iPad / iPod. Their issue with Java is precisely the fact that these are port of Android apps.
There is an easier approach. Compress margins. Apple's gross margins are around 70%. They iPhone 4S could be $0 after carrier subsidy and Apple would still be making good money. If they worked at it, Apple could work with the carriers to offer a feature phone version sell it for $0 with a much lower subsidy and still make a slim profit.
2 weeks? Thats not a problem for iPhone. iPhone apps sell well for a few years with minor updates.
You got too many iPhone users, like me, who are old farts. I can't believe how much patient I am at 42 then I was at 18. At 18 I had to see the movie opening day. At 42... at heck it will be on TV in a few years.
Most of the Android phones were sold at carriers other than AT&T and at much lower price points. 2012 when Android comes up for renewal at Sprint and Verizon will be interesting. You'll have iPhones at all the same Android price points and available from the same carriers with the same data options. It really will be (excuse the pun) an apples to apples comparison. Many of the key aspects of Android like:
-- Poor / no updates for OSes -- Incompatibility for apps. -- Crippled features from carriers -- Microsoft taking a licensing fee
are starting to bite.
I sincerely hope that Android continues to do well. While I'm an iPhone user I think the competition is in every consumer's interest. And certainly when I bought my iPhone I looked long and hard at some HTC models. While software helped, it was ultimately hardware, not software, that made me pick the iPhone.
There was a period where Mac's market share fell to around 2%. But it was always a premium 2%. The 2% that were most interested in quality and willing to pay. That's why Mac when it was around 6-7% market share represented something like 50% of all the profits. Mac users, spend much more on hardware and software.
And that carries over. For example the iOS market place is 7x the size in dollar terms of the Android, Blackberry and Nokia marketplace combined.
The problem for the Linux market on the desktop is not just the lack of share but the lack of a market. Linux wants the low end. Microsoft however, unlike the server market, is willing to price themselves down far enough to compete for the low end. In the server market Microsoft (like Apple) choose margin over marketshare, on the desktop they choose the reverse.
The unity around Ubuntu was because Ubuntu was essentially universally regarded as the easiest desktop to use. One of two things will happen:
a) The move to Unity will be successful and Ubuntu will continue to be the easiest desktop. New users who don't Unity will discover that Linux means choice quickly.
b) The move to Unity will be unsuccessful and the beginner Linux distribution will fragment again, like it was before Ubuntu. That will create the space for lots of differently tailored beginner level Linuxes which aim for slightly different niches.
b2) The move to Unity will be unsuccessful and one beginner Linux distribution moves to the front of the pack almost instantly.
Just as an aside, Slackware doesn't even have Gnome as a package. You actually have to get one of the other distributions based on slackware to setup Gnome.
Though also it is important to note, this is about the degree of vertical integration required and predates the latest issues regarding Gnome2 -> Gnome3.
RedHat defaults to Gnome, a good chunk of Gnome development is paid for by RedHat. You are getting some good answers below. Suse being the big one. But Kanotix is the one I'd add to the list.
Absolutely and C++ developers frequently were coming over from the Windows world and had never done lower level stuff. So while they could make Linux applications the low level stuff required for a GUI was beyond them.
True but that was mainly over the whole problem of having to maintain an integrated software stack which goes against Slackware's philosophy. Patrick even pointed people to GNOME Slackbuild (GSB), GWARE, Dropline.... to get the integrated stack with a slackware feel.
Gnome was a pretty big project in its own right. I think if they were following the GNU coding guidelines, Gnome would be a lot more scriptable and there would be a lot more Scheme in Gnome. Frankly I wish they had been following the coding guidelines, a highly scriptable modern GUI would be wonderful.
As for GTK bindings and wrappers, GTK pre-existed Gnome. It was originally the "GIMP Toolkit". And I agree GTK has a long history of working pretty well with non C based languages.
There was a project to make an LGPL QT clone called Harmony. It didn't attract a ton of developers. Strategically the FSF (and Harmony was on board) was that the desktop needed to go first. Otherwise, Harmony would be chasing Trolltech and the free Harmony based desktop would be years behind the proprietary QT based desktop. The free version would be a poor quality knock off of the original.
That is essentially the situation that GnuSTEP has always found themselves in. They can't lead they have to follow.
So yes, what you are proposing was in fact what they were doing.
Not sure what iOS has to do with this. An end user does need to recompile their apps if the distribution discontinues support for it. Assuming the dependent libraries are still available on the new platform, or can be compiled against the new system without problems.
IOS has the same model you are saying is unacceptable. On IOS after a OS upgrade any number of apps can be broken. However, the end user redownloads those apps from the app store. Apple can and does break binary compatibility all the time and upgrading is a regular part of the process.
End users don't have any problem with it at all. They don't care about poor binary compatibility as long as they can redownload easily and their data remains intact. What IOS does is provides a counter example to your theory that end users demand binary compatibility over the Linux distribution system.
Linux does not have or want a binary compatibility model. If you disagree with that, you disagree with a fundamental design choice of the product. It is not a lack of change control but a lack of Linux doing something it has never asserted any interest in doing. For open source apps the model is the distribution recompiles. For closed source apps the model is becoming the app releases itself with a specific version of the OS and runs inside a VM.
I think you are wrong that it is a bad thing. I consider the distribution model to be a huge plus of Linux, one of my favorite thing about it, and when I'm on Windows or OSX and having to deal with 40 different vendors and contracts and license management and working through products that are only semi compatible and lie to me about when they are going to update, it is misery. And this vs. something like "yum update (package name)". I'll take the Linux model any day. But your problem isn't KDE or Gnome it goes back to Linus.
It is funny you mention BSDs. With BSDs I have to recompile all the time. I like ports but I don't find the platform binary stable. I actually find myself dealing with source related issues more often.
Anyway... a very stable Unix with good binary support I'd have said Solaris. Probably the only one left at this point is AIX that really does what you want, and IBM's commitment to AIX is iffy.
and this is why it is still a niche (and will remain so). End user doesn't care for recompiling apps,
The model of having to redownload apps is the model on iOS. End users seem fine with it. Under the Linux distribution model end users don't have to recompile apps but distributions do.
If you want slower turnover, and don't want to recompile, and long term app support have you considered Debian?
Its probably fixed now, but an instance of my point about stuff being rushed out the door "half-assed".
I agree all the desktops don't do much Q&A. I find the general lack of quality in Linux pretty infuriating. But I understand, QA takes a lot of money and a lot of time. That being said, things working the same way for almost 20 years is why I like WindowMaker when I can.
The issue is that Windows adopted the Aero interface. That means both Mac and Windows have interfaces vastly more sophisticated than what is available on KDE2. If they didn't do the eye candy work Linux desktop would look a decade behind minimum.
Also there have been features added, for example unified notification, that is all applications being able to send messages to end users in a unified way that is configurable by the end user. That is a major shift for both KDE and Gnome.
As for old apps, this is Linux apps should recompile and be sent out by the distribution. Linux has never sought binary compatibility.
At the time objective-C developers were, if they were interested in desktop development working on Mac, OpenSTEP or GNUStep. Java was too slow for a desktop and had bad Linux support, though this was a major consideration. QT was amazingly good for C++, Gnome couldn't compete. So they created a system for C programmers who didn't know C++.
They had to reinvent the wheel because they had to recruit people. There were a lot of C programmers that were willing to work on Linux desktop apps.
Gnome is doing a major breaking change, that is going to cause a huge shift in its user base. Ubuntu created a whole generation of Linux users that were Gnome users by default. Those people are now picking their replacement desktop and well.... that's a good chunk of the/. Linux users.
Nothing went wrong, Gnome's intention was to make this change.
They have one oxygen-gtk makes KDE look a lot like gnome 2, including the apps.
Anyway I wouldn't worry: openSuse is a big player. Mandriva, Linpus, Kanotix aren't small. And there are plenty of other distributions like Mint which have KDE versions. Frankly I think Ubuntu is about to hit a brick wall with their approach to Gnome which leaves Kubuntu as a possible alternative.
I don't think there is any doubt they mismanaged the transition. I think that a lot of the Gnome users will end up moving to KDE. Lets not forget that Gnome 2 is still available, in other 2 years.... And people who have been with Gnome for years are experienced enough to enjoy the features of KDE. Don't forget that Ubuntu is currently forking Gnome in a way that most likely will not work for them. I can see Ubuntu being forced to switch over.
It wouldn't shock me at all to see KDE move back into first place by say 2015.
As far as not listening... you have to remember that KDE came in for some massive criticism that was very unfair (from their perspective) in the late 1990s due to the whole QPL issue and the failure of United Linux. They had fallen from the Linux GUI to deep 2nd place already. And they had to survive blistering criticism that was biased. I think the thick skin from the Gnome / KDE wars didn't serve them well here. They didn't really understand that level of antagonism that the move would generate, they screwed up but won't make that mistake again.
In the end I think KDE is the only full featured high end power desktop left.
why do all that work (loading, configuring, and then swapping it to disk) just so you have something that you don't use.
Because loading a few parts of a fully configured subsystem from virtual memory is much cheaper than loading tons of libraries, initializing them, using them, and then shutting them all down. Frequently used services should start up. Absolutely that leads to slower startup of OSes, it also leads to more responsive OSes.
Further it makes the system scale up well to the next hardware level. Getting more hardware leads to a sharp improvement in performance. A system designed to really function well at say 256mb won't likely see any difference between 4g and 8b.
I'd say that's a rather specific complaint about a massive overhaul of an entire GUI. That might be configurable, it might not but that amounts to disliking a new model of a car because they changed the shape of the ashtray. The ashtray might be a real problem but...
KDE 4 at this point is better than 3. It just took time.
As for Knuth, in practice that is not what happened. TeX got replaced with AMSTeX and LaTeX then AMSLaTeX, adding huge numbers of macros. This moved on towards LaTeX 2e adding even more. The engine itself got replaced with the PDFTeX engine which is now standard. That caused Metafont to mostly be dropped which was key to the whole system, as well as DVI. The entire character system got chucked with the Omaga and now XeTeX / XeLaTeX.
So how exactly in a practical sense is Knuth not an example of upgrading overtime with dramatic replacement?
Slashdot Mirror
Good analysis.
There issue with #2 is not so much the speed of Java as the fact that Java developers by their very nature are cross platform developers. Apple wants to avoid cross platform applications because such applications tend not to take maximum advantage of the underlying hardware and OS. I don't think they would object to a 3rd party "Apple Java", like the old Microsoft Java, designed to link directly with Cocoa and tuned for the iPhone / iPad / iPod. Their issue with Java is precisely the fact that these are port of Android apps.
There is an easier approach. Compress margins. Apple's gross margins are around 70%. They iPhone 4S could be $0 after carrier subsidy and Apple would still be making good money. If they worked at it, Apple could work with the carriers to offer a feature phone version sell it for $0 with a much lower subsidy and still make a slim profit.
2 weeks? Thats not a problem for iPhone. iPhone apps sell well for a few years with minor updates.
You got too many iPhone users, like me, who are old farts. I can't believe how much patient I am at 42 then I was at 18. At 18 I had to see the movie opening day. At 42... at heck it will be on TV in a few years.
Most of the Android phones were sold at carriers other than AT&T and at much lower price points. 2012 when Android comes up for renewal at Sprint and Verizon will be interesting. You'll have iPhones at all the same Android price points and available from the same carriers with the same data options. It really will be (excuse the pun) an apples to apples comparison. Many of the key aspects of Android like:
-- Poor / no updates for OSes
-- Incompatibility for apps.
-- Crippled features from carriers
-- Microsoft taking a licensing fee
are starting to bite.
I sincerely hope that Android continues to do well. While I'm an iPhone user I think the competition is in every consumer's interest. And certainly when I bought my iPhone I looked long and hard at some HTC models. While software helped, it was ultimately hardware, not software, that made me pick the iPhone.
There was a period where Mac's market share fell to around 2%. But it was always a premium 2%. The 2% that were most interested in quality and willing to pay. That's why Mac when it was around 6-7% market share represented something like 50% of all the profits. Mac users, spend much more on hardware and software.
And that carries over. For example the iOS market place is 7x the size in dollar terms of the Android, Blackberry and Nokia marketplace combined.
The problem for the Linux market on the desktop is not just the lack of share but the lack of a market. Linux wants the low end. Microsoft however, unlike the server market, is willing to price themselves down far enough to compete for the low end. In the server market Microsoft (like Apple) choose margin over marketshare, on the desktop they choose the reverse.
kanotix might be worth trying if you like Knoppix and KDE.
The unity around Ubuntu was because Ubuntu was essentially universally regarded as the easiest desktop to use. One of two things will happen:
a) The move to Unity will be successful and Ubuntu will continue to be the easiest desktop. New users who don't Unity will discover that Linux means choice quickly.
b) The move to Unity will be unsuccessful and the beginner Linux distribution will fragment again, like it was before Ubuntu. That will create the space for lots of differently tailored beginner level Linuxes which aim for slightly different niches.
b2) The move to Unity will be unsuccessful and one beginner Linux distribution moves to the front of the pack almost instantly.
Just as an aside, Slackware doesn't even have Gnome as a package. You actually have to get one of the other distributions based on slackware to setup Gnome.
Though also it is important to note, this is about the degree of vertical integration required and predates the latest issues regarding Gnome2 -> Gnome3.
RedHat defaults to Gnome, a good chunk of Gnome development is paid for by RedHat. You are getting some good answers below. Suse being the big one. But Kanotix is the one I'd add to the list.
Absolutely and C++ developers frequently were coming over from the Windows world and had never done lower level stuff. So while they could make Linux applications the low level stuff required for a GUI was beyond them.
True but that was mainly over the whole problem of having to maintain an integrated software stack which goes against Slackware's philosophy. Patrick even pointed people to GNOME Slackbuild (GSB), GWARE, Dropline.... to get the integrated stack with a slackware feel.
Gnome was a pretty big project in its own right. I think if they were following the GNU coding guidelines, Gnome would be a lot more scriptable and there would be a lot more Scheme in Gnome. Frankly I wish they had been following the coding guidelines, a highly scriptable modern GUI would be wonderful.
As for GTK bindings and wrappers, GTK pre-existed Gnome. It was originally the "GIMP Toolkit". And I agree GTK has a long history of working pretty well with non C based languages.
There was a project to make an LGPL QT clone called Harmony. It didn't attract a ton of developers. Strategically the FSF (and Harmony was on board) was that the desktop needed to go first. Otherwise, Harmony would be chasing Trolltech and the free Harmony based desktop would be years behind the proprietary QT based desktop. The free version would be a poor quality knock off of the original.
That is essentially the situation that GnuSTEP has always found themselves in. They can't lead they have to follow.
So yes, what you are proposing was in fact what they were doing.
IOS has the same model you are saying is unacceptable. On IOS after a OS upgrade any number of apps can be broken. However, the end user redownloads those apps from the app store. Apple can and does break binary compatibility all the time and upgrading is a regular part of the process.
End users don't have any problem with it at all. They don't care about poor binary compatibility as long as they can redownload easily and their data remains intact. What IOS does is provides a counter example to your theory that end users demand binary compatibility over the Linux distribution system.
Linux does not have or want a binary compatibility model. If you disagree with that, you disagree with a fundamental design choice of the product. It is not a lack of change control but a lack of Linux doing something it has never asserted any interest in doing. For open source apps the model is the distribution recompiles. For closed source apps the model is becoming the app releases itself with a specific version of the OS and runs inside a VM.
I think you are wrong that it is a bad thing. I consider the distribution model to be a huge plus of Linux, one of my favorite thing about it, and when I'm on Windows or OSX and having to deal with 40 different vendors and contracts and license management and working through products that are only semi compatible and lie to me about when they are going to update, it is misery. And this vs. something like "yum update (package name)". I'll take the Linux model any day. But your problem isn't KDE or Gnome it goes back to Linus.
It is funny you mention BSDs. With BSDs I have to recompile all the time. I like ports but I don't find the platform binary stable. I actually find myself dealing with source related issues more often.
Anyway... a very stable Unix with good binary support I'd have said Solaris. Probably the only one left at this point is AIX that really does what you want, and IBM's commitment to AIX is iffy.
Compiz subs out fine for Mutter or KWin. Almost all the distributions include it. Mandriva and Suse both do this subout.
The model of having to redownload apps is the model on iOS. End users seem fine with it. Under the Linux distribution model end users don't have to recompile apps but distributions do.
If you want slower turnover, and don't want to recompile, and long term app support have you considered Debian?
I agree all the desktops don't do much Q&A. I find the general lack of quality in Linux pretty infuriating. But I understand, QA takes a lot of money and a lot of time. That being said, things working the same way for almost 20 years is why I like WindowMaker when I can.
The issue is that Windows adopted the Aero interface. That means both Mac and Windows have interfaces vastly more sophisticated than what is available on KDE2. If they didn't do the eye candy work Linux desktop would look a decade behind minimum.
Also there have been features added, for example unified notification, that is all applications being able to send messages to end users in a unified way that is configurable by the end user. That is a major shift for both KDE and Gnome.
As for old apps, this is Linux apps should recompile and be sent out by the distribution. Linux has never sought binary compatibility.
At the time objective-C developers were, if they were interested in desktop development working on Mac, OpenSTEP or GNUStep. Java was too slow for a desktop and had bad Linux support, though this was a major consideration. QT was amazingly good for C++, Gnome couldn't compete. So they created a system for C programmers who didn't know C++.
They had to reinvent the wheel because they had to recruit people. There were a lot of C programmers that were willing to work on Linux desktop apps.
Gnome is doing a major breaking change, that is going to cause a huge shift in its user base. Ubuntu created a whole generation of Linux users that were Gnome users by default. Those people are now picking their replacement desktop and well.... that's a good chunk of the /. Linux users.
Nothing went wrong, Gnome's intention was to make this change.
They have one oxygen-gtk makes KDE look a lot like gnome 2, including the apps.
Anyway I wouldn't worry: openSuse is a big player. Mandriva, Linpus, Kanotix aren't small. And there are plenty of other distributions like Mint which have KDE versions. Frankly I think Ubuntu is about to hit a brick wall with their approach to Gnome which leaves Kubuntu as a possible alternative.
I don't think there is any doubt they mismanaged the transition. I think that a lot of the Gnome users will end up moving to KDE. Lets not forget that Gnome 2 is still available, in other 2 years.... And people who have been with Gnome for years are experienced enough to enjoy the features of KDE. Don't forget that Ubuntu is currently forking Gnome in a way that most likely will not work for them. I can see Ubuntu being forced to switch over.
It wouldn't shock me at all to see KDE move back into first place by say 2015.
As far as not listening... you have to remember that KDE came in for some massive criticism that was very unfair (from their perspective) in the late 1990s due to the whole QPL issue and the failure of United Linux. They had fallen from the Linux GUI to deep 2nd place already. And they had to survive blistering criticism that was biased. I think the thick skin from the Gnome / KDE wars didn't serve them well here. They didn't really understand that level of antagonism that the move would generate, they screwed up but won't make that mistake again.
In the end I think KDE is the only full featured high end power desktop left.
Because loading a few parts of a fully configured subsystem from virtual memory is much cheaper than loading tons of libraries, initializing them, using them, and then shutting them all down. Frequently used services should start up. Absolutely that leads to slower startup of OSes, it also leads to more responsive OSes.
Further it makes the system scale up well to the next hardware level. Getting more hardware leads to a sharp improvement in performance. A system designed to really function well at say 256mb won't likely see any difference between 4g and 8b.
I'd say that's a rather specific complaint about a massive overhaul of an entire GUI. That might be configurable, it might not but that amounts to disliking a new model of a car because they changed the shape of the ashtray. The ashtray might be a real problem but ...
KDE is thriving at this point. They took a huge hit from the KDE 3 -> KDE 4 transition but now they get to enjoy the fruit of the upgrade.
KDE 4 at this point is better than 3. It just took time.
As for Knuth, in practice that is not what happened.
TeX got replaced with AMSTeX and LaTeX then AMSLaTeX, adding huge numbers of macros. This moved on towards LaTeX 2e adding even more.
The engine itself got replaced with the PDFTeX engine which is now standard. That caused Metafont to mostly be dropped which was key to the whole system, as well as DVI.
The entire character system got chucked with the Omaga and now XeTeX / XeLaTeX.
So how exactly in a practical sense is Knuth not an example of upgrading overtime with dramatic replacement?