The Next Big Thing for Windows, after Security?

So what you think should be the Focus of Windows. I think Security is a good primary focus but one or two of these should be a new focus without forgetting security.

A: A version of Windows that centers on fixes, updates, upgrades and polish. Big new features postponed to next major version. If XP would have a dozen big SP2 style updates, the difference between the old and the new version should equate to that dozen. Lot of attention to details and polish and postponed things.

B: UI and usability re-think for other than just Office. Change pixel relative controls to more advanced ones that do not base on the display resolution. Intro controls that change with the interface but surface the same data to developers, so that whether you use keys, mouse, touch or even some more in the future ones such as eye and brain-connection to operate the control, it will always be the best possible usability, accuracy, ease, speed and such for the program without changing line of the underlying program code.

C: Start over with the OS in such manner that no native code is carried over but pure-.NET, zero pinvoke programs would still work more directly, meaning little or no porting work for most .NET apps. .NET itself would be thoroughly ported to the new foundation, which for example could be based on the Singularity research. Native apps or services calling to Win32 would still be available for use but in their own context. ISV and third parties can do the primary work where direct bridging of data between applications and services running in the legacy context and the new is necessary.

Legacy drivers would continue to work in the new OS inside a legacy context and have an abstracted bridge for communication. If for example legacy network driver "bsods", it would only bsod one legacy driver context in the background, disabling the network connectivity for the new OS.

Thus a huge number of legacy drivers would continue to work where there is a generic bridge available but be isolated from the new foreground OS and other drivers. The bridge could use the existing abstractions by mapping current win32/user mode calls to new communication pipe between the legacy context and the new OS.

D: Suggest?

I think your suggestion about making windows over in .NET is great.
But i dont see it's gonna happen just yet. I think .NET should be more mature before you go into such a big project.

I think this subjects are very importent.

1. Stabillity. A program crashing isn't acceptable. Windows XP as it is now, is way too unstable imo. It often happens to my dad that IE crash, which make him go "omfg i hate computer".

2. Administration. Remove the Registry. Find another way to store this data, this would also enable us to do different kind of things where we can do backup of our settings. Also the Windows folder should be cleaned up. Its ugly as it is.

3. But i do think that we should begin to see BIG parts of the main system, be over time switching over to the .NET platform. Lets slowly make everything .NET. This would aswell help for 1.

Just what comes to mind.

D. After going through so much evolution up to this point in Windows, going back to the basics (perhaps with Singularity) and rebuilding Windows bit by bit with everything (security, UI, etc) in mind. Simplifying everything further. 

Ooh, and here's a good, fun one: use more AI in the OS.  Make Windows learn more about the user and predict what she wants (in a less annoying way...no dog or paperclip please).  If a device driver is bad, have Windows construct its own temporary driver until the manufacturer releases an update. Have Windows not just recognize speech and ask predefined questions, but also give Windows a certain personality that adapts to the user. Let Windows be your friend, let it engage it conversations, etc.

I may not understand things perfectly since I havent developed software in a long time; but am I right that developing Windows in .NET would entail:---

* making it incredibly easy to create extensions or modifications to the interface, like in Firefox,

* makin the interface accessible to object-based scripting hosts, like Monad,

?

Basing Windows on WinFS2* with through-out integration of it all across the system and shell.

* = NTFS + SQL Server merged, not piggybacked.

mVPstar wrote:

D. After going through so much evolution up to this point in Windows, going back to the basics (perhaps with Singularity) and rebuilding Windows bit by bit with everything (security, UI, etc) in mind. Simplifying everything further. 

Ooh, and here's a good, fun one: use more AI in the OS.  Make Windows learn more about the user and predict what she wants (in a less annoying way...no dog or paperclip please).  If a device driver is bad, have Windows construct its own temporary driver until the manufacturer releases an update. Have Windows not just recognize speech and ask predefined questions, but also give Windows a certain personality that adapts to the user. Let Windows be your friend, let it engage it conversations, etc.

Rebuilding might sounds like a good idea, but it is not.

The first reason is that even if you rebuild windows you are still going to run into the same problems that needs to be solved, sometimes problems are conflicting and you just have to choose one or the other. The myriad of problems in windows is not because it's badly built but rather that it is big and has to satisfy such a diverse customer base (pc users, hardware manufacturers, partners ....) not to mention backward compatibility. without this microsoft will lose customers.

Another reason why rebuilding windows or any software of such scale is not a good idea is because no single person/architect is capable of comprehending every single aspect of the system. Thus rebuilding windows is not neccessarily going to result in a superior system.

Thus rebuilding windows from scratch will never happen. Unless if microsoft create a R&D team that is specifically brought together to look at how to create windows from scratch and market it as a different operating system all together. No why would they do that?

I used to be a researcher of AI and feel the "intelligent software agent" is a promising direction.

http://en.wikipedia.org/wiki/BDI_software_agent

blowdart wrote:

and with it we get security problems now and again. Me, I'd like to see the XP/2000 subsystem in a virtual PC in a rewritten OS. Safe, isolated, backwards compatibility.

Yeah, this was actually what I was leaning towards: a rewritten OS with VPC for old apps.

The next big thing should surely be a new Ballmer video...

SECURITY, SECURITY, SECURITY, SECURITY

zhuo wrote:



Rebuilding might sounds like a good idea, but it is not.

The first reason is that even if you rebuild windows you are still going to run into the same problems that needs to be solved, sometimes problems are conflicting and you just have to choose one or the other. The myriad of problems in windows is not because it's badly built but rather that it is big and has to satisfy such a diverse customer base (pc users, hardware manufacturers, partners ....) not to mention backward compatibility. without this microsoft will lose customers.

Another reason why rebuilding windows or any software of such scale is not a good idea is because no single person/architect is capable of comprehending every single aspect of the system. Thus rebuilding windows is not neccessarily going to result in a superior system.

Thus rebuilding windows from scratch will never happen. Unless if microsoft create a R&D team that is specifically brought together to look at how to create windows from scratch and market it as a different operating system all together. No why would they do that?

I disagree. Microsoft’s D&R teams are working on things like Singularity because they realize that the structure of windows is flawed and bloated. And in one of the C9 interviews they admitted that the registry is a total mess and they wish it was never invented.

It’s far superior to have each app store its variable data in their own data stores like they do on Linux for example. This way programs can’t interfere with other programs settings. And if a program gets all muddled up, you don’t have to search through the registry to remove its keys. All you have to do is delete the programs directory and recopy it over.

These are the things that Microsoft would love to redesign. Now that they have had a good decade to realize there mistakes.

But use will say, “OHHH Compatibility”.
Really that’s not an issue at all. They could make an entirely new OS and simply embed a windows virtual machine into it. And place this VM in a sort of sandbox so that it can never affect the host OS.

With the latest Processor advancements in Multi Core and hardware accelerated Virtualization. You would hardly notice that stuffs running in a VM. And over time they could then remove the VM since everything would be rewritten for the new OS and there wouldn’t be a need for backwards compatibility.

zhuo wrote:



mVPstar wrote: D. After going through so much evolution up to this point in Windows, going back to the basics (perhaps with Singularity) and rebuilding Windows bit by bit with everything (security, UI, etc) in mind. Simplifying everything further.  Ooh, and here's a good, fun one: use more AI in the OS.  Make Windows learn more about the user and predict what she wants (in a less annoying way...no dog or paperclip please).  If a device driver is bad, have Windows construct its own temporary driver until the manufacturer releases an update. Have Windows not just recognize speech and ask predefined questions, but also give Windows a certain personality that adapts to the user. Let Windows be your friend, let it engage it conversations, etc.

Rebuilding might sounds like a good idea, but it is not.

The first reason is that even if you rebuild windows you are still going to run into the same problems that needs to be solved, sometimes problems are conflicting and you just have to choose one or the other. The myriad of problems in windows is not because it's badly built but rather that it is big and has to satisfy such a diverse customer base (pc users, hardware manufacturers, partners ....) not to mention backward compatibility. without this microsoft will lose customers.

Another reason why rebuilding windows or any software of such scale is not a good idea is because no single person/architect is capable of comprehending every single aspect of the system. Thus rebuilding windows is not neccessarily going to result in a superior system.

Thus rebuilding windows from scratch will never happen. Unless if microsoft create a R&D team that is specifically brought together to look at how to create windows from scratch and market it as a different operating system all together. No why would they do that?

Nicely said.  Also, there has to be a focused reason.  Just rewriting is not the answer.  That assumes prior developers where somehow "bad" and current developers are better somehow.  I don't think you can make that case.  So it is silly to talk about a rewrite unless your reasons are to go to a totally different dev platform that provides other value - such as .Net.  The work in that area is Singularity, but that may never actually be a product - at least not for a long time.

CoLD FiRe wrote:

I disagree. Microsoft’s D&R teams are working on things like Singularity because they realize that the structure of windows is flawed and bloated. And in one of the C9 interviews they admitted that the registry is a total mess and they wish it was never invented.

It’s far superior to have each app store its variable data in their own data stores like they do on Linux for example. This way programs can’t interfere with other programs settings. And if a program gets all muddled up, you don’t have to search through the registry to remove its keys. All you have to do is delete the programs directory and recopy it over.

These are the things that Microsoft would love to redesign. Now that they have had a good decade to realize there mistakes.

But use will say, “OHHH Compatibility”.
Really that’s not an issue at all. They could make an entirely new OS and simply embed a windows virtual machine into it. And place this VM in a sort of sandbox so that it can never affect the host OS.

With the latest Processor advancements in Multi Core and hardware accelerated Virtualization. You would hardly notice that stuffs running in a VM. And over time they could then remove the VM since everything would be rewritten for the new OS and there wouldn’t be a need for backwards compatibility.

i think there are benefits to a centralized registry, but obviously those disadvantages that you've mentioned. one of the promising things about winfs, is that if its implemented it could combine the benefits of a centralized registry and locally stored settings. the xml setting files would be indexed in a central store.

CoLD FiRe wrote:



zhuo wrote:  Rebuilding might sounds like a good idea, but it is not. The first reason is that even if you rebuild windows you are still going to run into the same problems that needs to be solved, sometimes problems are conflicting and you just have to choose one or the other. The myriad of problems in windows is not because it's badly built but rather that it is big and has to satisfy such a diverse customer base (pc users, hardware manufacturers, partners ....) not to mention backward compatibility. without this microsoft will lose customers. Another reason why rebuilding windows or any software of such scale is not a good idea is because no single person/architect is capable of comprehending every single aspect of the system. Thus rebuilding windows is not neccessarily going to result in a superior system. Thus rebuilding windows from scratch will never happen. Unless if microsoft create a R&D team that is specifically brought together to look at how to create windows from scratch and market it as a different operating system all together. No why would they do that?

I disagree. Microsoft’s D&R teams are working on things like Singularity because they realize that the structure of windows is flawed and bloated.

That is not true.  They never said that.  Singularity is a logic reasearch project to work on.  Any OS could become bloated, even Singularity over time.  So starting over is not the answer for bloat - if, in fact, that was true.  Its easy to solve a bloat issue - remove features.  So a balance must be struck.

The Coffeehouse.

The trend towards communitys will only continue to rise.
Already we are seeing users design their own ads for corporations (gm..firefox...)...and become more popular - or a slap in the face.

MS execs will continue to wring the monopoly washcloth of customers (with wga, multiple win versions "upgrade ready"..) until they have had enough - and as a result - will hopefully finally react to what has been being said and posted here for 2 years.

The developers, fans, customers, designers in the coffeehouse - old and young - even troll or no troll - have been right about so many things - it is almost funny to watch  (if it wernt so sad)

Seeing as Martin Taylor is leaving today - everyone knew/knows get the facts was a smoke screen.  Even mary joe foley - who does delight in any misteps ms makes - opined today the same thing :

"10.No Microsoft Linux!: ...Microsoft could have thrown a real monkey wrench into Linux companies' plans. Instead, Microsoft continues to spend lots of money, time and attention fighting open-source software on a whole host of fronts. They should have joined the camp, rather than obsessing on beating them."

Not sure they need to go that far - but they are breaking their cardinal rule: negativity loses. To win - you need to be possitive / unrestrictive / open ..  Long story short.  There are obviously many people who come here - and while they may be "nerds" the majority are not idiots.

That's why i humbley submitt for the next phase in MS's larger than life existance:  The Coffeehouse.

Use the user luke


edit: PS - re: "after security" .. ha - youd better believe we'd fix all the misguided security restrictions to boot


edit 2:  Channel 9 is PEOPLE....



*couldnt resist

blowdart wrote:

Me, I'd like to see the XP/2000 subsystem in a virtual PC in a rewritten OS. Safe, isolated, backwards compatibility.

Virtualization, while a great idea, isn't the magic bullet for backwards compatibility. Oh, it's fine if you want to run a legacy application by itself, but that's seldom the case. You want that legacy application to share data with other apps - whether it's the clipboard, access to the filesystem or whatever. You have a lot of interoperability problems here if you want to give legacy support without losing a lot of the user convenience that you have today.