I guess I'd call that paper an example of large amounts of sloppy guesswork. He doesn't make any observations of an actual Vista box to see if any of his conjectures or expected outcomes were true. Overall the paper hasn't aged well. It's like claiming that new cars can only go a max of 20 mph while commuters are daily using them and going 55 on the Interstate. Occationally someone has a engine malfunction or runs out of gas and suddenly believes "He was right! My car can't go 55", immediately forgetting what they just did yesterday, but that's religion for ya.
It more then a style choice, it's a reflection of the customer base and support model.
Windows does corporate support which forces there to be a large number of fixes are done for individual customers or given out only if you a specific problem (to minimize risk to the general population and lower testing costs). This prevents a single release wide number that could be incremented. The service pack release collapses all these combinatorial configuration possibilities back into a single point and sometimes adds back ports of features in later OS releases.
Apple on the other hand doesn't really do corporate support, so they can go with the simpler model of keeping everyone on the same linear update branch.
> What incentive would anyone have now to provide "mature" drivers for Vista?
Because they want thier products to get shipped by OEMs and bought by users; if they don't improve thier drivers performance they lose on benchmarks, if they don't improve driver quality OEMs get upset. There is also marketing dollars when you get a "Designed for Windows Vista" Logo.
>How can any manufacturer know that their stuff will even be allowed to work with Win 7, when it comes out, or that the driver API won't be as changed as to make the improvements they're currently making to their drivers obsolete?
I would expect that any new OS version would enable new opportunities for performance tuning, but I seriously doubt that there will be another fundimental driver api shift since we just went through that with Vista and we still have plenty of milage to get out of the current generation.
There was a serious credability gap with the Vista CTP/Betas in when the OS was going to ship (even internally). So when it did ship many of our partners were caught unprepared. It has taken a year for everyone to catch up. PDC and WinHEC are pretty close together this year (Late Oct, early Nov), so the industry will have a gut check at that time to determine if the ship date is creditable.
A test suite wouldn't have helped. Win2k worked just fine with normal kerberos as a client and as a server. The problem was that if you wanted to deal with domain based groups you needed an extension, something that MSFT wasn't intrested in letting people have for free.
> No company is going to be allowed to implement it and become a competitive threat to Microsoft. Microsoft will shut them down with Patent Violations.
Again BS: The OSP pretty much crushes that arguement. Even the main point in the SFLC anaylsis about future versions has been addressed.
Do I implement a brand new spec that is clean because it ignores and wishes away all the legacy of the past, or do I implement a more complex spec that transitions them?
Developers like the earlier Users like the later.
OSS is typically dev oriented. Commercial software is user oriented (because they have to be).
You want to start experiencing what a real Mesh feels like, install foldershare. You can use any application to work with your data rather then be tied to one application (like google docs). It's been killer for keeping everything between my work, home and laptop machines up to date and also sharing full quality photos to my extended family. Once we all add in the ability for web applications to access and stay in sync with my data and we will be cooking with fire.
>2) If you use or redistribute software, and the software is released under the GPLv3 license, you have legal assurances that you will not wake up one morning and find that the software you have come to rely on is now subject to patents that the vendor received.
Actually for number 2, it only helps you if the person who wrote/distributed the software has the patents. You still have just as much risk against anyone else (which is where the real risk is anyways).
That doesn't help you if you aren't tying yourself to the GPL, so if I want to release a version under the BSD I'm just as potentially SOL with the IBM or Sun patent pledges as the Microsoft one.
>Doing mysterious DRM checks while copying files at a rate that would embarrass a TRS-80 Model 1
What are you talking about? There are no system "DRM checks". There is an API that an application may request protection for the content it is actively playing, which btw has shown up on OSX recently on the apple TV.
We are talking about network protocols where there is one server implementation and one client implementation. So long as it worked, it didn't matter much to do formal documentation. There is an ocean of a difference between the scant internal documentation and what Microsoft reverse engineered out of source and on the wire behavior that is now up on MSDN.
There were periods of time where/. would block Microsoft because some tool thought all the requests coming from microsoft's proxy servers were flood attacks. So I'm guess there are quite a bunch of us.
Not to deflate your rant, but IE/wininet acutally does do address fallback (ipv6 addresses are just more address in the list). I would guess that most people wouldn't want to wait the full connection timeout time for that to happen though.
You need to check out the the Windows RE (http://blogs.msdn.com/winre/default.aspx) that comes with vista. It's almost eactly what you are looking for, a winpe based (dvd bootable) repair environment.
> Yes, that's very true. But they are OPEN STANDARDS. You don't have to give IBM oodles of money, you can just figure it out for yourself.
For that definition of open standards, then OOXML is just as good, prehaps better since it is better documented (the good part of 6000 pages) and just as free to implement, versus paying IBM to figure out the missing bits in the spec (ex: openoffice specific implementation details).
Of course this is skipping the whole point that IBM has thier own office suite software that they sell per copy.
Vista has a lot of instrumentation for when things go wrong, more then any previous OS release. It's amazingly productive and useful when you can prioritize and fix issues based on real world data instead of guesswork and vague bug reports.
Why should anyone be suprised that Microsoft is proud of that ability and it's results in SP1?
To the contrary, since you brought up that anti-trust might play a role in this game, I'd like to hear why you think that or what "moves" you expect that would make it relevant. Otherwise it just sounds like you are saying "M$".
> If they don't or there are obvious means which show they are leveraging their monopoly position to promote a product with a competitive element( Adobe Flash )... well then it must be stopped and the market must be allowed to decide what they want.
I'd like to hear the reasoning on how an optional download not on Windows Update or a paticular website constitutes leveraging a monopoly.
Silverlight is an independant implemenation of the CLR and does not depend on whether or not the Full Windows CLRs are installed or not on a machine. The complete size (of the downloads) for Silverlight 1.0 is ~1.4MB and for 2.0 is ~4MB. Also, I've personally never had a reboot when installing silverlight.
You don't get it, HTTP 1.0 has to work with people speaking HTTP 0.9, HTTP 1.1 has to work with people speaking HTTP 1.0.
Backwards compatability is everything, warts and all. OpenXml is effectively a new version of the office file formats designed to be well documented and generically implementable such that it can be a standard. Imagine using a new firefox that could ONLY render perfect valid latest standards HTML, would you bother using it with the vast majority of the Web not working?
If you disagree with this philosphy, go ahead, but that is no reason to actively block standard proposals that do care about people's existing data.
> Microsoft has been legally found to be a monopoly, Apple has not. End of story
Lack of a statement is not the same as a negative statement. If you want to use the court as the main criteria, you need an anti-competative complaint against Apple that was dismissed because they were found to not be a monopoly in the market.
Slashdot Mirror
> Is google singling out IE users?
Firefox defaults to google's search, IE doesn't (at least until the OEM gets paid).
I guess I'd call that paper an example of large amounts of sloppy guesswork. He doesn't make any observations of an actual Vista box to see if any of his conjectures or expected outcomes were true. Overall the paper hasn't aged well. It's like claiming that new cars can only go a max of 20 mph while commuters are daily using them and going 55 on the Interstate. Occationally someone has a engine malfunction or runs out of gas and suddenly believes "He was right! My car can't go 55", immediately forgetting what they just did yesterday, but that's religion for ya.
It more then a style choice, it's a reflection of the customer base and support model.
Windows does corporate support which forces there to be a large number of fixes are done for individual customers or given out only if you a specific problem (to minimize risk to the general population and lower testing costs). This prevents a single release wide number that could be incremented. The service pack release collapses all these combinatorial configuration possibilities back into a single point and sometimes adds back ports of features in later OS releases.
Apple on the other hand doesn't really do corporate support, so they can go with the simpler model of keeping everyone on the same linear update branch.
> What incentive would anyone have now to provide "mature" drivers for Vista?
Because they want thier products to get shipped by OEMs and bought by users; if they don't improve thier drivers performance they lose on benchmarks, if they don't improve driver quality OEMs get upset. There is also marketing dollars when you get a "Designed for Windows Vista" Logo.
>How can any manufacturer know that their stuff will even be allowed to work with Win 7, when it comes out, or that the driver API won't be as changed as to make the improvements they're currently making to their drivers obsolete?
I would expect that any new OS version would enable new opportunities for performance tuning, but I seriously doubt that there will be another fundimental driver api shift since we just went through that with Vista and we still have plenty of milage to get out of the current generation.
There was a serious credability gap with the Vista CTP/Betas in when the OS was going to ship (even internally). So when it did ship many of our partners were caught unprepared. It has taken a year for everyone to catch up. PDC and WinHEC are pretty close together this year (Late Oct, early Nov), so the industry will have a gut check at that time to determine if the ship date is creditable.
A test suite wouldn't have helped. Win2k worked just fine with normal kerberos as a client and as a server. The problem was that if you wanted to deal with domain based groups you needed an extension, something that MSFT wasn't intrested in letting people have for free.
> This "standard" is completely irrelevant as a standard. No one, absolutely no one, is going to implement it. Not even Microsoft.
BS: Microsoft has pledged to support it in public. So assuming that the OpenOffice implementation of OOXML continues, it will support it too.
> No company is going to be allowed to implement it and become a competitive threat to Microsoft. Microsoft will shut them down with Patent Violations.
Again BS: The OSP pretty much crushes that arguement. Even the main point in the SFLC anaylsis about future versions has been addressed.
Your core arguements seem to be strictly Proof by Assertion
Do I implement a brand new spec that is clean because it ignores and wishes away all the legacy of the past, or do I implement a more complex spec that transitions them?
Developers like the earlier
Users like the later.
OSS is typically dev oriented.
Commercial software is user oriented (because they have to be).
You want to start experiencing what a real Mesh feels like, install foldershare. You can use any application to work with your data rather then be tied to one application (like google docs). It's been killer for keeping everything between my work, home and laptop machines up to date and also sharing full quality photos to my extended family. Once we all add in the ability for web applications to access and stay in sync with my data and we will be cooking with fire.
>2) If you use or redistribute software, and the software is released under the GPLv3 license, you have legal assurances that you will not wake up one morning and find that the software you have come to rely on is now subject to patents that the vendor received.
Actually for number 2, it only helps you if the person who wrote/distributed the software has the patents. You still have just as much risk against anyone else (which is where the real risk is anyways).
That doesn't help you if you aren't tying yourself to the GPL, so if I want to release a version under the BSD I'm just as potentially SOL with the IBM or Sun patent pledges as the Microsoft one.
>Doing mysterious DRM checks while copying files at a rate that would embarrass a TRS-80 Model 1
What are you talking about? There are no system "DRM checks". There is an API that an application may request protection for the content it is actively playing, which btw has shown up on OSX recently on the apple TV.
MS: 3.87%?
EU: Lower
MS: 2.98%?
EU: Lower
MS: 0.5%?
EU: Lower
MS: 0.4%?
EU: BZZT! Too late, we are going to fine you a Billion $s.
We are talking about network protocols where there is one server implementation and one client implementation. So long as it worked, it didn't matter much to do formal documentation. There is an ocean of a difference between the scant internal documentation and what Microsoft reverse engineered out of source and on the wire behavior that is now up on MSDN.
There were periods of time where /. would block Microsoft because some tool thought all the requests coming from microsoft's proxy servers were flood attacks. So I'm guess there are quite a bunch of us.
Wouldn't it be easier to just turn back on the "run" menu option on the start menu?
Not to deflate your rant, but IE/wininet acutally does do address fallback (ipv6 addresses are just more address in the list). I would guess that most people wouldn't want to wait the full connection timeout time for that to happen though.
You need to check out the the Windows RE (http://blogs.msdn.com/winre/default.aspx) that comes with vista. It's almost eactly what you are looking for, a winpe based (dvd bootable) repair environment.
> Yes, that's very true. But they are OPEN STANDARDS. You don't have to give IBM oodles of money, you can just figure it out for yourself.
For that definition of open standards, then OOXML is just as good, prehaps better since it is better documented (the good part of 6000 pages) and just as free to implement, versus paying IBM to figure out the missing bits in the spec (ex: openoffice specific implementation details).
Of course this is skipping the whole point that IBM has thier own office suite software that they sell per copy.
Vista has a lot of instrumentation for when things go wrong, more then any previous OS release. It's amazingly productive and useful when you can prioritize and fix issues based on real world data instead of guesswork and vague bug reports.
Why should anyone be suprised that Microsoft is proud of that ability and it's results in SP1?
What makes you think that every possible thing on the disk is getting loaded into RAM, or even a majority of it?
To the contrary, since you brought up that anti-trust might play a role in this game, I'd like to hear why you think that or what "moves" you expect that would make it relevant. Otherwise it just sounds like you are saying "M$".
> If they don't or there are obvious means which show they are leveraging their monopoly position to promote a product with a competitive element( Adobe Flash )... well then it must be stopped and the market must be allowed to decide what they want.
I'd like to hear the reasoning on how an optional download not on Windows Update or a paticular website constitutes leveraging a monopoly.
The parent post is completely wrong.
Silverlight is an independant implemenation of the CLR and does not depend on whether or not the Full Windows CLRs are installed or not on a machine. The complete size (of the downloads) for Silverlight 1.0 is ~1.4MB and for 2.0 is ~4MB. Also, I've personally never had a reboot when installing silverlight.
You don't get it, HTTP 1.0 has to work with people speaking HTTP 0.9, HTTP 1.1 has to work with people speaking HTTP 1.0.
Backwards compatability is everything, warts and all. OpenXml is effectively a new version of the office file formats designed to be well documented and generically implementable such that it can be a standard. Imagine using a new firefox that could ONLY render perfect valid latest standards HTML, would you bother using it with the vast majority of the Web not working?
If you disagree with this philosphy, go ahead, but that is no reason to actively block standard proposals that do care about people's existing data.
> Microsoft has been legally found to be a monopoly, Apple has not. End of story
Lack of a statement is not the same as a negative statement. If you want to use the court as the main criteria, you need an anti-competative complaint against Apple that was dismissed because they were found to not be a monopoly in the market.