Next Version of Windows? Call it '7' 488
CNet has the news that Microsoft is currently aiming to release the next version of the Windows operating system in about three years. Previously known as Vienna, the OS is now simply known internally as '7'. After achieving a quality product, the article states, Microsoft's big goal with 7 is to recapture a regular release schedule for their operating system product. From the article: "Like Vista, Windows 7 will ship in consumer and business versions, and in 32-bit and 64-bit versions. The company also confirmed that it is considering a subscription model to complement Windows, but did not provide specifics or a time frame. Next up on Microsoft's agenda is Service Pack 1 for Windows Vista, which is expected before year's end. The discussion of Windows' future isn't surprising, given that Microsoft has been criticized by business customers for delays related to Vista. Many business customers pay for Microsoft's software under a license agreement called Software Assurance."
32bit? (Score:4, Interesting)
History lesson (Score:5, Interesting)
From your profile it is clear you are too young to remember it first hand so I'll educate instead of flaming ya.
Recall that there were versions of Windows prior to Windows 3.1, the first clue to which should have been the version number. Moving to Win32 was a major upheaval in the software world, keeping compatibility with Win16 and more importantly, DOS were the major selling features of Windows 3.1. By 3.1 a lot of major software was running in Windows 16-bit AND business depended on a lot of DOS code, home users depended on DOS for the majority of games, etc. Heck, most of the software people were actually running on WinNT was 16-bit code. And most games were DOS based well into the Win95/Win98 era. It wasn't until XP was looming and game makers saw sticking with DOS as a death sentence that they drank the DirectX Kool-Aid for any project not depending on 3D.
And there were a LOT of 286 based machines not only in the installed base but still being sold. For example on the day Win3.1 shipped I was working at a Radio Shack in the D/FW area and the only 386 class machine in the store was the SCO Xenix box in the stockroom running the store. To buy a 386 class machine from Tandy you had to go to a Business Computer Center.
Re:Leave it to computer geeks.... (Score:2, Interesting)
Swi
3 years? then why Vista? (Score:3, Interesting)
Re:Didn't we just leave this party? (Score:1, Interesting)
Forcing
Your rose coloured glasses betray you. (Score:3, Interesting)
Uh, considering 2000 was just NT 4.0 with the IE 4.0/Explorer shell jammed on and a bit of tweaking under the covers for DirectX, I'm pretty disappointed. NT 4.0 ran reasonably well on a 486 with 24mb of RAM, and even better on a K6 233 with 128mb of RAM. However, without DirectX support newer than 3.0, the only real thing that I use Windows for (since the Wine/Cedega DirectX support lags a bit), I was forced to start using Windows 2000 -- an OS which does not run nearly as well as NT 4.0. NT 4.0 was also hobbled a bit by a lack of USB support. If you were to take a computer from anywhen between 1998 and 2001, and compare its performance under Windows 2000 and NT 4.0, you'd find it was not as close as you might think. By that metric, Windows NT 4.0 appears to be better than Win2k, and thus makes it the best Microsoft release ever.
Windows XP is only a minor revision to Windows 2000 (far more minor than Windows 2000 is over NT 4.0) -- which is why the internal nomenclature for the two is Windows NT 5.1 and Windows NT 5.0, respectively. Activation and a fisher-price interface (which you can disable) are the big differences, although the broken VM (minimizing a window to the taskbar lets it tell the VM to pageout its memory to swap -- even if you have many gb of RAM free!) and some other "tweaks" are also "features" of Windows XP. In any normal setup I've had, the only really bad difference between XP and 2K was that XP was limited to 10 TCP/IP connections at a time OOTB.
You could even argue that Windows Server 2003 is the best Microsoft release ever -- it's definitely the successor to Win2k in terms of no fisher-price UI, and the code tree used inside. Have you tried any of these, or are you making your claim purely on XP vs. 2K? I don't consider your Win9x experience to count, because that's a completely different codetree/build from Microsoft.
Of course, YMMV, since I only run Windows inside virtualization with either MacOS X or Linux as the real host operating systems (no troubles with search or sleep inside MacOS X -- although I disable Spotlight due to its rather large and unwelcome metadata cachces).
Re:Didn't we just leave this party? (Score:3, Interesting)
I doubt that... (Score:3, Interesting)
Now Windows 64-bit is a different set of circumstances. Most drivers have source guarded by the hardware vendor, and most of these hardware vendors ever really cared about support Windows and the only platform where Windows dominated was x86. Thus the situation is pretty grim for those companies, many of which still don't care about 64-bit, and the ones that care being ill-equipped for knowing the sorts of things that break in a platform change of this sort. The fact that the driver API is so close to what they've been using, it means the logical schedule move for them is to try to port their existing code. Of course, as Vista has shown, the commercial vendors have even more of a hard time getting it right porting from XP driver model to Vista (even without an arch change) than 32 bit to 64 bit within the same driver model.
Re:Didn't we just leave this party? (Score:5, Interesting)
Did you read the issues around Vista that were posted on Microsoft employee blogs by Microsoft employees?
And this was AFTER Jim Allchin told Gates the thing would never be done if they didn't change their methods.
So they changed their methods.
And what happened was that an incredibly broken testing system was put in place which was so badly broken that it delayed Vista release for months, while the development managers certified test builds as "Approved" even when they failed numerous - possibly even ALL - tests.
The problem with Microsoft is their PEOPLE - not their system. Their corporate culture is broken - and broken beyond repair as long as Gates and Ballmer are running the show. The stockholders should be in revolt - if they knew anything about the company.
The bottom line is that Vista has twice the code in it that Linux does - and does not do twice as much (unless you count the ridiculous DRM checking mechanisms, I suppose.) There's no way one company can develop a system with even more code in it and come up with something that works. I doubt it's even feasible for the Linux community unless the industry comes up with better development and testing methods. When Linux hits the amount of lines of code in Vista, it will be BETTER than Vista - but even I doubt whether it will be adequately reliable for normal use.
The problem is that the industry - especially Microsoft (and with an even worse attitude) - is pushing the limits of the current software development technology. The result is what we see everywhere: "Nothing works and nobody cares."
Re:What Microsoft needs to do (Score:5, Interesting)
Well other then Losing Ballmer they need to do a few things and your right one is to just throw the shit away and start fresh.
But the problem is far beyond backward compatibility.
First of all they have to pick a model on which to base this From Scratch new operating system. Now when it comes to this it is not as wide open as one would think. First of all they have to have a hardware platform on which to build. The current hardware world is basically divided between Intel Processors & Chipsets and IBM processors and chipsets. AMD's are clones of Intel processors with some interesting modes. So where to you choose to host this thing? If you go with Intel do you collaborate with them on an entirely new instruction set and design? Do you go with IMB and the PowerPC line and it's successors such as Cell Technology? If you are going to push out something totally new then these are questions you have to ask. Even Microsoft with all their money and clout has to have a partnership with the hardware producers.
In this theoretical new hardware environment what do we do? Do we come up with an entirely new bus? I mean please, anything to replace the PCI bus. And what about external communication? Do we still keep forging along with USB? Embrace FireWire? Do we go back to having dedicated interfaces for things like Keyboards and Mice?
Personally I think something like a fiber-optic switch being the heart of a "computer" and things like Processors, Ram, Storage Devices, et all each having a fiber connection. If you need more of anything you just plug it into a port of the fiber switch and off you go.
The problem with that brilliant idea is that you only switch so fast from fiber to electrical signals and back so fast. Thats one of the problems with digital is the switching and the converting, which is one of the reasons IMO why we are still stuck with things like PCI, it may suck, but it has 32 or more discrete pathways that carry all the various signals, in parallel to the various main components.
Parallelism was what made SCSI so superior to IDE or MFM along with the fact that it has its one on-board processing and command set. However along with all that came the problems of clocking and this that and the other. SAS will push the speed up, put at some point there is a limit as to how fast you can signal before it just becomes a radio transmitter and the physical layer will become to cumbersome with all the shielding and the power consumption since the faster you signal the more it just like DC to the copper.
Processor speed is starting to approach the end of its limit and its time to start looking for better ways to line instructions up across a broad front and process them at once and resolve the dependencies accordingly which brings us right back to software.
Can software, an OS, solve some of these problems? yes but not alone. To re-invent an entirely new OS will take years unless you want it to be a re-hash of the Unix model which is what, like it or not, everything out there is based upon. It will take bold new thinking in both the hardware and software fields to bring us the truly next generation of computers. Some of the people at Microsoft, IBM, Novell, The Linux community are capable of some bold steps but the the companies themselves as entities are not IMO capable. In the hay day of advancement we had places like PARC, AT&T Labs, MIT, UC Berkeley, The Government ( indirectly through grants ), and a few others doing PURE research and that was where all the great advances really came from. If we are to do so again, those places must be renewed, funded and staffed with people with imagination and the drive to take any direction that shows promise and do the basic research.
The last thing that has to happen to get all these great labs back in top shape is patent reform. These days research is often scuttled because some MBA or Lawyer cannot see the road to lock this in so lots of money can be made.
Re:Didn't we just leave this party? (Score:1, Interesting)
meanwhile, out in the world (Score:5, Interesting)
"My mum upgraded and it's exactly the same, except now it's got 'rounded corners', big deal it's the same thing except the 'corners are round'!!!"
I was quietly suprised, but she went on..
"My mum has gone throught the same hassle everytime she decides to upgrade her computer, she spends a whole lot of money, a whole lot of time and in the end the result is the same thing, except 'the corners are rounder'".
That is the perspective of a average computer user with no technical interest, I simply agreed with her and said I had noticed pretty much the same thing.
For as long as I can remember M$ have underdelivered. I don't even support windows users anymore, it's simply not worth the effort, if I fix it, it will break again subject to the three R's of windows;
Reboot the machine.
Reload the application.
Reinstall the Operating System.
I can charge them for it, but I usually just make suggestions on how to fix it so they have to go through the hassle themselves, after all it was their choice.
Nowadays, I just give people a Ubuntu live install to try, I tell them it will probably be a bit slow running of the CD or DVD and to focus on the way it works rather than the speed. I think that, despite the fragmentation in the Linux distribution's, I continue to notice a trend of installing more Linux, either Fedora or Ubuntu. This year I've actually had people asking me for linux installs, I haven't had any of these lay-users wanting to go back to windows even though I give them the option to. In reality, I think it comes down to this,
You can fool some of the people all of the time, or all of the people some of the time, but you can't fool all of the people all of the time.
I don't think this simple peice of wisdom factors into M$'s business plan.
Re:Didn't we just leave this party? (Score:3, Interesting)
Ambitious? The "innovations" that were abandoned when Longhorn morphed into Vista were WinFS, an object oriented filesystem which layers SQL Server onto NTFS, and has been dropped more times than Michael J. Fox's pen, and an upgrade to .NET which replaced GDI with Avalon and COM+ with Indigo.
These changes were essentially architectural bugfixes to repair design flaws inherited from the NT/2000 line, and wouldn't have significantly changed the user experience. Even WinFS, which had the most potential, was more for locking file attributes to Microsoft's filesystem than a way of achieving anything that couldn't be done with a less tightly coupled database layer.
No need to re-invent everything. (Score:3, Interesting)
What Microsoft needs to do is write a UNIX like O/S with the following features:
1) write the successor to the C language: a strongly and statically typed C derivative with none of C's deficiencies but allowing access to the bare metal, also incorporating functional features. They certainly have the stuff to do that. Then use this language to:
2) simplify the driver development system using microkernel techniques.
3) write a single tree file system, like in Unix, where filesystems are mounted/unmounted.
4) write a network-abstraction system on top of the file system described above, where resources of the system are abstracted over the network.
5) write an object/typed database layer on top of this network abstraction system, and offer it as the default. Use MIME as the typing mechanism.
6) write a Window System, ala X, which is a regular process sitting on top of the database system.
7) write a truly object-oriented toolkit which includes gui, xml, database, and everything else required for modern apps. See the Qt model for a good example.
8) use unicode throughout the system. Don't have 8-bit functions and wide-character functions. Make character a single 32-bit data type which can host all unicode characters, so you don't have a problem on how strings are handled. Forget C string handling, and do it in the modern way.
9) provide garbage collection where appropriate. This means that all code, except the microkernel and system drivers, should be garbage-collected.
10) use the Erlang model for multithreading. Provide userland multithreading libraries for the fastest multithreading possible.
11) virtualize the O/S for the user. Make it as if the user can read/write/execute everything, but any change will not be reflected to the system files or other users' files. Provide a ring security mechanism, like 80x86 rings, so as that networked applications can not hurt the rest of the system and can only communicate with it through specific call gates.
Microsoft's problem is entirely a software problem. They want to use 70's technology for the 21st century. It's doable, but only if UNIX like principles are followed.
(Thank God Microsoft does not read slashdot though, because there are quite a few interesting proposals here).
Progressivly better? (Score:3, Interesting)
First versions of 98?
Windows ME?
The patterns seems to be one big step forward, a few small steps back.
Win 2000 was the high point for me so far.