Microsoft says Windows 7 isn't fit for modern hardware and security

[bitbucket]

This is the problem here.

I don't know how to say this without being harsh, but it's the complete lack of awareness of your ignorance on the matter that fuels your viewpoint.

Do you know how many things have changed in this time? How something like Thunderbolt or USB C can create big challenges for kernel designers? What about CPUs? Do you have any appreciation the massive changes in architecture we've seen? Not even just on power management either.

Of course because you're 100% in the dark on all these things, you don't appreciate why they have happened. But it's incredibly frustrating when you are developer, that your project or product could be made with such ease if only people allowed you to use modern tools. But they don't because they think they understand what they are powerfully, willfully ignorant of.

I couldn't agree with you more. I wanted to say this myself but also didn't manage to find the words without being rude.

[Corky34]

This is the problem here.

I don't know how to say this without being harsh, but it's the complete lack of awareness of your ignorance on the matter that fuels your viewpoint.

Do you know how many things have changed in this time? How something like Thunderbolt or USB C can create big challenges for kernel designers? What about CPUs? Do you have any appreciation the massive changes in architecture we've seen? Not even just on power management either.

Of course because you're 100% in the dark on all these things, you don't appreciate why they have happened. But it's incredibly frustrating when you are developer, that your project or product could be made with such ease if only people allowed you to use modern tools. But they don't because they think they understand what they are powerfully, willfully ignorant of.

Isn't that more a design flaw of Windows though? I mean no one forced Microsoft to design Windows in such a way that made it nearly impossible for them to add, remove, and update features within the kernel or OS, the term DLL Hell wasn't coined for no reason.

[scaryjim]

I'm 5 beers into lunch.

Then I'd strongly advise against posting when drunk, if you can't avoid making inflamatory comments...

Well it's obvious you've had to much to drink ...

Wouldn't the "sober" response be to simply report such posts to the moderators, rather than responding in the thread and potentially exacerbating the situation?

I've cleaned the above mentioned posts out of the thread. Consider this a friendly warning, both of you. Don't make me give you an unfriendly one...

[TheAnimus]

It's rather hard to communicate a point when that person is making assertions that are born wholly out of ignorance... Even when sober!

I mean I can't begin to fathom the confusion of ideas to relate DLL Hell a filesystem vs binding problem, to radical kernel design changes. Let alone the backdrop of total ignorance of the NT Kernel being a great example of a widely used "hybrid microkernel" in the first place.

Bertrand Russell once said that The trouble with the world is that the stupid are cocksure and the intelligent are full of doubt. At times it's all the more exasperating once you've had a few.

[EndlessWaves]

All I really wanted from 10 was a more reliable hardware/software layer that looked like 7, but performed better. Ideally it would support all of the newest standards and technologies, be more secure than 7 was, and maintain some useful stuff like the clock, start menu, calculator and snipping tool. I don't know where Microsoft got the idea we all wanted and needed telemetry, Cortana, Edge, their own apps, a merger with Xbox, adverts on our own desktops...

Take a look around at what other software developers are doing and how users are reacting to that.

There's the PC gaming market for example. Gamers are perfectly happy to buy games that come with telemetry, app stores, console cross-linking and adverts. Many of them even praise the platforms that deliver such things.

If it works for Steam, Uplay and Origin on Windows then clearly it can work for Windows itself.

[nobodyspecial]

ROFL. If you're depending on windows built-in security crap as an IT manager you need to be fired.

Please list the windows 7 headaches...LOL.

No windows coming to my PC (new anyway) until 2020 at the earliest. Telemetry, terrible explorer, more clicks to get the information I use, hiding the information I need more often than not deeper (idiot proofing the OS?) for god knows what. MSFT can kiss my butt. IF you can't give it away for free for 18 months (still available free through accessibility crap) and barely meet half of the last OS (24% for win10, 49% for win7 still) that was not FREE, even based on their own numbers (netmarketshare IS microsoft), you have failed and need an OS we actually WANT rather than your idea of what we need.

Can't wait for Vulkan gaming to take over so I can chuck this OS train and start the *nix bandwagon (or whatever, tri-boot or something of multiple free OS's). MSFT has until 2020 to get a USER based OS out the door or I'm gone if they keep going down the current path. By that time Vulkan hopefully will give me all the reasons I need to jump elsewhere.

Enterprise BTW has a decimal point for win10 market share. Netmarketshare you'll notice (I'm mean microsoft) has zero info on enterprise. No point in bragging about a goose egg I guess...LOL. Not aware of many planning anything before 2018-2019. They're hoping microsoft will have a different option by then or extend win7 via force. That worked pretty well for xp/server 2003

[nobodyspecial]

This is the problem here.

I don't know how to say this without being harsh, but it's the complete lack of awareness of your ignorance on the matter that fuels your viewpoint.

Do you know how many things have changed in this time? How something like Thunderbolt or USB C can create big challenges for kernel designers? What about CPUs? Do you have any appreciation the massive changes in architecture we've seen? Not even just on power management either.

Of course because you're 100% in the dark on all these things, you don't appreciate why they have happened. But it's incredibly frustrating when you are developer, that your project or product could be made with such ease if only people allowed you to use modern tools. But they don't because they think they understand what they are powerfully, willfully ignorant of.

http://www.wegotserved.com/2016/01/1...products-2016/
You can get thunderbolt 3 and usb3.1 on Z170 etc boards by adding it, usb3.1 already on them all support win7.
https://www.newegg.com/Product/Produ...82E16813998030
https://www.newegg.com/Product/Produ...82E16813995053


Not sure what you're talking about. Note the LONG list of boards the Asus addons work with and all show win7 32/64bit. Can't be bothered to look up if you can do this on many others, but many newer z170 board ship with 3.1 so...Whatever

Massive changes in architecture? Skylake (and apparently Kaby also on Z170) works just fine in windows 7 and Z170 boards have both technologies you mentioned. Your complaints seem to show you sir are in the dark or "ignorant" not the OP.

You're worried about being harsh? LOL. Please tell me you guys don't have all these stupid snowflakes we have here being pumped out like crazy from our USA colleges...ROFL. Jeez, you put up a chalk trump sign here, and these fools need play-doh, crayons+coloring books, therapy puppy dogs, skip finals etc. ROFL. I hope trump denies funding for his first 4yrs at least, to any school who used this crap for their kids over his election. $20 says tuition fees come down immediately after across the country if he does it.

[nobodyspecial]

https://www.newegg.com/Product/Produ...-001-_-Product
Asrock has Thunderbolt 2 cards, so won't be surprised if they too add T3 card soon. Again, not like win7 can't support the stuff you mentioned previously. Besides that, ever heard of "the customers is always right?". Microsoft forgot this. Seems you have too.

[Corky34]

Wouldn't the "sober" response be to simply report such posts to the moderators, rather than responding in the thread and potentially exacerbating the situation?

It would but i had assumed it was just the drink talking so gave TheAnimus the benefit of doubt, although i see as (s)he has continued to personally attack someone (s)he's knows nothing about even after being advised about inflammatory comments i see i was mistaken.

It's rather hard to communicate a point when that person is making assertions that are born wholly out of ignorance... Even when sober!

It's even harder to communicate with a drunkard who resorts to accusing people (s)he knows nothing about of lacking knowledge or information on a subject, especial when that drunkard is standing in a glass house.

I mean I can't begin to fathom the confusion of ideas to relate DLL Hell a filesystem vs binding problem, to radical kernel design changes. Let alone the backdrop of total ignorance of the NT Kernel being a great example of a widely used "hybrid microkernel" in the first place.

I can't fathom how someone can conflate a filesystem with DLL's (Dynamic link library) when the clues in the name and then try's to juxtapose those two disparate ideas with how the human brain segregates elements in complex patterns of sensory input.

I also can't fathom how someone thinks adding USB3 support to a kernel requires radical change and an entirely new OS when that's the job of drivers, like the ones you can slipstream into the Windows install media, compile them into a Linux kernel and install them in OSX.

It's obvious you've been listening to far too much marketing hype as it seems you believe a "hybrid microkernel" is actual a thing when it's just marketing, even a hybrid kernal is a controversial classification, let alone the made up term of "hybrid microkernel", there are monolithic kernels, there are hybrid kernels, and there are microkernels, the only place a "hybrid microkernel" exists is in the mind of marketing people, but don't take my word for it, let's ask Linus Torvalds.

As to the whole "hybrid kernel" thing - it's just marketing. It's "oh, those microkernels had good PR, how can we try to get good PR for our working kernel? Oh, I know, let's use a cool name and try to imply that it has all the PR advantages that that other system has"

Or how about we ask Sonny Rao of the IBM LTC Kernel Performance File/IO team.

It's wrong. I've read the same crap many years ago when I still bothered to use windows in some book. I can't remember what book, but IIRC it was fairly "well respected" (whatever that means in the windows world).

What's funny is people say the same thing about the Darwin kernel, and they're equally wrong there. I think this talk of micro-kernels was born from some kind of half-assed marketing game to try and make XXX kernel seem "more advanced." "Micro" and "hybrid" are always better/more advanced than "Monolithic" right?? Certainly sounds that way. :-)

Like i said people in glass houses shouldn't throw stones.

[Ozaron]

Take a look around at what other software developers are doing and how users are reacting to that.

There's the PC gaming market for example. Gamers are perfectly happy to buy games that come with telemetry, app stores, console cross-linking and adverts. Many of them even praise the platforms that deliver such things.

If it works for Steam, Uplay and Origin on Windows then clearly it can work for Windows itself.

While I understand what you're trying to say here, I don't feel like the example given is a working one. You're right in that people are perfectly happy to buy said games with said "features", but they do so based more on a compromise IMO. If you offered the same game twice, one with and one without telemetry, ads, microtransactions or in-app stores, etc., and they were accurately represented on the boxes / store pages? Anyone who knew the alternative existed wouldn't pick the one with those features (though sometimes you do get people who are happy with paying for success in F2P games to either win or support the game devs).

If we solely count the distribution platforms, well, Steam's end user system audits have proven somewhat useful to game devs who want to include as many players as possible. Not to mention outsiders who reference the success or failure of hardware and software devices. They keep it open and available to anyone. The app stores for each platform are a core part of delivering games to the customers, the platform is nothing without them. Adverts, sure, but not in the games, and they're frequently so well targeted that nobody minds. So here, you're right. But the difference is that people sign up to these services voluntarily because they accept what is offered under its terms. I know that there are at least a few on Hexus who refuse to do so on principle, and those are the kinds of people who most vehemently object to having such services forced upon them by an OS.

As someone who has used all of your mentioned platforms I know that my choices when I buy a game are based entirely on the compromise of whether the hassle is worth the content. If you could provide evidence of people being pleased about a PC title containing ads or telemetry I'd appreciate it.

I think the evidence that disproves your last statement is right here in this comment chain. I haven't seen anyone praise Windows Store, their attempt at providing on-desktop Steam / Origin game stores. Nor have we seen them happy about Edge... yet. I haven't spoken to anyone anywhere who appreciates Cortana. And so it goes..

[Ozaron]

ROFL. If you're depending on windows built-in security crap as an IT manager you need to be fired.

Please list the windows 7 headaches...LOL.

Enterprise BTW has a decimal point for win10 market share. Netmarketshare you'll notice (I'm mean microsoft) has zero info on enterprise. No point in bragging about a goose egg I guess...LOL. Not aware of many planning anything before 2018-2019. They're hoping microsoft will have a different option by then or extend win7 via force. That worked pretty well for xp/server 2003

Luckily, nobody in that position relies on Windows. That said, there's a difference in trying to secure a version like 7 compared to a version like Vista, at release. I hope I don't need to link you a history lesson on how full of holes it was.

As to Enterprise, if the vast majority of those who were OK with the concept of W10 from the start are therefore people who upgraded at home, then I have reservations on how many Enterprise based systems would have been in the pool of home-use by average users. The vast majority of businesses, as you know and mentioned, haven't upgraded and plan to hang onto 7 bearing in mind they currently have no reason to leave it (this article's reason for being). I don't know how many got forced into moving from XP to 7 but they likely won't want to move again after such a short time.

[SineWave]

It's funny reading all these comments while using Debian Linux, and I've been using it for more than 3 years now. Couldn't be happier. Microsoft, Windows, blue screens, all kinda cock ups... it all seems like a bad dream and distant past now. Enjoy your Windows chaps! You just seem to need longer to get wiser.

[TheAnimus]

It's even harder to communicate with a drunkard who resorts to accusing people (s)he knows nothing about of lacking knowledge or information on a subject, especial when that drunkard is standing in a glass house.

But the problem is you don't understand what you are talking about.

I can't fathom how someone can conflate a filesystem with DLL's (Dynamic link library) when the clues in the name and then try's to juxtapose those two disparate ideas with how the human brain segregates elements in complex patterns of sensory input.

OK, let me explain DLL hell for you.

Back when making the PE format, someone made a mistake. But it's sort of understandable, computers were slow back then. Anyway, a PE file has a section called the IAT or Import Address Table, it looks something like this

{
    int id
    lpstr libraryName  // The name of the library, ie "nt.dll"
    lpstr functionName // The name of the run time bound function, ie "GetProcAddressA"
    farproc functionAddress // This is put in by the runtime binder.  It will be the address of the function in the process memory.
}

So I'm guessing you can see what the problem is right? The lack of unicode support. Jokes aside there is no version stamp, just a name of a DLL. So when bound at runtime if the wrong DLL was picked up, it would create havoc.

So this is why it was basically a filesystem vs binding problem. If everyone distributed every dll that the entire dependency tree needed in the exact same folder, there would be no binding failures.

Then there was the mistake of the search behaviour, this is why such horrible things as SafeDllSearchMode exist or LOAD_WITH_ALTERED_SEARCH_PATH.

You can have these kind of binding problems, and you do, with every platform except the NT kernel itself but I'm sure you don't need me to explain that difference for you do you? Because you clearly know what you are talking about....

I also can't fathom how someone thinks adding USB3 support to a kernel requires radical change and an entirely new OS when that's the job of drivers, like the ones you can slipstream into the Windows install media, compile them into a Linux kernel and install them in OSX.

Not for the host controller. Seriously, read the WDK.

At what point do you declare a new version and not have the full backwards compatibility? When I am lucky enough to find myself writing code I'm often fighting hard against the mistakes that still exist for backwards compatibility.

It's obvious you've been listening to far too much marketing hype as it seems you believe a "hybrid microkernel" is actual a thing when it's just marketing, even a hybrid kernal is a controversial classification, let alone the made up term of "hybrid microkernel", there are monolithic kernels, there are hybrid kernels, and there are microkernels, the only place a "hybrid microkernel" exists is in the mind of marketing people, but don't take my word for it, let's ask Linus Torvalds.

Or how about we ask Sonny Rao of the IBM LTC Kernel Performance File/IO team.

So let me get this right? You are saying because people who advocate an idealistic method of software engineering, don't think it's a thing, it's not a thing?

Linus clearly outlines what he thinks the differences are, and why he doesn't agree with them. I would argue the WDDM 2 and later show that to be demonstrably false, but it's not that nice having something you said 10 years ago dragged out to be proved wrong.

The distinctions are always going to be hard, I would argue that it's simple enough when you are making great efforts to compartmentalise and stay out of ring 0, you are at least attempting a microkernel.

Like i said people in glass houses shouldn't throw stones.

The problem is you appear to dislike the company and know little about the technologies. I have to know lots about them it's kind of been my job. this is also why I hate them with the burning fires of a thousand suns.

Because trust me, they are all crap, they are all monuments to compromise.

What amazes me here is we have people who demonstrably don't understand what they are on about saying that the vendor is lying. I'd wager you couldn't write a device driver for a simple USB HID that lived in kernel mode and say toggled the LEDs on a keyboard, but you are apparently an expert in Kernel Design more than the vendor.

[TheAnimus]

http://www.wegotserved.com/2016/01/1...products-2016/
You can get thunderbolt 3 and usb3.1 on Z170 etc boards by adding it, usb3.1 already on them all support win7.
https://www.newegg.com/Product/Produ...82E16813998030
https://www.newegg.com/Product/Produ...82E16813995053


Not sure what you're talking about. Note the LONG list of boards the Asus addons work with and all show win7 32/64bit. Can't be bothered to look up if you can do this on many others, but many newer z170 board ship with 3.1 so...Whatever

Massive changes in architecture? Skylake (and apparently Kaby also on Z170) works just fine in windows 7 and Z170 boards have both technologies you mentioned. Your complaints seem to show you sir are in the dark or "ignorant" not the OP.

You're worried about being harsh? LOL. Please tell me you guys don't have all these stupid snowflakes we have here being pumped out like crazy from our USA colleges...ROFL. Jeez, you put up a chalk trump sign here, and these fools need play-doh, crayons+coloring books, therapy puppy dogs, skip finals etc. ROFL. I hope trump denies funding for his first 4yrs at least, to any school who used this crap for their kids over his election. $20 says tuition fees come down immediately after across the country if he does it.

Don't conflate an implementation with a good implementation.

People at the moment are doing everything they can to manage power consumption for consumer devices. Apple for instance brought in GCD a while ago. Now it's completely possible to do periodic or event based tasks before. The only difference is now in OSX it got a little bit more efficient thanks to a lot of work.

The changes that have happened in the NT Kernel to allow these technologies to be used efficiently are the things that require a lot of redesign. Now could you port that kernel back? Probably. But you'd be making a fair few patches along the way, you can't blame vendors who want to move them forward unless you think 10 years isn't enough.

So don't confuse "just fine" with "working fully". Remember the Skylake bug? How many people who had defective CPUs actually noticed, how many stated they were "working fine". That's because they weren't even using the features fully.

[Corky34]

But the problem is you don't understand what you are talking about.
OK, let me explain DLL hell for you.

Back when making the PE format, someone made a mistake. But it's sort of understandable, computers were slow back then. Anyway, a PE file has a section called the IAT or Import Address Table, it looks something like this

{
    int id
    lpstr libraryName  // The name of the library, ie "nt.dll"
    lpstr functionName // The name of the run time bound function, ie "GetProcAddressA"
    farproc functionAddress // This is put in by the runtime binder.  It will be the address of the function in the process memory.
}

So I'm guessing you can see what the problem is right? The lack of unicode support. Jokes aside there is no version stamp, just a name of a DLL. So when bound at runtime if the wrong DLL was picked up, it would create havoc.

So this is why it was basically a filesystem vs binding problem. If everyone distributed every dll that the entire dependency tree needed in the exact same folder, there would be no binding failures.

Then there was the mistake of the search behaviour, this is why such horrible things as SafeDllSearchMode exist or LOAD_WITH_ALTERED_SEARCH_PATH.

So in other words a design flaw, you can go into the minutia of what that design flaw is and why it happened but none of that matters, what matters is that Microsoft have spent the last 23 years putting sticking plasters on the problem instead of fixing it.

You can have these kind of binding problems, and you do, with every platform except the NT kernel itself but I'm sure you don't need me to explain that difference for you do you? Because you clearly know what you are talking about....

Sorry didn't you start your post by saying that i don't understand what you I'm talking about, now you're saying i do, i wish you'd make up your mind.

Not for the host controller. Seriously, read the WDK.

At what point do you declare a new version and not have the full backwards compatibility? When I am lucky enough to find myself writing code I'm often fighting hard against the mistakes that still exist for backwards compatibility.

Yes for the host controller, the kernel loads the driver as it does for all drivers.

And the point you declare a new version and/or have full backwards compatibility is however long you choose to do so, no one is forcing third parties to support older versions of software and no one is demanding full backwards compatibility, the extent of support has always been in the hands of those third parties.

So let me get this right? You are saying because people who advocate an idealistic method of software engineering, don't think it's a thing, it's not a thing?

Linus clearly outlines what he thinks the differences are, and why he doesn't agree with them. I would argue the WDDM 2 and later show that to be demonstrably false, but it's not that nice having something you said 10 years ago dragged out to be proved wrong.

The distinctions are always going to be hard, I would argue that it's simple enough when you are making great efforts to compartmentalise and stay out of ring 0, you are at least attempting a microkernel.

No I'm saying a "hybrid microkernel" is marketing talk, that the closest thing to a "hybrid microkernel" is a monolithic kernel with elements of microkernels incorporated into it, that doesn't make it a "hybrid microkernel", the two things a distinctly different.

I'm not saying Windows doesn't use a hybrid kernel I'm saying it doesn't use a "hybrid microkernel", they attempted that years ago and the performance overheads were to great so they abandoned the idea and just incorporated some parts of a microkernel into a monolithic kernel.

The problem is you appear to dislike the company and know little about the technologies. I have to know lots about them it's kind of been my job. this is also why I hate them with the burning fires of a thousand suns.

Because trust me, they are all crap, they are all monuments to compromise.

What amazes me here is we have people who demonstrably don't understand what they are on about saying that the vendor is lying. I'd wager you couldn't write a device driver for a simple USB HID that lived in kernel mode and say toggled the LEDs on a keyboard, but you are apparently an expert in Kernel Design more than the vendor.

It's got nothing to do with my personal feelings towards the company, it's got to do with my opinion that, like you say, they're monuments to compromise but instead of fixing those compromises they slap sticking plasters over them leading to yet more compromise and more plasters.

What we're talking about has nothing to do with peoples knowledge of kernel design or driver development, it's that people understand compromises have been made and they're the ones who have to live with the consequences of those compromises, it's that instead of making excuses they expect those issues to be fixed so they don't cause further problems.

[TheAnimus]

it's that instead of making excuses they expect those issues to be fixed so they don't cause further problems.

That's exactly what's been done with a bunch of Kernel APIs, it's a new version. You are complaining about the exact thing they are doing.

As for DLL Hell. Oh boy did they solve it, by arguably making worse! Side by Side... DLL Hell isn't a design flaw, it's a feature. That's the horrible thing about the problem, how many times has it been (badly) solved? Look at NPM 2 vs 3, look at dpkg. Heck look at NuGet graph theory makes for fun times with a filesystem.